rangifer’s diary: pt. cix

Errata for “Heal”

The previous entry in this diary has an essay entitled “Heal”. You should read it now, if you haven’t already. And if you have read it already, then you should know that I accidentally forgot one kinda important part of the essay, which I’ve gone back & added to it: Heal can be used when unarmed. See, in particular, the “As dissociation” section.

All Hallows’ Eve

It’s time for another one of MapleLegends’s Hallowe’en events! Wow…!

“Wait”, I hear you say. “Another Hallowe’en event? There was one last year??” Yes, actually, there was. Unfortunately, I was unable to attend, & to this day I still regret missing out on the Soft Face… 😔

But that’s okay, because I’m here now. And most importantly, that means that I get to do the Hallowe’en quests!!

Traditional Hallowe’en quests

There are a handful of quests associated with the Hallowe’en event that take place in a location that nevertheless exists year-round: the Haunted House of Masteria’s Phantom Forest. I’ve known that these quests existed ever since I joined MapleLegends, but I never got the chance to do them myself. This was a source of some sadness for obvious reasons, & also, these quests are the only way to obtain a Toymaker Hammer! The only two-handed blunt weapon equippable by anyone other than warriors, archers, & thieves!!^[1] To make matters worse, the first Hallowe’en event that I did experience didn’t feature these quests…

This time, however, I finally get to do them! And better yet, they’re permanent year-round quests now!! That’s what we like to see: less event junk, & more real game content!

Poofcakes’s Halloween: Haunted Mansion quests^[2] is a fairly comprehensive guide to these quests. These quests are covered within the “In quest impressed, & all the rest” and “Ever still to haunt my house” sections below.

Entering the Haunted Mansion

The Haunted Mansion quests

The Butler Party Quest

Hanamiru’s Halloween 2023 — Butler PQ guide (archived) is an excellent guide to this version of BPQ.

Stage 1: Parallel JQing

Stage 2: Dodging Mirror Ghosts

Stage 3: Riddles four

This stage is, in a way, the opposite of the previous stage: whereas stage 2 is cleared as soon as someone is fast enough to clear it, stage 3 is cleared on a “weakest link” basis, meaning that it’s only cleared as quickly as the slowest person in the party. Everyone must visit their own sequence of bookshelves, meaning that there’s not generally any way to do someone else’s work for them, unlike in every other stage of the PQ. This, combined with the fact that stage 3 makes up the bulk of the PQ, & the fact that BPQ is — as we’ll see — quite time-sensitive, means that stage 3 puts quite a bit of pressure on players to not be responsible for slowing down their entire party.

When I started BPQing, Hanamiru’s guide didn’t exist yet, & so I learned stage 3 “the hard way”. I solved almost all the riddles myself (excepting two or three that I consulted party members for), and I found the locations of bookshelves by just… walking around & looking for them. Finding bookshelves is very non-trivial even when you know what you’re looking for, partly because the map is absolutely huge, & partly because navigating the map is itself painful.

After some runs of learning the hard way, I pieced together my understanding of this stage, & ended up running the vast majority of my BPQs with stage 3 pretty much memorised. I was later somewhat surprised to check Hanamiru’s guide and see that there are 28 bookshelves — that’s a lot of riddles & a lot of bookshelves to memorise!

When I talk to the Butler, I jot down quick notes so that I remember the list that I was given. I usually end up with something like (using the list from the above image as an example):

guard
toy
mirr
empty

Bonus stage: EPQ stage 4… but with ghosts

And now, we arrive at the tragic fatal flaw of BPQ: its bonus stage.

Butlerewards

But what are the rewards for doing BPQ? After all, it gives absolutely no EXP whatsoever! Well, the reward structure can be split out into two parts: rewards obtained by clearing, & rewards that require the bonus stage.

Clear rewards

Bonus stage rewards

As you’d expect, a higher tier corresponds to better rewards, where duds < raffle tickets ≪ MCP1s < MCP2s, & more NX is obviously better than less. However, even at tier 5, you can still get duds, albeit with a lower probability than with lower tiers. As we’ll see, the raffle for this event is pretty dry, so turning in 50 Ghost Sacks & getting 10 Malady’s Candies for it is a bit like throwing food directly into a toilet: you knew that you were going to get shit anyway, so you may as well skip the middleman.

Milking the bonus stage!

My strategy was to not bother turning in Ghost Sacks unless I had at least 40 of them, & of course to ideally turn in 50 at a time. Tier 4 is similar enough to tier 5 that I didn’t think that it was worth doing another entire BPQ & bonus stage just to get the remaining 1〜10 Ghost Sacks that I needed for tier 5.

However, the time necessary to farm ≈50 Ghost Sacks is very much not trivial. This is why BPQ is so time-sensitive: if you don’t clear the PQ fast enough, then you’re robbing yourself & your entire party of the precious time necessary to get good bonus stage rewards. I found that, in general, a party needs somewhere around ≈5.5 minutes or more to comfortably farm 50 Ghost Sacks per person. Note that this is more than half of the entire BPQ timer, which is only 10 minutes. The result is that, unless you’re just skipping the bonus stage (as a minority of parties do), more time is spent in BPQ’s bonus stage than in all three BPQ stages combined.

This is why I refer to the bonus stage as the “tragic fatal flaw of BPQ”: it turns what would otherwise be a fun little character-inclusive PQ that you can clear without much effort, into a slow Ghost-capturing slog that’s interspersed with episodes of frantic PQ speedrunning. Moreover, the speedrunning aspect makes bonus-stage-included BPQ uncomfortable or inaccessible for some players: they see the letters ⟨PQ⟩, wait around in the channel 1 lobby until they can find a party, thinking that this event PQ will be a fun little romp à la KPQ etc., & then quickly realise that “casually learning as you go” might not be up to the standards of seasoned BPQers.

Stage 3 theory

Although the speedrunning aspect of this PQ can make it both tiring, & inaccessible to some players, it is also… speedrunning. And some people like that! Heck, I’ve even done some speedrunning of my own in several past entries within this diary, not the least of which being the “Hinamatsuri” section of the previous entry!

Getting to the bonus stage with 7:35 on the clock implies a clear time of 2:25. I suspect that this is pretty close to an ideal run.

The reality is that the time that it takes to clear BPQ is heavily influenced by RNG:

• Because PCs are spawned into the bottom-left corner of stage 1, the most favourable outcome is that chest #1 is the correct one, simply because JQ #1 is the fastest JQ to get to the beginning of.
• The closer to the centre of the map that the correct mirror is in stage 2, the better.
• Stage 3 can take considerably differing amounts of time, depending only on how much the PCs are forced to walk to get to their four bookshelves in order.

Because stage 3 tends to be the bulkiest of the three stages, & because it necessarily directly exposes all players to the effects of its RNG, it’s perhaps the most interesting. So, welcome to the first of several sections of this diary entry that analyse PQ optimisations!

The problem

It’s clear that, for any given riddle list, there is a well-defined shortest path that takes the PC from the spawnpoint of stage 3, to all four bookshelves in order, & then to the hearth. The length of such a path can be understood as the time (in seconds) that it takes to traverse it. What we want to ask, then, is: how is this length distributed?

In particular, we mostly just want to know three summary statistics: the minimum, the maximum, & the expected value. These correspond to the best case, the worst case, & the average case, respectively.

Graphs

A first step might be to formalise stage 3’s map (Library Room) as a mathematical object of some kind. The obvious choice is a graph, where the vertices are locations throughout the map. Then, we can connect the vertices with edges that are weighted according to how long it takes to travel between those two vertices. The set of vertices, then, would look something like the union of all of the following:

• The spawnpoint.
• All 28 bookshelves.
• All 3 upward teleporters & their destinations (four vertices in total, after deduplication).
• One vertex for each point on the vertical line passing through each bookshelf — & through the spawnpoint — if that point is on a floor, is strictly below the bookshelf (or spawnpoint), & isn’t already accounted for. This is sufficient to allow for the necessary down-jumping.
• The hearth.

We quickly realise that the edges of this graph are going to need to be directed: for example, doing the reverse of a down-jump generally takes considerably longer than the down-jump itself, so down-jump edges should be directed downwards.

This graph has two small flaws: it cannot encode teleporter delays, & doesn’t include the nigh-unused downward teleporters. This latter flaw is one that I’m simply going to ignore, for two reasons: basically no one uses these teleporters anyway, and more importantly, they’re only faster than down-jumping — & only slightly, thanks to teleporter delays — in somewhat rare cases.

What about teleporter delays, though? Taking an upward teleporter to go up one floor is basically instantaneous. However, also taking the next teleporter, so that you can go up a total of two floors, incorporates one teleporter delay into the time taken. After taking one teleporter, there’s a brief cooldown before you can take another, which I’m here referring to as teleporter delay. But if the edge connecting a teleporter to its destination has a weight of 0 (effectively instantaneous), then traversing two such edges in sequence has a path length of $0+0=0$. Where’d the teleporter delay go? This can be ignored, effectively removing teleporter delays from the model, or can be taken into account with ad hoc rules about traversing the graph.

(Quasi)metrics

So far, we haven’t specified that this has to be a complete digraph, with the implication being that we can calculate the directed distance from any given vertex to any other (or possibly the same) given vertex by finding the shortest path, & summing the weights of its edges. This is starting to smell a lot like a metric space, but don’t get too excited.

Firstly, although a complete digraph is indeed unnecessary, we do need enough edges to allow the representation of every possible in-game shortest path between the spawn point & any bookshelf, between any two bookshelves, and between any bookshelf & the hearth. I’m starting to get the idea that the graph representation may or may not actually come in useful, given that it’s not yet doing anything that our basic geometric intuition isn’t already capable of…

But second of all, unfortunately for any aspirations of “just using a metric space instead”, we most certainly do not have a metric space here. Consider the requirements for $𝑑$ being a metric over our vertex set, which we’ll call $𝑉$. Let ${𝑣}_{𝑖}\in 𝑉$, for any index $𝑖$:

• ✅ No dislocation: $𝑑\left({𝑣}_0,{𝑣}_0\right)=0$.

  Easy enough.

• ✅ Positivity: ${𝑣}_0\ne {𝑣}_1\to 𝑑\left({𝑣}_0,{𝑣}_1\right)>0$.

  This is satisfied by ensuring that no two distinct vertices refer to the same position within the map, & that instantaneous travel is not possible, i.e. all edge weights are strictly positive. This can be achieved either by making teleporters almost — but not quite — instantaneous, or by collapsing all upward teleporters & their destinations into a single vertex. The former option is probably preferable, albeit somewhat less simple.

• 🚫 Symmetry: $𝑑\left({𝑣}_0,{𝑣}_1\right)=𝑑\left({𝑣}_1,{𝑣}_0\right)$.

  This one is an absolute no-go, for the same reason that the graph must be directed.

• ✅ Triangle inequality: $𝑑\left({𝑣}_0,{𝑣}_2\right)\le 𝑑\left({𝑣}_0,{𝑣}_1\right)+𝑑\left({𝑣}_1,{𝑣}_2\right)$.

  This amounts to saying that detours don’t help. This is ensured by the fact that we’re defining $𝑑$ in terms of shortest paths.

But we’re kinda almost there? We’re just crucially missing that symmetry, & for good reason. Therefore, although $\left(𝑉,𝑑\right)$ is not a metric space, it is a quasimetric space, which is defined by throwing out exactly that symmetry axiom.

Finding shortest paths

We’re defining $𝑑$ in terms of shortest paths, but how do we actually find those shortest paths to begin with? For this problem, our intuition already gives us everything that we need to know. In particular, the BPQer already intuitively understands this simple algorithm:

1. If you’re on a higher floor than the destination, then down-jump until you’re on the same floor as the destination.
2. If you’re on a lower floor than the destination, then walk directly to the upward portal that’s on your floor. Take the portals upwards until you’re on the same floor as the destination.
3. Walk directly to the destination.

Here, “walk directly” is taken to mean either uninterrupted walking purely leftwards, or uninterrupted walking purely rightwards.

Generating riddle lists

Randomly generating a riddle list sounds pretty simple: just pick any 4 distinct riddles out of the 28 riddles, in any order. But we’re going to go further by assuming that this process is uniform, meaning that all possible riddle lists are equally likely to be chosen. This makes the probability of choosing any given riddle list equal to $\frac{1}{𝑛}$, where $𝑛$ is the total number of distinct possible riddle lists.

But what is $𝑛$, exactly? If we briefly ignore that each riddle list is an ordered list, then counting the lists is simple: it’s just the binomial coefficient:

Once we take into account that the lists are actually ordered, we see that each unordered list can be made into $4!=24$ possible ordered lists. Thus, we have $𝑛=\mathrm{20 475}\cdot 24=\mathrm{491 400}$, which is roughly half a million.

491 400 is a lot, but it’s also small enough that we could, if we were so inclined, brute-force calculate the total travel distance for every one of those riddle lists…

Constructing the graph concretely

Before we can get carried away with dumb computer programs that do stupid things, though, we need to make the aforementioned graph less abstract. To do this, we can associate each vertex with a position within some coördinate system, & also associate a speed with each mode of travel.

For the purpose of extracting positions, I actually made a full-sized panorama of the Library Room very carefully stitched together using only screenshots of the game:

Wow… Cool… 🤓

We’re going to put the origin at the very top-left corner of the image, and then have values along both axes increase as they move away from the origin. Our x-values will be in pixels, but y-values will simply be floor numbers: 0 for the top floor, 1 for the next floor down, all the way down to floor 3 at the bottom of the map. The result is that each coördinate will be an ordered pair of natural numbers, but the two axes will not be comparable, so the pair is heterogeneous.

For the purpose of extracting movement speeds, I took a short screen recording with OBS Studio at a smooth 60 FPS. By analysing this video, I extracted the following times:

• From spawnpoint to Jonas bookshelf: ≈11.917⁢ s.
• From Jonas bookshelf to just above clock bookshelf: ≈14.000⁢ s.
• Down-jump: ≈0.800⁢ s.
• From clock bookshelf to vase bookshelf: ≈6.900⁢ s.

Based on these times, I estimate that it takes roughly ≈800⁢ ms to down-jump from one floor to the floor immediately below it, & roughly ≈8⁢ ms to walk a single unit (pixel) of horizontal distance.

Assigning a time to teleporting is more difficult for the reasons discussed above, but we’ll actually still be able to accurately account for teleporter delay. I used a teleporter in Henesys to analyse the teleporter delay, & found that when holding down ↑, the delay is exactly 600⁢ ms.

The force of a brute

I painstakingly recorded the fairly exact in-game positions of every vertex of our digraph, & wrote a Python script that does the riddle list brute-forcing for me. It defines a function called d that is the previously-discussed quasimetric function. I’ve left some comments in there to hopefully make it readable, so be sure to check out the script (bpq_stage_3.py) here if you’ve any idea how to read Python code! [:

Although I wrote it in Python, the script manages to run in about half a second on my machine. Not bad! As we take a look at the results, remember that all these values are just for the walking involved, and thus do not take into account NPC chat delay (much of which is simply due to networking latency), nor the time that it takes to read & jot down the riddles.

Let’s start with the expected value: ≈64.012⁢ s. And how about a five-number summary?:

Wowza… The minimum & maximum values are pretty extreme! The worst-case scenario is about 8.7 times slower than the best case! In particular, the best case is:

1. Empty.
2. Clock.
3. Ladder.
4. Newspaper.

And the worst case…:

1. Ludmilla.
2. Destruction.
3. Sophilia.
4. Beachball.

2:21 of just walking is pretty brutal!!

But for good measure, let’s plot this bad boy in a good ol’ histogram!:

(See: riddle_list_plot.py.)

Hmm… Looks vaguely Gaussian, but it’s pretty rough & spiky. Plus, it’s lopsided, leaning more towards lower times… a bit right-tailed, if you will. For reference, I’ve used dashed orange vertical lines to mark the minimum, median, & maximum.

Other sources of event currencies

Reward structure

Raffo

Because this event’s raffle didn’t have any MCPs nor CSs/WSs, the only raffle prizes that most players actually cared about were the pink capes (PAC/PGC) and the three chairs. These things were largely not present in the raffle pulls, & as a result, overall raffle engagement was relatively low.

There were actually some neat Hallowe’en equips in the raffle pool, however. We got things like:

• Pumpkin Basket, everyone’s favourite goofy box-opening weapon.
• Broomstick, which has SPEED on it (for the islanders), & also it’s cute.
• Pumpkin Spear, which is like the Fish Spear, but… with a punkin, instead of a fish. This makes it really super accurate, so it goes pretty hard. Great swashie weapon!
• Pumpkin Lantern, which I personally was very excited to see. It’s cool, it’s got a jack-o’-lantern, & it hits like a truck. This two-handed sword is an incredible gish weapon, & even if you ignore its MATK, it still puts almost any other level 30-ish melee weapon to shame.
• Talking Witch Hat. I don’t like this hat. I also refuse to elaborate.
• Spider Venom Earrings are more generally notable, because they can — if you get really lucky — be up to +2 allstat clean. They’re untradeable, but I still like having the chance to get these bad bois on some of my characters, & I’m sure many other people feel the same way.
• Worn Halloween Badges are kinda like untradeable Warm Mufflers, but… OK yeah, just worse than Warm Mufflers. That being said, they do look cooler, & I couldn’t get a goddamned Warm Muffler to save my life in this past CNY event…

Pointz

Pumpkin Coins go towards cosmetics like NX equipment, face/hair coupons, event pets, etc. But they also go towards Zombie Army Rings. 310 Pumpkin Coins gets you a clean ring (which is OoaK & untradeable), & 130 gets you a Zombie Essence that you can use to scroll it. A fully-scrolled Zombie Army Ring, then, costs 700 Pumpkin Coins.

Many people will want to retry their ring until they can make a perfect one; that is, passing Zombie Essences on all three slots. Because Zombie Essences have a 75% pass rate, spending 700 Pumpkin Coins immediately on a ring & three essences has a…

…Probability of resulting in a perfect ring. That’s not fantastic, as it means that you’re more likely to fail than to succeed.

Of course, if you’re dead-set on a perfect ring, then this isn’t how you do it. As soon as a zEss fails, you drop the ring & buy a new one. You might wonder how many “attempts” it will take to succeed. Well, the maths works out basically the same: the geometric distribution gives the number of attempts necessary for a success^[4], & the expected value is simply:

But with this strategy, what an “attempt” entails can vary. What we really want to know is: how many Pumpkin Coins do you spend to succeed?

What? Don’t look at me. I’m not a mathematician. I don’t fucken know.

Well, okay. We know that it takes $\frac{64}{27}$ attempts on average, so… how much does an attempt cost, on average? In terms of Pumpkin Coin expenditure, there are only three possibilities per attempt: that the first zEss fails, that the second zEss fails, or that we buy all three zEsses (regardless of whether the third one fails). These correspond to costs of 440, 570, & 700 Pumpkin Coins, respectively. The expected cost per attempt, then, can be calculated like so:

Assuming that we’re justified in simply multiplying these two expected values together(‽), the expected cost of making a perfect ring is then:

1 447.4 is no small quantity of Pumpkin Coins, & worse yet, that’s just the average! With a little bit of bad luck, it could take a whole lot more than that. 🙂 Fun!

That feels like a lot of probability theory, though. Maybe I can check my work by writing another dumb computer program to do it the dumb the Monte Carlo way…? Nothing a few more lines of Python can’t do:

from random import random

costs = []
for _ in range(10_000_000):
    spent, passed = 310, 0
    while passed < 3:
        spent += 130
        if random() < 0.75:
            passed += 1
        else:
            spent, passed = spent + 310, 0
    costs.append(spent)

That was easy, I guess. And it looks like I did my probability theory correctly, because the average value spit out by this little program is just about ≈1 447.4. Of course, it’s all Monte Carlo, so the actual exact average is randomly different every time that you run the program. But it’s close enough to confirm that I was on the right track.

Really, storing each & every individual result in one big fat list is not the best idea here. Instead, we can just use an associative array (Python calls these dictionaries) to record how many times that we see each particular outcome. And with that, well… let’s plot it?

(The orange dashed line is the average. See: zring.py.)

Kind of a weird-looking bar chart! As you can see, you’re not actually that unlikely to need an amount of Pumpkin Coins so large that you’d better be spamming some BPQ for most of the event, just to make your ring. All of those little bars add up!! When I initially tried generating some random trials, the seventh one resulted in spending 4 120(!) Pumpkin Coins. Also note that I’ve arbitrarily cut the chart off at 6k Pumpkin Coins; if I don’t do this, then the extreme cases are way out there in the tens of thousands, & the chart isn’t so readable anymore.

My impressions of the event

In many ways, this was a pretty PQ-heavy event. If you want raffles, your best bet is likely BPQ spam, & if you’re not in level range for ordinary PQs, then it’s definitely your best bet. Even if you are in level range for, say, OPQ, BPQ spam is so powerful that — with a good party — you can generate many more raffles than the OPQers. And if you BPQ spam with a good party & you skip the bonus stage consistently, you can generate even more Pumpkin Coins than the OPQers — albeit giving up raffles in the process.

Although it might sound like I’m describing an event full of mostly just BPQ spam, taking Maplers away from the Maple World and into the world of the Haunted Mansion & the Butler’s mind, this is certainly no Retro World event (see the “In which MapleLegends is swallowed up by an arcade cabinet” section of pt. cv of this diary). I actually liked this event quite a bit because of its emphasis on, you know, MapleStory, & because of its reasonable raffle prize pool. It’s nice to have a raffle that just offers some nifty stuff, instead of the promise of getting rich quick.

With the exception of mostly just my Viclockers, I rarely felt like I really “had to” spam BPQ. Of course, for my characters like rusa (my darksterity knight) and axisaxis (definitely not my character… trust…) who are ineligible for ordinary PQs, but also don’t do bosses frequently enough, I did go and farm a few BPQs for Pumpkin Coins — & a handful of raffles, to boot. Ordinary PQs were lively as a result of awarding Pumpkin Coins & raffles, as well as the event just generally being noob-friendly, which is always good to see. Bosses were emphasised not only by making them award Pumpkin Coins, but also by — at least, supposedly… — increasing MCP droprates for the duration of the event. And of course, the extra +10% bonus EXP is nice as well!

Footnotes for “All Hallows’ Eve”

1. [↑] MapleLegends does not have the Sake Bottle. Then again, the Sake Bottle is beginner-only, whereas the Toymaker Hammer has no job requirements whatsoever.

2. [↑] This guide somewhat erroneously refers to the Haunted House as the Haunted Mansion, whereas here I will be using Haunted Mansion to refer to a location distinct from the Haunted House, as consistent with the in-game nomenclature.

   The guide is archived here.

3. [↑] See the “Are you a robot?” section of the previous entry of this diary.

4. [↑] Sometimes, the convention of “failures before the first success” is used for the geometric distribution.

Islandry

Prior to the rollout of the Hallowe’en event, I spent some quality time on the Victorian Archipelago. Slaying golems at FoG, of course! On my Vicloc dagger spearwoman d34r!! Who hit level motherfucken one hundred!!!:

Triple digits, baby!!! Ergoth is crying & pissing himself rn!!!!

But before I even had the chance to go & stab Skeleton Daddy™ in his despicably sexy skeleton face, the Hallowe’en event was already upon us.

Now, as mentioned in the “Raffo” section above, this event’s raffle was offering Worn Halloween Badges. Of course, the Archipelago naturally offers no pendants whatsoever, so the prospect of getting a permanent pendant on my Viclockers — after getting screwed out of a Warm Muffler by this past CNY event — was sounding very juicy indeed. Unlike the aforementioned CNY event, this event puts no particular limitation on how many raffles that you can effectively turn in per unit time — other than your ability to, uhm, spam BPQ (see the “The Butler Party Quest” section above), of course.

So you’ll never guess what happened…!

…I spammed the everloving fuck out of BPQ!! Like, I mean a lot of god-damned BPQs!!! 31 clears to get the I Heart the Masked Gentleman! chair? Done & done.

Now, I’m turning in like… one or two dozen raffle tickets at a time here, like an absolute lunatic. And you’d better believe I’m getting some stuff. And things. Lots of things, really. But no Worn Halloween Badges…

I got a Flame Chair…?:

Pretty cool, actually! And made by none other than Slime (OmokTeacher, Thinks, JumpQuest, Slimu, Slimusaurus)!!

But I was genuinely beginning to worry that level 100 was too high to have any significant chance of getting these damned pendants from the raffle. I double-checked & triple-checked that Worn Halloween Badge was indeed on the list when I talked to Kaleido-Spinner to ask about which items were in the prize pool. I even went into the MapleLegends Discord™ server (that’s how you know I’m desperate…) to ask if anyone managed to raffle the badges on level ≥100 characters. Shout-out to 39daph (daph39, PQMuIe) for confirming — even including photographic evidence! — that it was indeed possible.

Well, fast-forward to very many hours of BPQ later, and I’ve turned in 73 raffle tickets. Seventy-three. Just on this one character. And BPQ is a bit stingy with those tickets sometimes, too. Especially because most of this BPQing was done before the BPQ update… But you know what they say: (seventy-)third time’s the charm!:

My god, finally!! And of course, the stats are… not good! Lowest possible STR, only average DEX… But hey, it could be worse: at least the LUK is maxed (gotta get that +0.5 WACC!) and so are both the WDEF & the MDEF! A great little upgrade for d34r. [:

You’d’ve never guessed, but I actually got a second Worn Halloween Badge later in the event, from those filthy FoG raffle tickets! And naturally, the stats were somehow even worse!! [':

But okay, surely I got some other Vicloc raffle spoils, including from the raffle tickets that I accumulated on my clericlet d33r in the process of getting her a Masked Gentleman chair… right? Right. Like these two egg yolk Full Moon Night Chairs!:

For whatever reason, these things got super expensive, so I’m probably sitting on like 300M worth of chairs in the above image…

Oh, and the Worn Halloween Badge wasn’t even the only priceless equipment upgrade that d34r got from this raffle! I just got a lot more of these Spider Venom Earrings than I did the badges…:

Stromk!!

And of course, with the weight of all of this new gear, d34r headed back to þͤ olde FoG. Including this FoG session that I did with fellow Viclocker bak2monke (snakewave, oley, yeIlo, VAPORWAVE)!:

Because 100 just ain’t enough:

😎

Ever still to haunt my house

(ℹ️ See the “Down the rabbit-hole” section of pt. cvii of this diary for the previous instalment in this series.)

The one for whom the Fates write ’cross the sky,
Whose puzz’ling journey leaves us asking why
No peril & no threat doth e’er she meet,
Without again & time doth Death she cheat.

O axis, axis, come once more to now;
Such med’ling in the past you’ll disavow.
And yet you’ll find the town of toys a bore;
To distant lands adventure, girl, once more.

Masteria^[1], the land so often curs’d,
Whose hist’ry in which thou art not well vers’d;
Her tales Ossyrian^[2] quartos fail to tell,
That ever & anon strange woe befell;
A vast land haunted, tide by tide consum’d,
Until her sunken body were exhum’d;
It fell & rose, or so the story goes,
’Tis here the Fate of axisaxis flows.

So close, yet far, to cities forests are;
The New Leaf outskirts axis reach’d by car.
A wind-swept wetland forest of Bent Trees,
Their lifeless branches leering in the breeze;
As turns her head, so too the plants do peek,
She’s thankful that they do not also speak!

But now, dear axis came across a carr,
In which she learnt of wandèrer’s devoir:

She’d ne’er seen so peculiar a place
To rant & rave as this here Tombstone’s face.
But of her worries, epigraphs were least,
For it was now that axis face’d the beast!

On bright pink steed the headless fiend did ride,
His collar pop’d, but nothing was inside.
Instead, the vacant horseman’s head was squash —
His face so glew, ’twas all in flame awash!

No sooner did the pumpkin man come forth,
Than someone fearlessly came walking north.
A woman clad in garb quite odd — but bold —
She spake, & words of partnership she told.

Engross’d in battle both of them became,
So Killian^[3] hurl’d forth his “head” of flame;
An awesome fight the dullahan advanc’d;
Betwixt the fire & the blades they danc’d.

The mad equestrian spat one last curse:
“If you’ve my head, you’ll ſuffer even worſe.
In hell of hells I’ll ſee you both, my friends;
But we ſhall never, ever make amends!”

No sooner did the pumpkin man so speak,
Than empty clothes fell suddenly so weak.
His flame extinguish’d, crumpl’d to the ground;
And so gan ceann^[4] was nothing but a mound.
LeFay^[5] was spook’d — her lover’d met decease —
And left unlikely allies there in peace.

So axis & miss Taima^[6] parted ways,
The both aware that in the coming days,
A party would be held at mansion near,
To celebrate All Hallows’ festive cheer.

In front the mansion, Old Man Tom did warn:
“In this accursèd house, a jinx was born.
If you’ve got any sense left in your head,
Regard this house with great respect & dread.”
But to wise words, dear axis paid no heed;
And to the invitation she’d accede.

The costume party was a hit indeed,
And with her appetite the food agree’d.
But when the mansion axis did explore,
She got a whole lot more than she ask’d for.

She found a little girl tuck’d in her room,
And on her face was nothing else than gloom.
With axis as her witness, she did cry:
“Where went my mum & pop? Oh why, why, why?!”

With manner quick did axisaxis say:
“Oh don’t you worry, it will be okay!
I’ll go & find your mum & pop, I swear;
I’ll be right back, so please just stay right there.”

Around the Haunted Mansion axis snuck,
Where she would stumble ’cross some things by luck.

Into a secret door she found her way,
And found herself in giant chamber grey.
So well chaotic was the scenery —
The only sounds were of machinery.

Then saw a man at work with fierce intent;
On making toys the fellow was hell-bent.

“Who goes there?” bark’d the man, his body tense.
“’Tis only axis — I pose no offence.”
“Wherefore, then, art thou present in my mill?”
“I met a girl whose wish I must fulfil:
The poor soul’s starv’d of both her parents’ love…”
“Is it perhaps my daughter thou speak’st of?”

’Twas not long ’fore the man name’d Jonas said:
“Outside this factòry I will not tread —
Not till I make the perfect doll for her;
So priceless that she’d nothing else prefer.”

The passion in that voice, she had not miss’d;
And thus she felt no choice but to assist.
Around the spacious workshop axis went,
And in each walking doll she made a dent.

No simple toys these wickèd puppets were;
The magick springs inside them made them stir.
As axis bop’d them one by one, they turn’d;
And for retaliation each one yearn’d.

But all the cloth she’d gather’d was for nought;
The doll so made, he threw out on the spot.
“It’s not the perfect doll, why can’t you see?
Go bring a portrait of her back to me.”

So on behalf of Jonas axis went,
To find the portraiture for which he sent.
She met the room where picture frames abound,
And search’d until the daughter’s face she found.

But all the glass was smash’d, the frame was broke,
And when ’twas brought to Jonas he then spoke:
“For use as ref’rence, this here art will do;
But sadly, there is something else askew.
This gadget here & instrument right there,
The both are in acute need of repair.
If thou canst find the spring for my device,
To make the perfect doll it will suffice.”

Around the workshop & the house she went,
Into each chimney & out of each vent.

The mansion vast, its parts put up a fight;
It was poor axis whom they try’d to smite.
The pipes & valves she wrestl’d past in spite,
To crawl her way around with all her might.

At last, she found the resting place of springs,
Plus nuts & bolts, & many other things.

But when she brought it back to Jonas now,
He took it wordlessly, & creas’d his brow.

“Your daughter’s visits you must please allow!” —
Said axis: “You do this, for her? But how?”
Said he: “What did I say? What have I told?”
And in his steely voice was only cold.

Dejected axis then return’d to her,
And once again the crying did recur.

The Haunted Mansion was quite large indeed —
Perhaps a sign of avarice & greed.
But ’twasn’t that the poor girl here was lost;
Her father’s shoulder, cold, left only frost.

But then the party’s host approach’d to speak;
From ’hind his mask his shifty eyes did peek.
Mysterious & eerie was his voice;
To axis he gave this outlandish choice:
“I hate to see my guests feel so forlorn;
But I’ve a remedy for what you mourn.
Give me a chance, & I’ll have thee reborn;
Yet first, there’s something ’bout which I must warn:
The things you’ve seen & done here on this day,
My magick aims to make them go away.”

Upon this proposition axis thought,
And judg’d that to accept she really ought.
With snap of fingers & a tip of hat,
She was made retroactive, just like that.

Now with determination fill’d anew,
Once more she went to see what she could do.

Although the father may have been detach’d,
Perhaps the stepmother was better match’d.
In bedroom lavish, under vaulted roof,
Lyudmila^[7] stood, amidst an air aloof.

Mistaking axisaxis for a maid,
Lyudmila bark’d commands, requesting aid.
Forgetting where her earrings she had laid,
To fetch her jewell’ry pearls she axis bade.

Suspicious beasts the earring hoarders seem’d;
With virid ghosts the western hallways teem’d.
As axis struck each Poltergeist, it scream’d;
The earrings from these phantoms she redeem’d.

But Mila^[8] could not even spare a glance,
Before requesting music for the dance.
An inauthentic party it would be,
If there weren’t even dancing here to see!

So axis left to seek a worthy sheet —
Of music — one that surely would move feet.
And in the Libr’y many sheets she found,
There’s just one catch: they’re flying all around!

So breathlessly the folios she fought,
Till fin’lly she a score of music caught.

She place’d the sheet on grand piano’s tray,
Awaiting someone else to come & play.

But when to Mila axis did arrive,
Another busy task she did contrive:
“The guests are parch’d, they need something to drink;
A bev’rage Hallowe’en, what do you think…?
Who am I kidding? What do I expect?
Too much from someone of your intellect…”

For Mila, any poor folk is a dope;
Her insults & her sneers left little hope.
In desperation, axis would avail,
And with a “fellow” Maid, brew’d special ale.

With pumpkin beer distributed to guests,
Poor axis had enough of Mila’s tests.
So with the grumpy wife she now did plead:
“Your daughter! She’s of parents’ love in need!”

But these concerns Lyudmila did dismiss:
“That pamper’d rotten child is remiss.
If all she wants is playing with her dolls,
For all I care, she’ll stay within those walls!”

At wits’ end, axis beg’d again the host:
“Your magick tricks I’m now in need of most!”
So once more work his magick did the man —
And this time, axisaxis had a plan.

From dear Sophelia^[9] she did request:
“Could you please choose the doll that you like best?
I’ll show it to your father — he’s upstairs;
I’m sure he’ll come & see you if he cares.”

She doll in hand, with Jonas met againe;
The puppet seen, his face collaps’d in pain:
“She kept that oldest puppet all these years?”
His voice did fail, so muffl’d by the tears.

He look’d now ’pon his fact’ry with disdain:
“These horrid monsters I’ve made all in vain.
This workshop is my prison & my bane;
I shall — I will! — see my creations slain!”

So axis seeks these blasted, cursèd toys,
And with her pinwheel, each one she destroys.

The rocking horses from a nasty dream,
Becloak’d in night — more dang’rous than they seem!

The giant creepy puppets made her start! —
Then flaming pinwheel tore them all apart.

The jacks in boxes had their springs all wound,
And when she struck them, they made quite a sound.

Each Psycho Jack with music box was arm’d;
The song they play’d made axis quite alarm’d:^[10]

X:1
T:Psycho Jack-in-the-box theme
C:traditional song; lyrics by deer
L:1/8
M:6/8
I:linebreak $
K:C
V:1 clef=treble
(G, | C2) C D2 D | (EGE) C2 (G, | C2)  C DDD    | E3 C z (G, |
      C2) C D2 D | (EGE) C3     | !>!A z z D2 F | E3 C z z   |]
w: For|you your dad- dy|made all these toys, e-|quip’d with en- gines of|Die- sel. His|soul your fate so|neat- ly de- stroys.|“POP!” GOES THE|WEA- SEL!!|

Once axis had all of the playthings fell’d,
Return’d to him, whose head in hands he held:
“I cannot thank you quite enough, my friend;
My only solace is to see this end.
I have no things to give, no things to lose;
But you may take my hammer, if you choose.”

But axis had one question left at hand:
“What happen’d to her? I don’t understand.”

“I’d answer you — I really, really would —
But thinking on it has done me no good.
So all that I can do is fin’lly rest;
Retire to my long-forgotten nest.
And when there’s no life left, not e’en a mouse,
I’ll be here ever still to haunt my house.”

Footnotes for “Ever still to haunt my house”

1. [↑] /məˈstɛ(ə̯).ɹiˌə, mæ-, -ɪ.ə/.
2. [↑] /ə(ʊ̯)ˈsɪɹ.j(ə)n/, not /ə(ʊ̯)ˈsɪɹ.i.ən/.
3. [↑] /ˈkɪ.li.ən/, Anglicised from Irish Cillian /ˈcɪ.lʲi͜ən̪ˠ/. Note the addition of a syllable. That’s his actual name in MapleStory, I didn’t make that up. Cillian is just a normal Irish name, but the Irish notion of “headless horseman” is called dúlachán /ˈd̪ˠuː.l̪ˠa.xaːn̪ˠ/ — whence English dullahan /ˈdu.ləˌhɑn/.
4. [↑] /gən̪ˠ caːn̪ˠ/ “headless”.
5. [↑] /ləˈfeɪ̯/. Eccentric sorceress, disciple of Ridley, namesake of the LeFay Jester & the LeFay Jewel, and Killian’s lover. Transmuted into a pink horse by the Alchemist, & doomed to roam the Phantom Forest with her lover on her back.
6. [↑] /ˈtaɪ̯.mə/, Anglicised from Japanese 大(たい)麻(ま) ⟨taima⟩ /tä.i.mä/ “hemp; cannabis”. Note the loss of a syllable.
7. [↑] /l(j)udˈmi.lə, l(j)ʊ-, -ɪ.lə/, Anglicised from any of various Slavic languages.
8. [↑] /ˈmi.lə, ˈmɪ.lə/, standard short form of Lyudmila/Ludmilla/etc.^[7] From PSl. *milъ “dear, sweet, cute, beloved, pleasant”.
9. [↑] /sə(ʊ̯)ˈfi.lɪ.ə, -li-/, with variant spellings ⟨Sophelia⟩ & ⟨Sophilia⟩.
10. [↑] The tune here is a very slightly modified version of the tune traditionally used for jack-in-the-boxes: “Pop! Goes the Weasel” (Roud 5249), which is of unknown origin — probably three or four centuries old. The tune gained its current name, & became a standard in Anglophone countries, only in the 1850s. The first version with lyrics probably(?) only had the lyric “Pop! Goes the weasel”. Later versions, however, gave it full verses, of which there are many variations (such as this one!).

The d00rs of perception

In the “Hysteria in the area of Masteria” section of the previous entry in this diary, we saw my clericlet d00r accompany her partner in crime, ironman assassin GiIf (MiIf, Dakota), on some juicy Masteria questing adventures. Does that mean that it’s time for another episode of The Adventures Of GiIf & d00r™??

Yes! No. Maybe…

Victoria & defeatoria

But before we get there, I tried — & perhaps, failed… — to do some solo questing on our dear d00r. You see, unlike my ironman companion who wasn’t eligible for KPQ even at levels 21〜30, I did KPQ from level 21 to level ≈29 or so. This meant missing out on the Victoria Island questing that I’d usually do around those levels. If you’ll remember, I quested to the bottom of 21, KPQed to ≈29, & from there, pretty much just completed every goddamn Lūdus Lake quest that I could find.

So I thought that it was high time to head back to Vic. Over at Perion’s East Rocky Mountain III, I hunted Axe Stumps for their Plant Fossils & their Firewoods, for Winston’s Fossil Dig-up and “Thawee Fixing the Roof” &c., respectively:

For Winston once again, I did “Planting Trees” & “Shawn the Excavator’s Request” by heading into the Excavation Site:

And, whilst I was at it, I started Winston’s The Old Tree questline. After hunting those Very Dry Branches from Dark Axe Stumps, it was time to look for a Stumpy.

Unfortunately for me, ≥2 hours of holding mapowner in East Rocky Mountain V later, & I had yet to actually slay a Stumpy myself. Repeatedly getting area bosses stolen from me is really not worth my time — not that waiting for them to spawn was worth my time to begin with — and so I gave up & did something else.

The adventures of GiIf & d00r in the Haunted Mansion

Fast-forward a number of days, & the Hallowe’en event began. As mentioned in the “The Haunted Mansion quests” section above, this event came with its own exclusive questline. So, naturally, I teamed up with my time-honoured teammate to tackle this truly tangled tree of tough tests & trials:

A bit of careful re-typing, copying, & pasting later, we were enghostificated. Phantasmalised, if you will:

The Gatekeeper may be a big scary green goblin dude, but he nevertheless maintains a healthy fear of ghosts. So, he let us inside the Masked Gentleman’s party, out of respect. When we entered the Haunted Mansion, we ran into one Jonathan (Gumby, daggerknight, Edward, Kimberly) — a fellow ironman!:

Well, I’m not ironman, but you know. Sometimes I pretend like I’m an honorary member. [':

Four-way tragedy

My name is Jonas

Jonas was indeed in his workshop, hard at work intently crafting doll after doll that, to us, looked quite identical. Jonas, however, thought that each individual doll was separately unfit, time & time again taking a good look at each one, & then throwing it into his incinerator. When we spoke with him, he insisted that he was “only trying to make the perfect doll”, & challenged us to make such a doll ourselves. Although we only had a vague idea of what an ostensibly “perfect” doll would even be like, we did find some Busted Dolls from which we could salvage parts:

I knew that we’d end up needing one hundred or so of these cloth pieces, so GiIf & I spent some time here farming the Busted Dolls for two hundred in total. Using the basic doll skeleton that Jonas gave us, we could attach these pieces to the doll one by one:

And with each additional ten pieces, the Rag Doll started to look appreciably different:

We brought our well-completed ragdolls to Jonas, but he quickly threw them in the incinerator with the rest. We had made imperfect dolls.

Although we were certainly frustrated to see our hard work thrown away so easily, the look on Jonas’s face was one of even greater frustration. What could he possibly be doing wrong? If he, Jonas Prendergast the renowned toymaker, couldn’t make the perfect doll, then who could…? For better or worse, GiIf & I were going to find out the answer to that question.

But first, Jonas desperately turned to us for help. Unsure of what to say, I suggested that perhaps he didn’t have the right tools for the job — a good workman blames his tools, right…? Jonas recalled that, indeed, there was one tool in particular that he’d previously lost the use of. The apparatus itself wasn’t missing, but when Jonas had taken it apart for repairs & then put it back together, he’d lost track of its crucial spring in the process.

In the course of searching Jonas’s workshop, GiIf came across a low fissure in the eastern wall. Upon further inspection, we discovered that that wall was the only thing separating that side of the workshop from the mansion’s massive chimney. Enlarging the fissure even slightly had already compromised a sizeable section of the wall, and so GiIf & I decided to simply crawl our way in — perhaps the spring had fallen down here…?

After falling on our butts more than a few times, we managed to explore this chamber of the chimney. Near its apex, we finally found the spring that we were looking for.

Well… GiIf found it. I was not so fortunate…

Don’t worry; the chest respawned.

We brought the spring back to Jonas, who was delighted to have recovered it — or rather, we assume that he was delighted, as he immediately went to work fixing his long-disrepaired apparatus with a deep sense of preoccupation. Of course, we’d come here on Sophelia’s behalf, & so we asked Jonas if he could only come downstairs to see his daughter. But Jonas was more than preoccupied; he was resolutely dead-set on making the “perfect doll” before anything else. GiIf & I sensed a tinge of anger in his voice, & so thought it best not to push him any further.

Would you like to try again?

Stepmum

Quickly realising that the results were no better than last time, we joylessly made our way back to Sophelia. Sophelia was now even more dejected: Jonas wanted a portrait of her? Over his own actual daughter‽

At first, GiIf & I were a bit confused. Wouldn’t the in-house pianist, of all people, know where the sheet musics were stored? When we found the library ourselves, however, we immediately realised at least one issue:

Half of the books were on their shelves, like in any ordinary library, but the rest were flying about the room of their own accord^[2] — & at dangerously high speeds, to boot. After taking a few dozen whacks here & there — both at the books, & from them — we recovered some sheet music that was in readable condition. We placed the sheet music on the grand piano’s stand, in the mansion’s foyer, & returned to Lyudmila.

And oh boy, is this thing an absolute unit. GiIf & I spent no small amount of time climbing all over this oversized orange gourd, carving at it with improvised weapons of squashtruction. Once we’d made enough Pumpkin Juice for the whole party, we brought some back to Lyudmila. Lyudmila was, however, not easily impressed, & drinking the juice didn’t seem to noticeably improve her mood.

The grateful & the ungrateful dead

Now at wits’ end, GiIf & I complained to the Masked Gentleman. None of this made sense: everyone to whom we spoke seemed to refuse to leave the rooms that they were already in; Jonas’s inhuman obsession with the “perfect doll” had left the ultimate object of that obsession — his own daughter — in neglect; & moreover, why was a mysterious masked man with the magical might to modify memory, managing the mansion & moderating the meeting of so many multitudes of amused masses?

The portrait depicted Jonas, Sophelia, & a man who was presumably Peto. Sophelia could be seen tightly hugging a doll. When we brought the portrait to Sophelia, she was pleased to be reminded of good times, but insisted that she needed more candies to keep her energy up whilst she mused on the portrait. Although we were concerned that the child was now surviving on little more than a steady diet of Halloween Candies, we nevertheless obliged.

Ghost-hunting

At first, the chimney looked just like it had every other time that we’d been in it: a massive multi-chambered stone chimney, complete with pressurised gates, & soot marks that were impossible to clean off. After a few minutes, however, I heard GiIf emit a squeal. The ghost that she found was indeed a spooky scary ghost, although it did also have a face bearing an uncharacteristically jovial expression:

Joey?

Doing my best impression of Sophelia — aided by having a nearly identical copy of her voice — I greeted the ghost. The ghost in question indeed looked like a small boy, dressed all in black. But he was not as responsive as I’d hoped. After a few seconds of awkward silence, the ghost asked: “do you know who I am?”. Naturally, I really didn’t know, but thought that I could get away with “of course; you’re Joey, my favourite doll”. Unfortunately, the ghost was unconvinced by my dissimulation — his name was not Joey. The spectral boy’s tone as he rejected me outright was cold, bitter, & sinister.

Perturbed, I returned to GiIf & the Masked Gentleman. The gentleman reassured me that I could have another chance at speaking to the ghost boy, if only we could make a similar potion to turn us into Peto doppelgängers. With the assistance of the ghost of Peto — who, somewhat bizarrely, looked nothing like Peto — we did exactly that. With some more cloth, T put together a surprisingly well-made doll — impressive handiwork indeed, for a ghost — & we brought it to the Masked Gentleman.

Rain so loud that ghosts retire

With enough Peto potion for the two of us, GiIf & I both turned into frail old men:

We asked Jon for some protection, as we didn’t just look like infirm old codgers; our bones really did creak. Slowly & carefully making our way around the mansion, we spoke with the Butler, with Jonas, & with Lyudmila. Each one, in turn, expressed some form of horror at the understanding that Peto had returned.

When GiIf & I went back into Sophelia’s room one last time, we no longer heard the girl’s quiet sobbing. The room was empty.

The Masked Gentleman stopped us once again as we left the room. Now that we knew the story of Peto & the Prendergasts, and had brought it to a close ourselves, the mansion could no longer be called “haunted”.

As a reward for our efforts, the gentleman gave us each a very large gobstopper. These Hallowe’en sweets were so large, in fact, that they made for perfectly good chairs:

This was all a lot more than I expected from a Hallowe’en party. Nevertheless, ’twas no match for GiIf & d00r. 😎

Grindin’ With d00r

Although I very much enjoyed celebrating Hallowe’en, I was starting to worry that GiIf was getting much too far ahead of me in levels. I considered EPQ & OPQ, but ultimately decided that I had a better idea — plus, it’s not like I would be particularly useful in those PQs, anyway…

As part of my first determined effort to “just grind” on d00r, I seized upon the fact that, unlike my Viclocked clericlet d33r, d00r ain’t so area-locked. Indeed, she’s not area-restricted at all! So I figured it was time for a “world tour” (if you will…) of the undeads of MapleStory. Well, the grindable ones in my level range, at least…

Starting with Master Chronoses at Lost Time ⟨2⟩!:

Master Chronoses are most notable for dropping helmet DEX 60% scrolls.

Mostly because I managed to catch a handful of GM buffs that were the source of much of my grinding motivation, I ended up not moving from this spot…

…I wouldn’t want to waste my precious GM buff duration by travelling, would I? Plus, they drop LUK Crystal Ores as well…

At this point, I’ve farmed up a whopping six of those DEX 60% bad bois! Not bad!!

Next time, though, I’ll be sure to do a real “world tour” — & with EPM tests, to boot. Which map will win?? Tune in next time for more d00rventures!! 🧡

Footnotes for “The d00rs of perception”

1. [↑] In the Haunted House, this NPC is called Sophilia, & so too do the associated monsters — Sophilia Doll Ground & Sophilia Doll — use this spelling. In the Haunted Mansion, however, this NPC is known as Sophelia (note the ⟨e⟩).

   Like many things in the Haunted Mansion, the Sophelia NPC is not accounted for on MapleLegends’s Library, which is why I less-than-correctly link to the Sophilia NPC here. I do this throughout the text, linking to Haunted House counterparts when the Haunted Mansion versions do not exist in the Library.

   The Haunted House–Haunted Mansion correspondences, where they differ in name and/or ID, are as follows:

   House	Mansion
   Sophilia	Sophelia
   Jonas Prendergast	Jonas
   Ludmilla	Lyudmila
   Edmunds*	Butler
   Voodoo	Busted Doll
   Hoodoo	Destroyed Doll
   Glutton Ghoul	Green Phantom
   Purple Flying Book	Rideword P
   Orange Flying Book	Rideword Y
   Blue Flying Book	Rideword B
   Nightmare	Fire Steed
   Headless Horseman†	Pumpkin Knight

   • *Named Steward in later versions of the game.
   • †Not actually found in the Haunted House, but nevertheless found in the Phantom Forest more broadly.

2. [↑] These books are the Ridewords P, Y, & B. See above: ^[1]. These seem to be the only bizarrely-named NPCs/monsters in the Haunted Mansion. I thought that perhaps rideword was an actual word, but it appears not. Results from searching the WWW for rideword are all from Ragnarök Online, another Korean MMORPG that was released around the same time as MapleStory. In the original KMS, these are 라이드워드 ⟨raideuwodeu⟩ /ɾɐ.i.tɯ.wʌ.tɯ/; presumably from English ride /ɹaɪ̯d/ + word /wɜ(ɹ)d/. This would make it a pseudo-anglicism.

Accipiter

W—what if it were time for some Bossin’ Wif rusa™? Would you think that was cool…? 🥺👉🏽👈🏽

Okay, great. Let’s do it. Starting with… Ravpap, I guess? My darksterity knight rusa hasn’t actually fought either of these bosses in quite a while! And, you know, what else am I going to do with my Pink Bean buff…?

I joined STRginner Taima (Yunchang, Tacgnol, Boymoder, Girlmoder) & moderately INTelligent corsair Lvl1Crook (Level1Crook, xXCrookXx, Macer) in the ol’ Rāvvin’:

…And in the ol’ Pappin’:

I will say, for Pampered Lattice in particular, it is really quite nice to be able to simply cast stance on myself, instead of only getting stance when I’m surrounded by a weird blue bubble…

扶桑

Thailand & Lūdibrium are cool & all, but what about Japan? Alongside shadowers Harlez (Gock, Murhata, VigiI) & SinghNinja, I headed to a shopping mall in Roppongi [六(ろっ)本(ぽん)木(ぎ)] in the year 2102 to smack a big dumb robot cop with an oversized chandelier.

Unfortunately for everyone involved, however, the aforementioned robot cop did what robot cops do best: catastrophically malfunction. And I do mean catastrophically. The Royal Guard let us know, in no uncertain terms, that today was not the day to take on CBPQ:

I got a handful of good video clips of this bot being utterly broken, but this one in particular was notable because it managed to cast seduction whilst it was clearly stunned. Believe it or not, I survived this seduction despite my Hero’s Will being on cooldown, & despite getting seduced on top of this chunky lad who likes to touch attack me for 15k〜16k damage a pop. I somehow just got enough “MISS”es to make it work. 🤷🏽‍♀️

With CBPQ off the table, we instead did an approximation of an NT1 tour. Starting with Burger Moth…:

…Moving on to Nibble Lung…:

…And then Namie…:

…Where I embarrassed myself by activating my Zerk, and then swiftly having a pet autopot failure and IRL pet trying to climb onto my lap at the same time. Oups! 😭 Thankfully, although the sudden absence of HB was bad news for SinghNinja, they managed to survive for the rest of the fight with some clever manœuvring & a bit of luck. Phewf!

We also did some 2nasing alongside hero ChiR:

The highlight of this particular run was that I brought along my pugilist tarandus to provide SI to myself & ChiR, but nevertheless forgot to poke 2nas before it died… So no 2nas clear for her… Yet……

A little break from bossing

It wasn’t all bossing, though. Here I am with Lvl1Crook, doing some monk-slapping at 7 F!:

I also did the full Haunted Mansion questline (see the “The adventures of GiIf & d00r in the Haunted Mansion” section above) with MiIf (GiIf, Dakota)! Here we are doing BFJOLPQ (Big Fat Jack-O’-Lantern Party Quest, of course) together:

This “PQ” can be played as a party of one, which is how it seems to nearly always be done. It is a good deal faster as a real party, though.

’Tail

But would it even be a real episode of Bossin’ Wif rusa™ if there were no Horntailing? I did a run or two hosted by Diggy, as sed target, during which I had a bit of an 🤔interesting🤔 moment involving HT’s tail…:

Right, so, you can see Vorizon dead in this image. They’re a bishop, but they’re only in my party because they managed to die to the tail earlier, & this was a 10-Mapler run anyway, so we just moved them over to our party in anticipation of my sed bish Resurrecting them immediately after tail was to die.

Thanks to an unfortunate series of coincidences, I ended up getting right-sedded into tail whilst I was unroped & my Will was on cooldown. I’ve done many runs as sed target now, & I’m cautious enough that this really rarely happens — indeed, I think the only other time was one that I captured in a very juicy video clip that you can find in the “⁧روسا⁩” section of pt. ci of this diary. Like last time, I actually managed to survive somehow, despite not being assisted by my bishop. Unlike last time, unfortunately, I didn’t manage to record a video clip; all I have is the above screenshot.

Bizarrely, I kind of just waltzed right past the tail! By the time that my seduction wore off, I was somehow at around ≈20k HP remaining. I then almost immediately got seduced again, but this time on the “how the fuck did you get over there?” part of the map. This is why, in the above screenshot, you can see me rope-sedded rightward of the tail.

Pink Bean & jelly sandwich

But what about PB, tho? The bean of the pink?? The pale red legume??? The fuchsia pea????

In the «El haba rosada» section of the previous entry in this diary, I did my first ever runs of PB, as hosted by xBowtjuhNL (PriestjuhNL, BuccjuhNL) & Nonon (“Milky”, Ettyne, Lorenae). To give enough context for the reader, and especially to hash out my own perspective on & understanding of the PB fight, I went into some detail about what said fight actually entails. Before I talk about PB even more, I’ll quote myself to reiterate an important disclaimer:

⚠️NOTE:⚠️ In this section, I mention PB strategies & things related to them. The strategies represented here are applicable to MapleLegends specifically, even more specifically to the version of MapleLegends that is current at the time of this writing, & even more specifically to the 30-Mapler runs that I’ve personally joined. They are not necessarily applicable otherwise.

With that out of the way, I’d like to add a few bits of info that I neglected to include previously. The first two are courtesy of xBowtjuhNL:

• As mentioned previously, doing a “right bird last” (which I referred to as “left man first”) kind of strategy does mean eating seduces for basically the entire duration of stages 4 & 5. However, it’s worth noting that until you actually start attacking right bird, its seduces are generally just left-walk seds, which are significantly less dangerous than seds involving jumping, thanks to the touch-attack hitboxes of the statues not extending low enough to hit PCs that’re on the floor of the map. Moreover, a sed not being rightward implies that it won’t cause its victims to inadvertently take touch damage from Ariel during stage 5.

• One strategy that has been explored in xBowtjuhNL & Milky’s runs that I didn’t mention — because I’ve never actually seen it myself — is a “protected parties” strategy. The basic form of this is having two protected parties, with three bishops each. This allows for 100% uptime on Holy Shield within those parties. The protected parties of course enter the map first, to soak up as many seds as possible. With five bishops, one of the bishop positions can be occupied by a shadower who can then provide Smokescreen.

  However, this leaves all other parties bishopless (& thus Shieldless), which is problematic for the DPM output of those unprotected parties. Having shads in the unprotected parties can help to ameliorate that issue, in addition to providing some protection for seds that connect with unprotected party members (especially when left man is still alive, thanks to its stuns & 1⧸1s). Largely for these reasons, xBowtjuhNL & Milky have found the multi-bucc strategies to be largely superior.

• I later realised that my confusion over PB main body’s resistances was somewhat unwarranted: its resistance to “neutral” extends not only to elementally-neutral magic damage, but also to elementally-neutral damage in general, hence its 50% resistance to nearly all physical attacks. This would seem obvious from the fact that certain clearly physical attacks (e.g. Inferno) are nevertheless not elementally neutral (fire-typed, in this example), & thus physical attacks must interact with elemental profiles just as much as magical attacks do.

  However, I am still confused on this point: if this were an accurate & general explanation of all of PB’s damage resistances, then we’d expect for paladins to have an enormous advantage over all other physical attackers: because PB only resists neutral, fire, & ice (& has no immunities), they could simply use Divine/Holy charge (or really, any charge that isn’t either fire or ice) to bypass all of PB’s resistances, & thus deal damage normally. Unless pallies are just incredibly OP in PB, & I somehow didn’t know about it, this can’t be correct. There must be some mechanic that causes all physical damage — regardless of its element — to suffer resistance against PB’s body.

Also, in «El haba rosada», I referenced xiaoyaoz’s Pink Bean Animations Guide (archived), which covers much more than just “animations”, in spite of its title. I’d now also like to point to Milky’s brand-new PB guide that is specifically targeted towards active bishop players, but that is — naturally — also potentially useful for non-bishop PB runners. Unfortunately, this guide appears to only be posted on Discord™ at the moment, so I can’t link to it here. Maybe in the future…

Alright, that’s enough PB nerd shit for now. Last time, I was lucky enough to get my very own Timeless Pendant from PB! But I still had to level it up, so I promised to show the fully-levelled version this time. Well, here it is…:

Hmm… Not bad! I was really mostly looking for an upgrade to my Horntail Necklace, & I suppose this fits the bill: an additional 3 STR & 2 DEX, effectively the same AVOID & WDEF, and an additional 12 MDEF! I’m still not sure if I want to throw it away so that I can have a chance at a better one… 😅

In any case, I did continue PBing with this same group, & in one such run, we got really spoilt!!:

FS and MW30?? At the same time??? 🤑🤑🤑

In another run, I managed to level up at the end! To level 184!!!:

Whew. You know, this isn’t easy, considering that I only get ≈2% EXP from an entire PB run… Come back in five months for when I level up again! 😊

And we weren’t finished getting spoilt, either. In yet another run, we got two MW30s!!:

Wow. Rich…

Alright, how about even more PB content?:

Wait, what? This isn’t PB. That is clearly NMM. What’s going on here?

As it turns out, the easiest approximation to pinning PB’s main body is… Hamstringing NMM. So it’s a neat way to help train buccs for doing just that. And they needed HB, so I tagged along as well!

Part of the training involved Kasio practising barrel cancelling for the first time:

As a bucc player myself, I admit to being basically ignorant of these devious hacks used to bypass the delays intentionally built into many of the pirates’ attacking skills. I pretty much just press the attack that I want to use. I know, how naïve… But the fact that Kasio — a level 200 buccaneer — was largely unfamiliar with how to pull these tricks off at least made me feel a little better about myself. 🥲

After a while of using NMM as little more than a plaything — a seemingly bizarre thing to do — we finally decided to actually slay the thing:

Dyadic

In other news, I also did a duo Zaqqūm run with Harlez!:

Unfortunately, neither of us can reach arm 2. 🥲 Don’t worry — enough jump-Crushering & BStepping later, and it did die.

Unfortunately, I also died! Actually, I died when there were still seven or eight arms alive, but thankfully was revived by MineyBish (Minethrower), who was there (with a few mules) mostly to loot helms. Believe it or not, this marks the second time, in this entry alone — at least, as far as I know… — that I’ve died as a result of a bug with active Zerk. I managed to get video clips clearly demonstrating both instances, although I’m not including them as part of my diary for two main reasons:

• In order to actually demonstrate the bug, most of the game screen (1366×768, in my case) has to be shown. This would result in video files way too large to include here, as I already have compression troubles with much smaller video, as it is.
• They’re just not very interesting to watch. Bug occurs; rusa becomes gravestone. That’s pretty much it.

The first of these buggy Zerk deaths actually occurred during a PB run, where I was actively Zerked, at ≈12k HP, & got one-shot by an attack. Now, there’s a lot of stuff going on at any given time during a PB run, especially because this was during stage 4 or 5, so I sent the clip to Harlez to get a second opinion. Harlez said that it was clearly a bug.

In the Zaqqūm case, it happened a little differently. Whilst actively Zerked, I got hit by an attack. This nominally lowered my HP to 1 (which is correct), but also had the following other effects:

• My pet auto-HP failed, which is not part of the bug, but is nonetheless worth noting as part of what happened.

• I could not manually use any USE items.

  At the time that this occurred, I distinctly remember thinking to myself over the course of ≥1 000⁢ ms that Zerk had bugged and was going to kill me (& my micro-prophecy was soon fulfilled, of course…). I simultaneously mashed my Shift ⇧ key (where I put my main HP potions) as quickly as I could, getting in at least half a dozen or so presses before I finally bit the dust.

  Looking back at the footage, you can clearly see me stop attacking almost immediately after getting hit the first time. I then appear to be standing there doing nothing — but actually, mashing the HP potion button — for what seems like ages before I finally die.

• My HP bar glitched out. Although it did correctly display the number “1” as mentioned previously, the bar itself was havin’ a real goofy time goin’ all over the place, variously indicating that I had anywhere between ≈10% and ≈40% HP remaining.

The next attack that connected with me, then, did me in.

In the “DK rev. 3” section of pt. xcix of this diary, I listed a great number of both pros & cons of MapleLegends’s 3^rd version of the DK throughclass — the version that introduced active Zerk, & the current version at the time of this writing. The really important part of that essay is the “The unforgivably ugly” section, but the technical details prior to that section do touch upon several bugs with the implementation. Since then, one or two bugs with DK rev. 3 have been introduced & fixed — including the Beholder bug mentioned in the “My first Pink Beanpressions” section of the previous entry. Or not fixed…

But the point is this: in addition to being fundamentally broken on several levels as explained in my “DK rev. 3” essay, it’s clear that active Zerk is just a buggy mess that I can’t even use selectively, because I never know when it’s going to blindside me with a nasty bug. I’ve since removed Zerk from my keybinds entirely. Good riddance.

The good news is that Zaqqūm went down like a chump — the caveat being that it only dropped one or two helms, unfortunately for Miney.

But Zaqqūm wasn’t the only Mayan chunk of granite that I duoed with Harlez. …Actually, it was. But it wasn’t the only boss that we duoed. Here we are duoing Dunas Unit (effectively 2nas’s first body):

Although we were indeed duoing, Positivity (Disney) tagged along to get themself a 2nas clear.

And here we are, beating up 2nas himself:

Unfortunately, Harlez didn’t make it to the end of 2nas, & when Positivity tried to come Resurrect her, they… didn’t make it, either:

So I ended up soloing the remaining twenty or twenty-five percent of the damnable rascal Dunas II:

😎 I even managed to get a 2nas clear for tara this time…

Sorts, kinds, sets, & types

Hellow, & welcome to more PQing with my mfing PQ mules!! In the previous episode — the “Keming” section of the previous entry in this diary — my DEX page SGQ/KPQ mule kinds did some KPQing with friends, & then sold some very ✨unique✨ “KPQ service” to Danger (Hampy) & his quartet of hooligans: bandits Peruggia & Bonnot, and gunslingers Jennings & Worley.

LPQ gang rise tf up

With the level 31 → 35 gap bridged, it was time for the hooligans to do some LPQing. No “LPQ svc” here, though. Just some good ol’ fashioned LPQing with friends! I joined a number of runs on my DEX brawler LPQ/EPQ mule sorts:

Stage 4 of LPQ consists in slaying Shadow Eyes from Another Dimension & Dark Eyes from Another Dimension (whose names cannot be witnessed in-game) for the Passes of Dimension that they drop. In the image above, we can see Peruggia learning that Passes of Dimension aren’t the only things that these mysteriously disappearing eyes drop! Eyedrops, if you will!!

In this same stage, Peruggia also learnt from fellow bandit & notorious brigand Hitodama (Taima, Boymoder, Tacgnol, Girlmoder, deerhunter) another bit of eyedrop-related trickery:

Can you really call yourself a “thief” if you don’t even Steal?

In the next stage, I gave away some 💸free mesos💸! Wow!:

You’re welcome! F2

And here’s the gang, featuring brawler Murhata (Harlez, Gock, VigiI), taking on Lūdibrium’s most handsome whale:

When we were particularly starved of mages, we had the idea to try doing the mage portal of stage 5 The Gunslinger Way™. Is there such a thing as The Gunslinger Way™, though? Because of their access to the Wings skill, gunslingers are capable of completing the four “normal” (read: neither mage nor “thief”^[1]) portals extremely quickly — much faster than any non-gunslinger. For the mage portal, though, the idea is slightly different: because completing the mage portal consists in down-Teleporting through a series of equally-sized platforms (& hitting the boxes to obtain passes along the way), the gunslinger ostensibly could briefly walk off the edge of one platform, almost immediately Recoil Shot, & thus land on the immediately lower platform.

This is actually not the first time that I’ve been in an LPQ party that has tried this. Nevertheless, I couldn’t actually remember why it doesn’t (or does??) work. I didn’t mind giving it a go, though, so that’s what we did:

Right, so the actual obstacle to completing the mage portal sans Teleport is at the very beginning (read: the top of the map): there’s a top platform that is at least as wide as the map is, & with down-jumping obviously disabled — as that would trivialise the mage portal so that anyone could do it — there’s simply no way to get to any other part of the map, sans Teleport. Boooo…

So, maybe the mage portal really is as uninteresting as it seems: it’s no more than a Teleport check, similar to the right-hand portal of MPQ’s 4^th stage. But what about everyone’s favourite LPQ stage: stage 8?

Unlocking the mystery of LPQ’s 8^th stage

(Alternate title: “Counting is harder than you think”.)

Love it, hate it, or both, you can’t deny that stage 8 requires an unusually high level of coördination between party members. This stage is easily recognised as KPQ’s 4^th stage — which is itself a version of the preceding two KPQ stages on steroids — on steroids. And by “steroids”, I really mean that there are more possible combos:^[2] $\left(\genfrac{}{}{0ex}{}{4}{3}\right)=4$ in KPQ’s 2^nd stage, $\left(\genfrac{}{}{0ex}{}{5}{3}\right)=10$ in KPQ’s 3^rd stage, $\left(\genfrac{}{}{0ex}{}{6}{3}\right)=20$ in KPQ’s 4^th stage, & $\left(\genfrac{}{}{0ex}{}{9}{5}\right)=126$ in LPQ’s 8^th stage. This notably makes stage 8 the only reason why LPQ cannot be completed with a party of four or fewer (LPQ-sans-stage-8 would be duo-able).

Lexicography

Because of the analogy with KPQ stages 3 & 4 (not stage 2, because stage 2 does not explicitly number its slots), players are naturally tempted to solve LPQ’s 8^th stage in the same way: by traversing all possible combos in lexicographic order. Because the slots are neatly labelled, each with its own natural number, the usual ordering of the natural numbers yields the aforementioned lexicographic order on all possible combos. KPQ’s 4^th stage has only 20 possible combos, so I’ll list them here for illustrative purposes:

In this notation, each decimal digit represents one element of the combo. The order in which the digits are presented within a combo is strictly irrelevant from the game’s perspective, but by always listing them in ascending order, we can get the aforementioned lexicographic order.

In the jargon of formal languages, we are treating each combo as a string of length $𝑘$, where $𝑘$ is the number of slots (in KPQ’s 4^th stage: barrels) that have a PC occupying (read: standing on) them. We then require that all strings have their symbols in sorted order, & that a given symbol appears at most once within a given string (in other words: each string is actually a well-ordered set of cardinality $𝑘$). The alphabet from which the symbols are taken is the set of all positive integers that are less than or equal to $𝑛$, where $𝑛$ is the number of slots. This framing is what justifies my use of the phrase lexicographic order.^[3]

Shortest Hamming paths, flip graphs, & more maths

Ordering the combos is great, but doesn’t tell us how to actually play out those combos in-game. Thankfully, because we chose a kind of lexicographic order, the strategy is naturally suggested to us: we simply assign one PC to each index in the string. A given PC, then, only has to move themself during the transition from one combo to the next iff the corresponding strings differ w.r.t. the integer that they have in that PC’s index. For example, when transitioning from 123 to 124, the PCs assigned to the first two indices do not move, whereas the PC assigned to the third & last index must move from 3 to 4.

We get the vague idea that the fewer PCs who must move to transition from one combo to the next, the better. This amounts to minimising the Hamming distance from one string to the next. In the “For the Goddess” section of pt. cv of this diary, we saw a similar problem in the context of OPQ’s On the Way Up stage. When solving On the Way Up deterministically — using rogerhub’s method — each combo is a bit-string of length 5, & they are ordered as a Gray code. This allows for traversing all $2^5=32$ possible combos with there being a Hamming distance of exactly 1 from each combo to the next, which is clearly optimal.

Unfortunately, the usual binary Gray codes (which we’ll later refer to as RBGCs) aren’t gonna cut it for the class of problems in question. We can generalise Gray codes to arbitrary radices (that is, radices other than 2)^[4], but we’ll end up with combos like 333 that don’t make sense because they have more than one occurrence of the same symbol, as well as combos like 312 that are unsorted, & thus unwanted duplicates of their sorted counterparts (123, in this example).

We can instead think of each combo as a bit-string of length $𝑛$, where a 1 indicates that the slot is occupied, & vice versa. We then have a perfectly ordinary binary Gray code over the set of bit-strings of length $𝑛$, but this includes bit-strings that are unwanted because they don’t have exactly $𝑘$ 1’s in them. In the context of KPQ’s 4^th stage, this includes combos like e.g. 23 & 1234, that are invalid because too few & too many barrels are occupied, respectively.

In “For the Goddess”, we saw that rogerhub derived his Gray code method without knowing of the existence of Gray codes per se. Instead, he arranged the 32 combos as the vertices of a 5-cube^[5], & then found a shortest possible traversal — moving along the 5-cube’s edges — of all of said vertices.^[15] This is an example of a more general kind of mathematical object called a flip graph. A flip graph is a graph in which each vertex is some geometric or combinatorial object, & two vertices are connected by an edge iff they can be transformed between one another via some elementary operation called a “flip”. In rogerhub’s case, the “flip” is literally a flip, particularly of a single lever at the top of the map. In our case, a “flip” is the movement of exactly one PC from one slot to another.

In both rogerhub’s & our cases, what we’re looking for is virtually the same: a Hamiltonian path (a path that visits each vertex exactly once) in the flip graph. We could go even further & ask for a Hamiltonian circuit in particular. However, this is only theoretically an optimisation for cases where you fuck up & have to start over again, after you’ve just supposedly — but not actually — checked all possible combos. But, you know, if that happens, you’re already screwed anyway…

But it’s not looking so good. For starters, not all graphs even contain a Hamiltonian path. Worse yet, the problem of determining whether a graph has a Hamiltonian path — the well-known Hamiltonian path problem — is infamously $\mathrm{𝐍𝐏}$-complete in the general case, which is computational complexity speak for “good luck with that”^[6].

Don’t give up just yet, though! We’re not working with the general case, which means that we have a few particular affordances. For one, in the largest practical case for MapleStory — LPQ’s 8^th stage — we “only” have 126 vertices. That’s still a lot, but nonetheless better than “any number of vertices”. Furthermore, to quote Petr Gregor:^[7]

As an informal remark, many flip graphs are usually highly symmetric, in the sense they are vertex-transitive and sometimes Cayley graphs. In this context, the following conjecture makes us optimistic about the existence of combinatorial Gray codes for various classes:

Conjecture 1 (Lovász^[8]): Every connected vertex-transitive graph has a Hamilton path.

“Distance-2 Cyclic Chaining of Constant-Weight Codes”

That brings us to a particularly relevant 1973 paper by Donald T. Tang & C. N. Liu: “Distance-2 Cyclic Chaining of Constant-Weight Codes”^[9]. To understand the meaning of the title of this paper, we can revisit a remark that I made above:

We can instead think of each combo as a bit-string of length $𝑛$, where a 1 indicates that the slot is occupied, & vice versa.

Using the convention that we read bit-strings from right to left — that is, from the 1’s place, to the 2’s place, to the 4’s place, & so on — we can use the example from Fig. 1 to illustrate this representation:

Notice that some — but not all — transitions in this lexicographic sequence are the kind of transitions that we want. For example, the first two combos in Fig. 2 are 000111 and 001011, which only requires one PC to move (viz. from barrel 3 to 4). But in the bit-string representation, the Hamming distance between these two combos is not 1 as it was with the string-of-slots representation in Fig. 1; instead, it’s 2. This is because we have a bit complemented from 1 to 0 as the PC leaves the barrel, and then another bit complemented from 0 to 1 as the PC arrives at another barrel. This is the meaning of “Distance-2” in the paper’s title.

“Cyclic Chaining” basically means that this paper is concerned with Hamiltonian circuits, rather than just any Hamiltonian paths. And finally, “Constant-Weight” refers to the fact that each bit-string has the same Hamming weight; in the KPQ stage 4 example, this means that a constant number of barrels is occupied in any given combo.

The paper is originally concerned with analog-to-digital conversion, but also mentions the exact problem that we have:^[9]

The cyclic distance-2 chaining of constant-weight codes has applications in A/D conversions as well as in combinatorial problems involving the exhaustion of 𝑘-out-of-𝑛 combinations.

For our purposes, the main utility of this paper is that it not only proves that such Hamiltonian circuits exist, but also proves that simply taking an ordinary binary Gray code on bit-strings of length $𝑛$, & then removing bit-strings that aren’t of the desired weight, produces exactly such a Hamiltonian circuit! It’s that easy?!?

Well, not so fast. And I mean that literally; one of the issues is that naïvely generating the binary Gray code & then extracting the desired subsequence is quite inefficient. But also, the proof presented in the paper is not exactly trivial. I read the paper until I understood the proof — which took quite a bit of time & effort, even though the proof itself is only a few paragraphs of text! I’ll attempt to give a simplified sketch of the proof that doesn’t actually prove anything, & if you want a real proof, then check out the paper yourself.^[9]

Proof sketch

(NOTE: You can skip this proof sketch section, especially if it hurts your head to read anywhere near as much as it hurt mine to write.)

We’ll use ${𝐺}_{𝑛}$ to denote a binary Gray code whose bit-strings each have length $𝑛$. It’s clear that taking the subsequence of ${𝐺}_{𝑛}$ that contains exactly those bit-strings of weight $𝑘$ yields a sequence — which we’ll call the induced sequence — containing all the desired bit-strings exactly once. What we’re worried about is that they might be in the wrong order. If we have two such length $𝑛$ bit-strings $𝑥$ and $𝑦$ — both of weight $𝑘$ — that are adjacent in the induced sequence, & the distance between them is more than 2, then we’re shit out of luck.

Notice that the distance between $𝑥$ and $𝑦$ has to be even, due to the same reasoning as in the previous section. Half of the distance is from 1 → 0 transitions (leaving a barrel), and the other half is from 0 → 1 transitions (arriving at another barrel). So if the distance between $𝑥$ and $𝑦$ is more than 2, then it must be 4 or more. In this case, we’d better hope that there are other bit-strings in the induced sequence that go in between $𝑥$ and $𝑦$, so that $𝑥$ and $𝑦$ aren’t adjacent. But what does “go in between” mean, exactly?

Because our induced sequence is supposed to be a Hamiltonian circuit, it’s cyclic. Thus, we need not one, but two bit-strings — which we’ll call $𝑢$ and $𝑣$ — to “go in between” them…:

…So that we have $𝑢$ sandwiched between $𝑥$ and $𝑦$; and we have $𝑣$ sandwiched between $𝑦$ and $𝑥$ (note the order). In the above diagram, the arrows are dotted because there may be zero or more other anonymous bit-strings that are omitted.

The standard construction of ${𝐺}_{𝑛}$ is as a reflected binary Gray code (RBGC). We take ${𝐺}_{𝑛-1}$, & concatenate it with a reversed (“reflected”) copy of itself. In the reflected copy, the bit-strings are the same, but are listed in the reverse order. Then, we prefix each element of the original ${𝐺}_{𝑛-1}$ with 0’s, & those of the reflected copy with 1’s. For example, to construct ${𝐺}_3$:^[15]

The result is that, in ${𝐺}_{𝑛}$, the most significant (read: leftmost) $𝑚$ bits repeat themselves exactly $2^{𝑛-𝑚}$ times in a row, before proceeding to the next configuration & repeating themselves exactly $2^{𝑛-𝑚}$ times again, & so on. We can exploit this fact to show that $𝑢$ and $𝑣$ really must exist, & must be sandwiched between $𝑥$ and $𝑦$, as in Fig. 3.

Let $𝑓$ be the number of “flips” (in the sense of “flip graph”) necessary to go from $𝑥$ to $𝑦$ (or equivalently, vice versa). Clearly, $2𝑓$ is the Hamming distance between $𝑥$ and $𝑦$, and $𝑓\ge 2$. Then, $𝑥$ and $𝑦$ are the same in $\mathrm{𝑛}-2𝑓$ of their bit-positions, & we can make a subsequence of ${𝐺}_{𝑛}$ by “freezing” (if you will) all of those positions to the values that they have in $𝑥$ and $𝑦$ (that is to say, elements of ${𝐺}_{𝑛}$ that differ from $𝑥$ and $𝑦$ in any of those positions are deleted). Given the “reflected” structure of ${𝐺}_{𝑛}$ that we saw in Fig. 4, freezing the values of ${𝐺}_{𝑛}$ in all but $2𝑓$ bit-positions produces another Gray code ${\mathrm{𝐺\prime }}_{2𝑓}$. I’ve put a prime symbol (′) on $𝐺$ to represent the fact that we’ve re-indexed the bit-positions; for example, ${𝐺}_4$ has the bit-positions $⟨4,3,2,1⟩$, and if we freeze position 2, we obtain ${\mathrm{𝐺\prime }}_3$ with the bit-positions $⟨4,3,1⟩$, which is a re-indexing of the more usual $⟨3,2,1⟩$ where 2 becomes 3, & 3 becomes 4.

If we look at the most significant two bit-positions of ${\mathrm{𝐺\prime }}_{2𝑓}$, we obviously know that $𝑥$ and $𝑦$ have a Hamming distance of 2 in those positions. But because these positions form ${\mathrm{𝐺\prime }}_2$, there have to be corresponding two-bit versions of $𝑢$ and $𝑣$ interleaved between them, similar to Fig. 3. For example, if the two-bit version of $𝑥$ is 00, & the corresponding two-bit version of $𝑦$ is 11 (this is a valid example because 00 and 11 have a distance of 2), then we definitely can’t have a Gray code like ⟨…, 00, 11, …⟩; that’s not a Gray code. We’d need something more like ⟨00, 01, 11, 10⟩, where 01 & 10 are our “two-bit versions of $𝑢$ & $𝑣$, respectively”, so to speak.

But that doesn’t necessarily mean that these “two-bit versions of $𝑢$ & $𝑣$” are parts of larger bit-strings of length $𝑛$ & weight $𝑘$, right? Well, remember that each element of ${\mathrm{𝐺\prime }}_2$ is — within ${\mathrm{𝐺\prime }}_{2𝑓}$ — repeated $2^{2𝑓-2}$ times in a row. If the two-bit version of $𝑢$ is ${𝑢}_2$, then somewhere within its $2^{2𝑓-2}$ repetitions, there is a trailing string of $2𝑓-2$ bits that looks like $\overline{{𝑢}_2}\mathrm{𝟷}\mathrm{𝟶}\mathrm{𝟷}\mathrm{𝟶}\cdots \mathrm{𝟷}\mathrm{𝟶}$.^[10] Thus, the full element of ${\mathrm{𝐺\prime }}_{2𝑓}$ — viz. ${𝑢}_2\overline{{𝑢}_2}\mathrm{𝟷}\mathrm{𝟶}\mathrm{𝟷}\mathrm{𝟶}\cdots \mathrm{𝟷}\mathrm{𝟶}$ — has a weight of 2 due to ${𝑢}_2\overline{{𝑢}_2}$ (a bit-string & its complement always have a combined weight equal to the length of the original bit-string), plus $\frac{2𝑓-4}{2}=𝑓-2$ for the trailing 10’s, for a total weight of $𝑓$.

$𝑥$ and $𝑦$ are of weight $𝑘$, & they clearly both have the same amount of that weight within the bit-positions that they share — thus, they have the same amount of that weight within the $2𝑓$ bit-positions that they don’t share. Because in the latter positions $𝑥$ and $𝑦$ are bitwise complements of one another, we can again exploit the fact that a bit-string & its complement always have a combined weight equal to the length of the original bit-string. We then see that $𝑓$ of the weight is in the non-shared positions, and so $𝑘-𝑓$ in the shared positions. Seeing as our $2𝑓$-length versions of $𝑢$ & $𝑣$ were each of weight $𝑓$, the full $𝑛$-length versions must be the desired weight $𝑓+𝑘-𝑓=𝑘$ once we restore the “frozen” shared bits of $𝑥$ and $𝑦$.

Now that we see that any two bit-strings in the induced sequence that shouldn’t be adjacent ($𝑥$ and $𝑦$) aren’t (because they’re dutifully separated by $𝑢$ and $𝑣$), we know that the induced sequence has the right order! Wow!! Kill me!!!

What’s a little naïveté?

Above, I said…

[…] one of the issues is that naïvely generating the binary Gray code & then extracting the desired subsequence is quite inefficient.

But how bad can the naïve approach be? Tang & Liu give this no more than an unsourced passing remark:^[9]

However, this straightforward method can be shown to be rather inefficient as $𝑛$ becomes large. This is because we are interested in only $\left(\genfrac{}{}{0ex}{}{𝑛}{𝑘}\right)$ code vectors out of $2^{𝑛}$ possible ones, and the ratio of $\left(\genfrac{}{}{0ex}{}{𝑛}{𝑘}\right)$ to $2^{𝑛}$ tends to zero as $𝑛$ grows indefinitely.

Does it, though? For some reason, I felt the need to clarify this. Especially because we’re comparing $\left(\genfrac{}{}{0ex}{}{𝑛}{𝑘}\right)$ to $2^{𝑛}$, and the former is of two variables, whereas the latter is only of one variable. I think I know enough to investigate this!

If we get rid of $𝑘$ by letting it be constant, then I have no doubt that $2^{𝑛}$ wins. But I’ve seen Pascal’s triangle before, & I know that the fat numbers are in the middle! So what if we let $𝑘$ also grow indefinitely, by setting $𝑘≔\frac{𝑛}{2}$? Well, the formula for $\left(\genfrac{}{}{0ex}{}{𝑛}{𝑘}\right)$ is usually stated in terms of factorials like so:

By using Stirling’s approximation, we can characterise the asymptotic growth of $𝑛!$ in terms of some other elementary functions:

We don’t need those pesky constants, though.

Much better.

I won’t bore you with the half-dozen algebraic steps necessary to get from this, with $𝑘≔\frac{𝑛}{2}$, to something useful. This is what you get:

…Which would imply that Tang & Liu were correct in their assertion, & in particular, the ratio approaches zero at the same asymptotic rate that $\frac{1}{\sqrt{𝑛}}$ does.

Even for relatively small $𝑛$, the difference is considerable. For example, at $𝑛≔9$ (the LPQ stage 8 case), we have 126 and 512 (ratio of ≈4.1). At $𝑛≔20$, 184 756 and 1 048 576 (ratio of ≈5.7). At $𝑛≔100$, the ratio is ≈12.6; at $𝑛≔\mathrm{1 000}$, ≈39.6; & so on, increasing like $\sqrt{𝑛}$…

Homogeneity & nearing perfection

Tang & Liu’s paper goes on to prove the correctness of an algorithm that is much more efficient than the naïve method. But it’s super confusing (show me the pseudocode!!), & this is a pretty old paper anyway, right?

Well, I chased a lot of random papers, scoured the interwebz, & spent many hours just kind of scratching my head, for the most part. I found several promising items, but as far as I can tell, the best result that I found was — perhaps unsurprisingly… — in §7.2.1.3 of vol. 4A (“Combinatorial Algorithms, Part 1”; 2011) of Donald Knuth’s The Art of Computer Programming (TAOCP). This section also credits the theorem proved by Tang & Liu — as sketched in the “Proof sketch” section above — to an earlier 1971 Ph.D. thesis by J. E. Miller, citing Tang & Liu as having independently rediscovered it, but I couldn’t find her thesis.

In any case, whilst reading this section, I did some more head-scratching, & also found some interesting things that actually made some sense to me (rare!), but that aren’t really relevant here. What are relevant, however, are TAOCP’s notions of genlex order, homogeneity, near-perfection… — oh, & some pseudocode for an algorithm!

Although the path of combos that we’re after isn’t ordered lexicographically, it is in a genlex order. This means that every string with a given prefix occurs in a contiguous sequence. This is more or less just a consequence of recursively building up the order; see, for example, the classic RBGC construction in the “Proof sketch” section above. It’s not super important, but it does have the nice property of making the order amenable to a trie, where each node of the trie represents a common prefix.

But we can do better than that. Going back to the KPQ stage 4 example, consider the transition 001011 → 011001. This has a Hamming distance of two, which is what we want: we can see that the transition only involves a single PC moving from barrel 2 to barrel 5. But if we translate this into the usual string-of-slots representation, we notice something a bit weird: 124 → 145. In this representation, the Hamming distance is still 2, which is not what we want; we want a distance of 1. This is because the PC who started on barrel 2 had to “cross over” the PC who’s on barrel 4, in order to go from barrel 2 → 5. We can simply refer to this as crossing, & we definitely want to eliminate it, if at all possible. TAOCP refers to a path that eliminates all crossing as homogeneous.

Believe it or not, we can tighten the homogeneity requirement even further, if we so choose. A near-perfect path is a homogeneous path wherein each homogeneous transition either looks like 01 ↔ 10 or 001 ↔ 100. In the KPQ stage 4 example, a near-perfect path would mean that every transition involves one PC moving from a barrel to another barrel, where the barrels’ numbers differ by no more than 2.

In an attempt to take into account how long it takes for a PC to go from one barrel (more generally, one slot) to another, we can define a function $\mathrm{\delta }\left({𝑠}_0,{𝑠}_1\right)$ that takes two slots ${𝑠}_0$ and ${𝑠}_1$, and gives the “round-trip” time taken to move from ${𝑠}_0$, to ${𝑠}_1$, & back to ${𝑠}_0$ again. This attempts to smooth over the “quasimetric” problem, similar to that encountered in the “Stage 3 theory” section above, caused by the fact that it may take longer to go from one slot to another than it does to do the same in reverse. Then, we can order the slots (denote the set of slots by $𝑆\stackrel{\scriptscriptstyle\mathrm{def}}{=}\left\{{𝑠}_0,{𝑠}_1,\dots ,{𝑠}_{𝑛-1}\right\}$) in a way that minimises the following sum:

If we do this, then our slot ordering plays very nicely with near-perfect paths!

You might be wondering why they’re called near-perfect paths, or you already figured out that perfect paths would only have 01 ↔ 10 transitions. This is clearly worthy of the name “perfect”, but there’s just one problem: such paths usually don’t exist, excepting some special cases. In particular, TAOCP proves “Theorem P”, which largely boils down to (ignoring some trivial cases) perfect paths only existing when both $𝑘$ and $\left(𝑛-𝑘\right)$ are odd. Ouf. This means that there are no perfect paths for KPQ stage 3 nor LPQ stage 8, but there are perfect paths for KPQ stages 2 & 4 — granted, for KPQ stage 2, it’s trivial. However, perfect paths are vastly more difficult to generate than near-perfect ones, even when they do exist, so we’ll leave the optimisation of KPQ’s 4^th stage for… another time… 😉

Al Gore’s rhythm

TAOCP attributes the first construction of near-perfect paths to Phillip J. Chase (1976), & so calls such a path Chase’s sequence. It then gives “Algorithm C”, for generating Chase’s sequence.

I did find a free software implementation of Algorithm C — along with many other such algorithms — first in the form of The Combinatorial Object Server, a website originally launched by Frank Ruskey, & later relaunched by Torsten Mütze, Joe Sawada, & Aaron Williams (all theoretical computer scientists). You can find its source code here. For many of the algorithms, including Chase’s, the COS uses the implementations provided by Jörg Arndt’s FXT library, written in C++ & licenced under the GPL v3.

For various reasons, I decided to just implement Algorithm C myself! Check it out here, at src/algorithm_c.rs!! I think that my implementation improves upon FXT’s in some ways, & it’s pretty easy to use safely from any Rust code. For example:

use algorithm_c::AlgorithmC;

let mut ac = AlgorithmC::new(6, 3);
while let Some(combo) = ac.next() {
    println!("{combo:?}");
}

Which prints the following to stdout:

000111
001011
100011
010011
011001
101001
110001
100101
010101
001101
001110
100110
010110
011010
101010
110010
111000
110100
101100
011100

There’s also a little testing function called sanity_check that I used to ensure that my implementation was correct. If we inspect the above example output, we can see that it’s indeed a near-perfect Hamiltonian path through the bit-strings of length 6 & weight 3. Notice, however, that Algorithm C makes no attempt to produce circuits; the Hamming distance between the first & last bit-strings is 4. As mentioned previously, though, this is fine for our purposes; we really don’t need a circuit.

One possible optimisation that sticks out to me is to not use the bit-vec library, or to at least have a version of the AlgorithmC struct that doesn’t use it. Instead, AlgorithmC can directly store a single primitive integral type like u64 — or even u128 — to use as a bit-string. But my current lone implementation has the nice property that it easily handles any values of n and t (the name that TAOCP uses for what I’ve been calling $𝑘$) that you could really dream of.

In any case, with a working implementation to generate Chase’s sequence, let’s take a peek at what such a sequence visually looks like for LPQ stage 8’s parameters ($𝑛≔9$ and $𝑘≔5$):

Wooooaaaah… kinda cool.

“Chase”ing the optimal path

Remember these bad boys?

Using Chase’s sequence straightforwardly — by assigning the boxes labelled ⟨“1”, “2”, “3”, “4”, “5”, “6”, “7”, “8”, “9”⟩ to the bit-strings from left to right, or from right to left — is… something that you can definitely do. In fact, if you ignore the practical difficulties of getting people to memorise, & to coöperate in the execution of, the sequence (which we’ll get to), it’s probably really fast.

But as a point of attempted(!) optimisation, in the “Homogeneity & nearing perfection” section above, I suggested minimising the following sum:

The idea is that, if our Chase’s sequence — or whatever sequence of bit-strings we choose — is near-perfect or perfect, then we’re hoping that movements between slots are very small: a distance of one slot (one bit-position’s distance: 01 ↔ 10) in most or all cases, & a distance of two slots (001 ↔ 100) when necessary. Thus, if slots that are adjacent are also very close to one another in the game-physical world, then the stars should hopefully align, & we’ll get something close to maximal economy of motion.

That being said, we can already foresee at least a few potential problems with this approach:

• The function $\mathrm{\delta }$ is already papering over the “quasimetric” problem (again, see the “Stage 3 theory” section above), causing some slot-to-slot transitions to look slower than they actually are, & simultaneously making the corresponding reverse transitions look faster than they actually are.

  For example, the box 6 → box 1 transition is very fast, assuming that the PC has enough JUMP (or Teleport) to do it. On the other hand, the reverse transition — box 1 → box 6 — is considerably slower, as it requires falling twice, at the very least.

• Even the “real” time taken to transition from one slot to another can depend on incidental and/or situational factors, including — but not limited to — the SPEED & JUMP stats of the PC, whether the PC has Teleport, etc.

• A path being near-perfect doesn’t itself imply that there are “few” two-slots’-distance transitions. If the path is perfect, then this concern is clearly irrelevant; but for LPQ’s 8^th stage in particular, perfect paths do not exist. If there are actually more two-slots’-distance transitions than not, then perhaps the “$𝑖+1$” in the above sum should instead be $𝑖+2$ (along with $𝑛-3$ as the final value of $𝑖$) — or worse, something even more complicated.

• Minimising the above sum amounts to solving a travelling salesperson problem (TSP). Because I’ve defined $\mathrm{\delta }$ to symmetrise distances, the TSP would be symmetrical; it should be noted, however, that asymmetric TSPs are also well-known.

  Unfortunately, the decision problem version of the TSP is $\mathrm{𝐍𝐏}$-complete (much like the Hamiltonian path problem mentioned in the “Shortest Hamming paths, flip graphs, & more maths” section above), making the TSP itself $\mathrm{𝐍𝐏}$-hard (more specifically, ${\mathrm{𝐅𝐏}}^{\mathrm{𝐍𝐏}}$-complete). The best that can really be done for an exact solution is something like using dynamic programming^[11] to arrive at the Bellman–Held–Karp algorithm (BHK), which puts TSP in $\mathrm{𝐄𝐗𝐏𝐓𝐈𝐌𝐄}$ (better than a brute-force search) — with a runtime of “only” $\mathrm{\Theta }\left({𝑛}^2\cdot 2^{𝑛}\right)$, but also an unfortunate space complexity of $\mathrm{\Theta }\left(𝑛\cdot 2^{𝑛}\right)$.

• Even if this approach manages to work out quite well, it’s difficult to believe that it would be absolutely optimal, in the same way that (✨hypothetically✨) brute-forcing all $\left(\genfrac{}{}{0ex}{}{9}{5}\right)!\approx 2.37\cdot {10}^{211}$ (LOOOOL) possible paths would be.

But let’s take a look at those two-slots’-distance transitions, shall we? It’s easy enough to count them for a specific case, so I did just that for $𝑛≔9$, $𝑘≔5$: there are 34 of them. That’s out of 126 − 1 = 125, because we have 126 vertices, and we’re making a Hamiltonian path that isn’t a circuit. 27.2% isn’t terribly high, which is nice — it’s definitely a minority. So we’re just gonna call it good.

The travelling LPQer problem

As far as solving a TSP goes, we do have the benefit that $\mathrm{\delta }$ is a metric. This metricity puts the problem more specifically in $\mathrm{𝐀𝐏𝐗}$, & even more specifically, it’s $\mathrm{𝐀𝐏𝐗}$-complete. That means that it’s sorta approximable, but not as well as e.g. something in $\mathrm{𝐅𝐏𝐓𝐀𝐒}$ (e.g. the knapsack problem).

There are plenty of good heuristic/approximation algorithms out there, but you know what? Fuck it. Nine boxes isn’t that many. I started writing a BHK implementation myself, got pretty far into it, & then… thought better of it. Unfortunately, although the Concorde TSP Solver is super fast & source-available, it’s also proprietary. 😓 So instead, I used python-tsp to make things easier. It provides a function called solve_tsp_dynamic_programming that does what it sounds like it does: the BHK algorithm.

Of course, I need to know how to actually calculate $\mathrm{\delta }$. More effort can be put into this, but for now, I decided to just base it on the actual distances, in pixels.^[12] We also want to make sure that we’re solving an open TSP, meaning that we don’t need a circuit — just a Hamiltonian path. Because the starting — & thus ending — vertex of the TSP circuit found by BHK is always the same, you can just make that vertex have zero distance to it from any vertex — maintaining the true distances from it to other vertices. Moreover, we don’t care where we start the path, so I just ran the solver once for each of the nine boxes. Is that “the dumb way”? I’m not sure. In any case, by running travelling_lpqer.py, we get:

⟨1, 3, 6, 7, 4, 2, 5, 8, 9⟩.

(That’s a total travel distance of ≈410 pixels, by the way!)

You know… I don’t know what I expected. This is very nearly the JMS order (which we’ll get to…), & that’s something that we really didn’t need a TSP solver for — but hey. 🤷🏽‍♀️

The only difference is in how we visit box 8, but it probably isn’t all that significant. I think that the JMS order is clearly a bit better for going forward through the path, but maybe the above order has the advantage of being based on a metric (read: not a quasimetric)…? I do think that it has fewer “awkward moves” when going in reverse.

Note that although the Hamming distance of this order from the JMS order is 3, the Levenshtein distance from the JMS order is only 2, because 8 is simply shifted 2 places to the right.

Plus, if better measurements of box-to-box distances — as measured in ms, rather than px! — are obtained, then travelling_lpqer.py can be modified to produce something really optimal. Or, it won’t make much of a difference…

組合せ数学

Speaking of the JMS method, what is that, exactly?

In the “Lexicography” section above, I suggested that naïve LPQers would make an analogy between LPQ’s 8^th stage and the combo stages of KPQ (especially stage 4). This gets you the so-called GMS method: 12345, 12346, 12347, 12348, 12349, 12356, & so on… The GMS method is certainly straightforward, but it has just one problem: it sucks ass.

That’s why no one uses the GMS method, & we all use JMS instead. It all starts with an ordering:

Here, I divide the order into four segments, using grey arrows to connect the segments to one another:

Thus, the full order: ⟨1, 3, 6, 7, 4, 8, 2, 5, 9⟩.

JMS’s order is surely the primary, or the only, thing acting as a stumbling block for those learning the method for the first time. Although the JMS order geometrically makes more sense than the label order, its discoverability is severely hamstringed by the rather frustrating fact that the boxes are labelled with numbers. As soon as people start saying “1 3 6 7 4 8 2 5 9”, the poor noobs may as well be reading a tyre tread. In reality, the numbers could just as well be colours, shapes, words, or animals, for all we care, so long as they allow for the verbal identification of each box.

So perhaps a different diagram is in order…

In any case, with a box ordering established, JMS becomes a strange amalgam of the classic lexicographic method & something similar to a Gray code construction. The JMS method can be described as follows:

(In reality, this algorithm terminates due to the stage clearing. Otherwise, the “Lexicographic transition” step eventually cannot proceed because there is no “lexicographically succeeding combo”.)

Due to the third step (“Lexicographic transition”), four out of the five PCs move only according to a lexicographic scheme. On the other hand, the “Cycle” step generally visits many combos for each single lexicographic item; in some ways, this is similar to e.g. the RBGC construction, where we “freeze” a subset of the bits — that is, a subset of the symbolic positions within each combo string — whilst we visit all combinations of the other non-frozen positions.

Of course, in some sense, even a naïve purely-lexicographic strategy has this vague form. But in such a case, there’s only really one “active” PC at any given time, & other PCs almost always move simultaneously with the “active”/“leading” PC. Moreover, because of the nature of lexicographic order, the PCs are constantly encroaching on the “active” positions, forcing time-consuming (read: many PCs moving arbitrary distances) “avalanche” resets any time that the active positions can’t be squeezed further, which is quite frequent.

The Gray-code-construction-like quality of JMS, then, lies in the fact that there are effectively two active PCs at any time — with the caveat that $𝑐$ is sometimes 1, in which case the cycler is effectively inactive — & in the fact that the cycler’s active positions are never encroached upon, so that the cycler is never involved in costly avalanches. And this is all in spite of JMS also ultimately relying on a lexicographic sub-regime, of course.

There’s also a nifty bonus that I allude to in the “Cycle” step of Algorithm 2: “[…], in any order, […]”. This means that if the cycler is, for example, cycling through boxes {1, 3, 6}, then they can follow the repeating sequence ⟨1, 3, 6, 6, 3, 1⟩, so that they never have to move at the same time that the other PCs do, & also generally reduce their own movement. Notice that this is the same “reflection” utilised in the RBGC construction. In reality, it’s actually more like the repeating sequence ⟨1, 3, 6, 6, 1, 3, 3, 6, 1⟩, but that’s just the quasimetric problem (6 → 3 and 1 → 6 being far costlier than 3 → 6 and 6 → 1, respectively).

None of this is to downplay the JMS ordering of boxes. The JMS order is undoubtedly a massive advantage that the JMS method has over its naïve GMS counterpart!

Looking ahead, looking bhead, looking chead, …

We might think to take the naïve lexicographic method & augment it with minor optimisations, due to the unwanted avalanching. Using KPQ stage 4 as an example once again, consider:

For now, we’re ignoring the fact that this sequence often instead begins with ⟨123, 126, 125, 124, 134, …⟩, due to how the barrels are physically placed.

This starting sequence looks fairly economical due to going 126 → 136 instead of the lexicographic 126 → 134. However, by continuing this pattern straightforwardly, we hit 134 & have nowhere to go, thus the unfortunate 134 → 145 transition. By looking further ahead, we can try to circumvent this little issue:

…But then we have that horrible 156 → 234 transition. Do note that these kinds of optimisations are certainly not unheard of in practice. For example, I fondly recall being introduced to the so-called “WMS” method of LPQ stage 8. The ⟨W⟩, of course, stands for Wamu, an LPQ mule who — amongst many other things — enjoys leading the JMS combos, spicing it up & throwing in a few optimisations towards the beginning.

However, in order to continue optimising our KPQ-stage-4 strategy in this way, we have to finally relinquish any claim that “we’re basically just going in lexicographic order, but with some minor optimisations”. We’re leaving the realm of mere mortals.

Woah! It’s a near-perfect path now! What the fuck? There’s only ever one PC moving from combo to combo, & they never move from one barrel to another barrel whose number is more than 2 away!

Of course, this is not a recommendation of this exact method. In order to actually optimise this, we’d at least need to re-order the barrels to reflect how difficult it is to go between them. Moreover, we learned earlier that KPQ’s 4^th stage has a lucky configuration — $𝑘$ and $\left(𝑛-𝑘\right)$ are both odd — that actually allows for a ✨perfect✨ path. Still, it’s interesting that we got from a totally lexicographic order to a near-perfect path just by progressively working around the unsavoury combo transitions.

Putting it all together, threegether, fourgether…

Now that we’ve investigated the JMS method & near-perfection, let’s see what we can get out of it. We’ve seen basically three methods of doing LPQ’s 8^th stage: GMS, JMS, & some combination of Chase’s sequence with an open-TSP-like ordering of the boxes. Wamu named WMS after himself, so I’m reserving the name dMS just as well, for this third option!! F3

For now, I’m using the box ordering generated via open TSP for dMS’s order: ⟨1, 3, 6, 7, 4, 2, 5, 8, 9⟩.

So how do GMS, JMS, & dMS compare? Well, we can use some of the measurements that we’ve already worked with, in an attempt to answer that question — or at least part of that question. We’ll count the Hamming distance along a path to be the Hamming distance in its string-of-slots presentation (e.g. 12345, not 000011111). We’ll count its slotwise distance similarly, but taking into account the absolute differences between slots, within that method’s canonical ordering. And pixel distances will be calculated in the obvious way.

Check out together.py for the script that produced these numbers. In particular, it contains this implementation of the JMS method:

def jms(n=9, k=5):
    """
    Returns zero-normalised combos, e.g. [0, 1, 2, 3, 4] instead of the in-game
    labelling [1, 3, 6, 7, 4].
    """

    # 1. [Initialise:]
    c = list(range(k))
    c.append(n)

    while True:
        # 2. [Cycle:]
        for j in range(c[1]):
            yield c[:k]

            if j < c[1] - 1:
                c[0] = (c[0] + 1) % c[1]

        # 3. [Lexicographic transition:]
        i = k - 1
        while c[i] + 1 == c[i + 1]:
            i -= 1
        if i < 1:
            return

        c[i] += 1
        i += 1
        while i < k:
            c[i] = c[i - 1] + 1
            i += 1

The JMS method can here be generalised by passing non-default values of n and/or k. More obviously, the GMS method can also be generalised to pretty arbitrary values of n and k, & so too can dMS, as it’s really just Chase’s sequence applied to a bespoke well-ordering of the n slots.

Welcome to dMS

First off, I should make it very clear that this essay, for all of its exploration, is not intended to persuade the reader of actually using a “dMS method” in-game, for realsie reals. Part of why I feel comfortable giving it a silly name like “dMS” is because it’s largely a silly mathematical exercise. Consider that:

• The method as concretely presented here probably still needs some refinement, not the least of which being a proper measurement of realistic box-to-box timings & some experimentation with more than just one (1) box-ordering.
• Any version of this method is likely impractical, simply for reasons of coöperation: it relies on all 5 PCs independently memorising the sequence, or at least memorising the subsequence of moves that they personally have to make. This problem can be ameliorated by simply having a reference on hand, but again, probably not gonna happen.

Nonetheless, now that we’ve seen a brief comparison of the three methods, let’s take a look at what the current version of dMS looks like, in the usual notation:

It would be really nice to present this sequence visually, à la Fig. 5, but there’s a slight issue: the arrangement of LPQ stage 8 boxes isn’t flat. This isn’t a dealbreaker, but I figure it’s probably best to go for an animation…:

That’s 126 combos!! If that’s a bit fast, here’s a version that’s at a more leisurely pace:

Mystery unlocked?

So, how about that? I think I personally unlocked quite a few mysteries over the course of this little essay. The only mystery that I haven’t unlocked yet, is how exactly I blacked out for like a month, & then, when I came to, found that I’d written all of this. I guess nature is just full of mysteries!!!

Indeed, while dMS taught me a lot, one of the many things that it taught me was that the JMS method is really, really good. I mean, I kinda already knew that, but it’s worth noting how difficult it is to improve upon JMS without making it impractical for real in-game usage. WMS largely has the right idea: by pushing some minor optimisations onto the sole “combo leader” (the PC who starts on box 4), & then expecting the cycler to use “cycles” of the general form ⟨1, 3, 6, 6, 1, 3, 3, 6, 1⟩ (and its analogues for other values of $𝑐$), JMS remains practical but surprisingly efficient — especially when compared to GMS…

If you’ve read this far, thank you, & I hope you learned something, or at least… laughed at me, or whatever. If you find any errors, or are having trouble understanding any part of this essay — computer code included — just drop me a line! 🧡

S>EPQ svc

So… where was I…?

Oh, right. In the “S>PQ svc” section of the previous entry in this diary, I went over the KPQ svc that I sold to Hampy’s quadruplets (see also: the “LPQ gang rise tf up” section above). After enough LPQs to get each character a pair of Broken Glasses, it was time to move on to EPQ for those sweet, sweet Glittering Altaire Earrings.

Of course, sorts doubles as a quite effective EPQ mule, so I agreed to do some more “solo” PQing as a PQ-carry service! Stage 1, especially at first, ended up taking a good chunk of the time. Although I can SSK these Poisoned Lord Trees pretty effectively, there are just an absolute load of them, & the map that they’re spread over is quite large. So Danger found a way to make this easier:

By luring the monsters on the bottom floor to the right-hand side, I can focus on the top three platforms first, & then come down to the bottom-right & start SSKing these bad bois away.

Stage 2 was no problem, especially with my CSB. Similarly to what we saw with LPQ, Steal can be used to steal the Poisoned Stone Bugs’ Concentrated Poisons. Believe it or not, this actually came in handy quite a few times! The advantage is that, especially with CSB, you’re frequently waiting for the bug’s weapon cancel buff to wear off so that you can kill it, & drop its poison bottle onto the pond. With Steal, you don’t have to wait! With any luck! But also it only really helps on the final bug, because otherwise you have to wait to kill it anyway…

Hampy also helped with stage 3, & even stage 4. It can be helpful to have someone just spamming Purification Marbles, yeah?

He also noticed that, for some reason, the MapleLegends Library entry for the EPQ stage 4 map lists 20 spawnpoints for Poisoned Spright (and none for any other species), but the monster doesn’t look anything like the in-game version — instead, it’s some kind of purplish Nependeath-looking plant-monster. Moreover, the spellings of English sprite in the context of the Ellin Forest are inconsistent: here, it’s ⟨Spright⟩, for some reason…?

If we look at the KMS entity with the same MoID, we see that it has the correct sprites (in the graphics sense LOL) that we’re used to seeing in-game in MapleLegends. Moreover, its name is 중독된 스프라이트 ⟨Jungdokdoen Seupeuraiteu⟩ /t͜ɕuŋ.tok.twen sɯ.pʰɯ.ɾɐ.i.tʰɯ/ “Poisoned Seupeuraiteu”. 스프라이트 is clearly an adapted loan of English sprite /spɹaɪ̯t/, so we can only assume that whoever localised this into English saw ⟨-ㅏ이트⟩ and thought that it was a straightforward adaptation of ⟨-ight⟩.

This confusion is partly a result of the Great Vowel Shift of ca. the 15^th through 17^th centuries, whereby (amongst other things) MidE /iː/ diphthongised to ModE /əɪ̯/ > /aɪ̯/. MidE sprite (also spryt, etc.) would have been /spriːt/. In order to distinguish this use of ⟨i⟩ from its use for /ɪ/ (or even /i/) — compare sprit /spɹɪt/ — English orthography has adopted the significant use of a “silent ⟨e⟩”. On the other hand, words like bright, might, tight, light, etc. were generally *-ht- (e.g. *berhtaz) in PGmc, lost most or all of the suffix after the *-ht- by PWGmc (e.g. *berht), & were then realised in OldE & MidE as /-xt/ — e.g. OldE beorht (also bryht, byrht, etc.) /beo̯rxt/ > MidE bright (also briht, briȝt, etc.) /brixt/. By LMidE & EModE, h-loss had completed, and the presence of /x〜h/ (most commonly [ç], as in LModE^[13] human /ˈhju.mən/ [ˈçjum(ə)n]) in coda positions was lost, whilst spellings like ⟨bright⟩, ⟨knight⟩, etc. remained.

The result is that /aɪ̯/ — virtually /aj/ — is a common phoneme in ModE, but there somewhat confusingly remains no good way to spell it, apart from ⟨i⟩^[14] — indeed, the letter ⟨i⟩ itself is pronounced /aɪ̯/! English sprite is from OF esprit (Modern French esprit /ɛs.pʁi/), from Latin spīritus /ˈspiː.rɪ.tʊs/, making it a doublet of spirit /ˈspɪ.ɹɪt/ (which retained the Latin form). The spelling ⟨spright⟩, then, is unetymological — there was never a /x〜h/ — but the confusion is understandable. Indeed, ⟨spright⟩ is a now-obsolete spelling of the word, & it survives to the present day in sprightly!

EPQ made its GMS debut in v72 (2009-06-24). In the WZ data of that version, we can see that MoID 9300175 also looks how we expect it to — but its name is unexpectedly Poison Flower. Looking at MoID 9300174, we see the poisoned-Nependeath-looking monster named Poisoned Spright. It’s clear that their names were accidentally swapped, although it is possible that the accidental swapping already existed in e.g. KMS prior to the localisation into GMS, & was later fixed.

In any event, stage 5 works about how you’d expect; if we’re unlucky (read: ⅓ of the time), then the Purple Stone of Magic is on the platform that neither of us initially checked.

Then, I have to kill BBRG. During one run, Danger gave me an Onyx Apple (& MW30!), and so I got to absolutely melt this geological critter!:

SHEESH.

This final stage turned out to be more dangerous than it first seemed. There was one incident where one of Danger’s characters died because he’d moved them to the right-hand side of the map as a minor optimisation, & BBRG randomly decided that it really enjoyed the right-hand side of the map. It seems that BBRG somehow has a “re-aggro” mechanic similar to Pink Bean, where it can re-aggro to a PC that it “sees” (so to speak; meaning whatever’s right in front of it, in the direction that it’s facing), largely regardless of how much damage that PC is dealing. This would make at least some sense, considering that GMS released both EPQ and PB in v72…

Once we got the hang of it, most runs were below the ten-minute mark. Like last time, I tried to record times as accurately as I could manage, starting upon entering the PQ, & stopping when BBRG dies. You can see the raw data here:

This represents a total of 37 runs & 35 clears (2 abortive runs), with a fastest TTC of 8:25, a slowest TTC of 18:00, a mean TTC of ≈10:29, & a median TTC of 10:00. Not bad for “solo” EPQ svc! 🧡

🛑

That wasn’t all the sortsing that I did, though. I also partook in a li’l bit more of that LPQ action, this time alongside gunslinger Rokugatsu (Lvl1Crook, Level1Crook, xXCrookXx, Macer), permapirate Mommie (GiIf, MiIf, Dakota), & swashbuckler insist (Skjal, inject, inhale, Tetrin)!:

[:<

Naturally, no LPQ sesh is complete without some hoeing afterwards. So I headed over to the Haunted Mansion with insist & Mommie, to play Hallowe’en dress-up, and… to hoe. Duh.

Footnotes for “Sorts, kinds, sets, & types”

1. [↑] I write “thief” portal with scare quotes around “thief” because, although thieves are well-suited to using Dark Sight to complete this portal, many non-thieves are also well-suited:

   • Brawlers (such as myself) may use Oak Barrel.
   • Many — albeit certainly not all — warriors are capable of completing this portal simply by surviving hits from the King Block Golems from Another Dimension.

2. [↑] $\left(\genfrac{}{}{0ex}{}{𝑛}{𝑘}\right)$ (which can be read aloud as “$𝑛$ choose $𝑘$”) is the usual notation for a binomial coefficient, meaning the number of ways to choose $𝑘$ distinct elements — without regard to the order in which they’re chosen — from a set of $𝑛$ elements.

3. [↑] Note that this framing can also be interpreted in terms of positional notation. For example, we can interpret the combo string 123 as the number 123 (“one hundred & twenty-three”; a senary — or whatever radix — interpretation is also possible, of course). We then see that the combo strings are not only in lexicographic order, but are also in numeric (the usual ordering of the natural numbers) order. But we have to skip a lot of numbers (e.g. 131, 132, 133, etc.), so I prefer to think of it lexicographically.

4. [↑] See, for example, Clive Maxfield’s “‘𝑛-ary’ (non-Boolean) Gray codes (Part Deux)” (2008-10-07), a very brief blogpost on the subject.

5. [↑] A.k.a. penteract, regular deca-5-tope, or regular decateron; this is the 5-dimensional analogue of the 3-dimensional cube.

6. [↑] Unless it turns out that $𝐏=\mathrm{𝐍𝐏}$, in which case, it’s more like “good luck finding a polynomial-time algorithm for that that isn’t also galactic”…

7. [↑] Petr Gregor, scribed by Václav Vlček & Sasha Sami; “Lecture 10”; Hypercube Structures, Charles University, Department of Theoretical Computer Science and Mathematical Logic; 2021-12-12; archived.

8. [↑] Laszlo Lovász; “Problem 11”; Combinatorial Structures and their Applications; Gordon & Breach Science Publishers, New York; 1970.

9. [↑] Donald T. Tang & C. N. Liu; “Distance-2 Cyclic Chaining of Constant-Weight Codes”; IEEE Trans. Comput., vol. C-22, no. 2; 1973-02; doi:10.1109/T-C.1973.223681. (You’re gonna have to use Sci-Hub for this one.)

10. [↑] Using an overline to represent bitwise complement.

11. [↑] See my scroll_strategist for another application of dynamic programming!

12. [↑] When I took these measurements, I realised that not only is box 9 an extra four pixels to the right (which is fairly visible in-game), but for some godforsaken reason, specifically box 6 is misaligned by being one (1) pixel too far to the right. Hello? Wizet??

13. [↑] Excepting varieties that drop this /h/ unconditionally.

14. [↑] Or similarly confusingly, this phoneme is sometimes represented as ⟨y⟩, as in e.g. eye, why, try, dry, etc. This would suggest either *⟨spryte⟩ or *⟨spryt⟩.

15. [↑] Notice how the RBGC construction parallels the hypercube (as in e.g.: ^[5]) structure of its corresponding flip graph. In order to construct an $𝑛$-cube, we can take an $\left(𝑛-1\right)$-cube, duplicate it, and then add $\left(𝑛-1\right)$ edges, each edge being between a vertex in the original $\left(𝑛-1\right)$-cube & its corresponding duplicated vertex. These added edges correspond to the

    $$\begin{array}{cc}{2}^{𝑛-1}& 2^{𝑛-1}\\ \overbrace{\mathrm{𝟶}\mathrm{𝟶}\cdots \mathrm{𝟶}\mathrm{𝟶}}& \overbrace{\mathrm{𝟷}\mathrm{𝟷}\cdots \mathrm{𝟷}\mathrm{𝟷}}\end{array}$$

    sequence seen in the leftmost position of ${𝐺}_{𝑛}$.

    To make a Hamiltonian circuit (corresponding to ${𝐺}_{𝑛}$) on the $𝑛$-cube, then, we simply go around the original $\left(𝑛-1\right)$-cube (starting at whatever you decide is the “zero” vertex), take the available edge to the duplicate $\left(𝑛-1\right)$-cube, go around that in the reverse direction, & then take the available edge back to the very beginning (“zero”) vertex. For example, with $𝑛≔3$:

    

    The fact that we go in the reverse direction on the duplicate $\left(𝑛-1\right)$-cube (in this example, an $\left(𝑛-1\right)$-cube is a 2-cube, a.k.a. a square) corresponds to the “reflection” part of the reflected binary Gray code.

Thee ſubtile & vvonderfull aduentures ov ela & noto

As I briefly mentioned in the “My impressions of the event” section above, this Hallowe’en event made the PQs quite lively & rewarding. Although I did do some Hallowe’en-y LPQing in the “🛑” section above, I was hungery Hungary HUNGRY for something a little different.

And guess what? You’re not my mum, you’re not my boss, & you’re not my other mum, either (…unless you are. Hi mum!!). I can do whatever the ✨fuck✨ I want. So I made elaphus:

Accompanied by her trusty murid sidekick notomys, ela & noto set out to do some god-damned PQs. The plan could hardly be simpler: look cute, hit things with my red rose, show noto the delicate delights of party questing, kick some arse, be gay, do crime, et cētera, all that shit. You know the drill.

And who knows? I just might learn something new.

I previously believed that this route of OPQ’s Lobby stage — the 5^th from the left, for Thursdays — was impossible without either Teleport or Recoil Shot. After all, even with maximum (viz. 123%) JUMP, it’s just not possible to directly jump from this one platform to the next. Then, I heard someone claim that they’d done an OPQ with a mysterious archer who did this stage themself, on Thursday. Now that I really thought about it, I realised that there was a single pesky flying star obstacle that could — potentially — be used to bump the apex of the jump up a bit further.

As it was Thursday & there were no mages nor gunslingers forthcoming, I decided to give it a go anyway. After a few tries, I indeed made the jump, as you can see in the above video clip. Now we know that that archer wasn’t fly hacking! Probably!!

Indigoferous

In the “Cervidae” section of the previous entry in this diary, we saw my pure STR bishop cervid do some questing in Masteria, helping to uncover the mystery of the powerful magician Subani’s Legacy, & then meeting with John Barricade’s brother, Jack, to obtain a very handy Map of [the] Phantom Forest.

This time, I got to do a bit of fun grinding with fellow basic-attackers & Ctrl ⎈^[1] aficionados Taima (Hitodama, Yunchang, Boymoder, Tacgnol, Girlmoder, deerhunter) & Jestterz the STRginners! We did some classic trio grinding at Fancy Amps, in eye-melting Taiwanese shopping mall hell:

Whoa! I also gave cervid a makeover! Whoa!!

Anyway, a day or two later, we decided to do some more trio grinding, but this time at a ✨brand new✨ (woah!) grinding spot: Forgotten Passage! The rewards — especially including the EXP — for killing Thanatos, & its triocular cousin Gatekeeper, were bumped up by a huge amount in a recent update to MapleLegends. Naturally, this suggests a new grinding method, & we were eager to give it a go:

This is nothing like Fancy Amps, though. Instead of multiple platforms full of numerous little bad guys, the Forgotten Passage is one long floor that only has two spawnpoints! But Thanatos is pretty chunky, at 70k HP a piece, so we move around the map together as a trio, chipping away at each individual Thanatos like some kind of mini-boss.

Indeed, this training method is very active work. In fact, for cervid, it’s more than “active” — it’s quite the workout!! Thanks to the Seals, Curses, & Darknesses that these spoopy doods like to cast upon us, I get a ton of use out of my Dispel skill, & I have to be very quick to react! My only solace is Holy Shield, which gives me about half a minute of chill time, during which I can just attack. Well, even then, I still do a decent amount of Healing…

It might be exhausting, but that doesn’t mean that it isn’t pretty damned cool!

Footnotes for “Indigoferous”

1. [↑] The fact that I actually have it bound to x notwithstanding…

Bats spelled backwards is stab

Thanks in part to every diary entry that I make now taking ≈1 month’s worth of constant daily work to produce, & thanks in even larger part to MapleLegends being limited-time-event-driven (F7), I haven’t allocated much time to playing our favourite stabby girl alces.

Heck, last time, I even forgot to include the part where alces soloed Captain Lataniel twice daily until I finally finished up that card set!:

Speaking of Lataniel, I also did some Latting with STRginner Taima (Hitodama, Yunchang, Boymoder, Tacgnol, Girlmoder, deerhunter). That meant some more frantic ≈115-second clears so that we could keep the Holy Symbol buff that we got in the FM!:

Killing Lat in 115 seconds flat! How about that‽

And I did get just a tiny smol amount of card-hunting done. Over in Masteria once again, on the steep Mountain Slopes, I started the I.AM.ROBOT set:

Which also incidentally allowed me to start making progress on the Mighty Maple Eater set as well!:

Hopefully we’ll see more stabbing in the future!!

In quest impressed, & all the rest

In the “Accipiter” section above, I got some NT quest completions on my pugilist tarandus. So… how about that Questin’ With tara™ content…? Well, you’ll be pleased to know that this diary entry has a lot of Questin’ With tara™! Like a lot!!

But first, a little bossing. “Boooo!” I know, I know — we already saw rusa do a bunch of that. But can rusa do Bigfoot?

Actually, yeah. We’ve even seen rusa solo Bigfoot, at least a few times before. But soloing is boring, because there’s no one to talk to. That’s why I duoed some BF, alongside STRginner Taima (Yunchang, Tacgnol, Boymoder, Girlmoder):

And after a great number of BFs slain, I even levelled up — to 154~!:

Cool and nice.

I also did some BFs as a trio with Taima & somewhat INTelligent corsair Lvl1Crook (Level1Crook, xXCrookXx, Macer), during which, we uncovered Lvl1Crook’s true identity: purse pirate.

Incredible.

Also, Taima hit level 135?!?!?:

Whoa! That’s the Zak level!! Let’s… Zak?? o_o

And indeed, I was delighted to take part in just such a legendary permabeginner Zaqqūm moment with Taima, shadower Harlez (Gock, Murhata, VigiI), bishop misandrist (xRook), markswoman Temptress (Beotch, Chastity), & paladin 360Flip!:

Out of our party of six, 360Flip was the only one whom I could really call a “rando”, so it was a neat cosy pair of runs. Plus, with Harlez absolutely swaddling & carrying us, the runs were decently fast, to boot!

Like in the previous episode, I did a wee bit of Yāosēng duoing with Harlez. This time, even without using anything stronger than Ciders at all, we cleared with a little over a minute to spare each time!:

Stromk???

Also with Harlez & Taima, I did some o’ that good ol’ Rāvpappin’:

And Harlez helped me get my Portable Laser Guidance by doing just one or two NTPQs with me!:

Pretty luccy!! Oh, and I was really close to levelling up, anyway…:

155?!? That’s the HT level!! Will this mean that I get to HT on a character that isn’t rusa…?

Anywho, remember the first NMM run that I did in the 「扶桑」 section above? That’s a Namie clear for tara! NT1 done!! F5

Plus, I had a bunch of fun doing some Zaqqūming alongside Lvl1Crook & bishop MiIf (GiIf, Dakota)!:

Gang!!

And of course, level 155 just wasn’t enough…

The mystery of the Prendergast Mansion

But farming just a PLG is no more than the tip of the tip of the iceberg here. As mentioned in the “Traditional Hallowe’en quests” section above, this year’s Hallowe’en event brought back the Haunted House quests! That’s yuge!! I’ve really been wanting to do these quests ever since I joined the server, so naturally, I did them for the first time ever by doing them as tara.

Upon arriving at the Phantom Forest, just outside the previously opulent — but now dilapidated — Haunted House, I’m stopped by Old Man Tom. “A Word to the Wise” involves Tom repeatedly warning us that the house is haunted by the ghosts of the Prendergasts. Jonas Prendergast was a renowned toymaker who tragically lost his wife, & when he then tragically lost his child Sophilia to an unspecified accident, he was finally driven to madness. Jonas locked himself in his Toy Workshop, never came out, & it’s said that his factory continues churning out deranged “toys” to this day, which haunt the house in perpetuity.

I ain’t afraid of no ghosts, though. Tom becomes increasingly exasperated with me — just another “young’un” ignoring his words of warning.

You’ll be sorry! Don’t say I didn’t warn you!

But I’ve been in this Haunted House before: back in the “तारा” section of pt. xciv of this diary, we saw a 2^nd-job tarandus (li’l baby tara omfg 🥹) complete “Smores No More” by beating up a bunch of Nightmares^[1]. Well, for “A Cold Hearth”, I did just that, once again:

Of course, you could just bring Firewoods from, say, somewhere in Perion like East Rocky Mountain IV, but given that I’d only need 110 of them — 60 for “A Cold Hearth”, & 50 for “The Infernal Toy Machine” — farming Nightmares for them really wasn’t so bad.

Edmunds, the head butler of the household, has continued to service the mansion even after the demise of their darling Sophilia. This is no phantasmal manservant, however; Edmunds is just an old man who has serviced the Prendergast Mansion ever since its construction, and he — & the few other staff members who remain — have had to deal with the ghosts that now prowl the estate.

Edmunds fills me in about some other details relevant to the Prendergasts. After Jonas’s first wife passed, he remarried six times, with Ludmilla being his seventh & current wife. With Jonas being extravagantly wealthy from his highly successful toymaking business, he had plenty of suitors, but none could replace his lost first wife.

Now, the groundskeeper of the mansion has apparently… died? And it’s implied that the groundskeeper was slain by the ghosts and/or toys of the mansion! With no groundskeeper to cut wood for the fireplaces, Ludmilla has been complaining endlessly about the cold — hence “A Cold Hearth”.

Although the remaining mansion staff have plenty of their own troubles, they aren’t the only ones interested in the spooky scawwy skeletons of the Haunted House. Agent Falcon, who previously hired me to bring him those 25 Smores, is a Maple Bureau of Investigation (MBI) agent on a mission to interview a certain “Mr. Bones”, purportedly located within the mansion.

Normally, of course, I would never collaborate with cops. But like all cops in MapleStory, Falcon is comically incompetent — one of the game’s few concessions to realism. On an attempt to make his way through the mansion, he was spooked by a couple of Glutton Ghouls, & dropped his pages of a so-called “Paranormal Activity Report”. Thus, for “The M-Files” — so named for The X-Files, whose protagonists are fictional FBI agents, the IRL analogue of MBI agents — I headed to the Haunted House’s Library to farm Glutton Ghouls:

Unfortunately for me, farming Glutton Ghouls — to use a highly technical term — sucks big doo-doo. The map is oversized & difficult to navigate, & worse, there are these goddamned flying books everywhere. Unlike the flying books (“Ridewords”) of the Haunted Mansion — see the “Four-way tragedy” section above — the airborne tomes of the Haunted House are totally invincible. This means that if you accidentally hit one with an attack, it just follows you around forever — or at least, until you leave the map.

Also, why are there two species of Glutton Ghoul? The other one doesn’t drop any equipment, & doesn’t drop any of the USE items that the other one does… except for Maple Pops, for some reason? Whatever.

Oh, and “File Under Incomplete” asks for another hundred pages (for a total of 150), because… the report is a bit bloated as a result of most pages just being doodles.

For “Special Delivery”, we get the full Ludmilla treatment of “I wouldn’t usually converse with someone of your obvious low breeding”, “you’re quite an ugly one, aren’t you?”, & being called a “wretch” if we decline to assist her. I’m starting to think that Ludmilla is part of the reason why Jonas locked himself in that workshop… But she just wants five Cherry Pies, which is easy enough when Miki’s got your back.

Ludmilla tries luring Jonas out of the workshop with the smell of Cherry Pie, his favourite food, but unsurprisingly fails. Ludmilla immediately gives up, & for “Family Jewels”, asks me to recover her missing earrings. She wants twenty Tarrymore Earrings & ten Sepha Earrings… So many of the same exact two pairs of earrings? I guess she really loses these often! In any case, the Sophilia Dolls have them:

As recompense for my efforts, Ludmilla has a “ragged old doll” for me. This is actually Sophilia’s First Doll (named “Annabelia”), much to Sophilia’s delight in “A Lost Girl”. Wait, what? Sophilia is alive?? Something’s not adding up…

As it turns out, getting into Jonas’s workshop is… not terribly difficult? There’s a door on the top floor of the Library that leads into it, & the door has been left unlocked. Jonas is in there, & somehow manages to mistake me for his daughter. Rather than bringing him food, or water, or trying to get him to come out of his workshop for once, I set about assisting him in repairing his toy factory, in “The Infernal Toy Machine”. In addition to the aforementioned Firewoods, & a handful of Screws, Jonas needs 25 Loaded Springs to be recovered from his jack-in-the-box toys.

No ordinary jack-in-the-boxes, these tricksome musical boxes^[2] are nōn compos mentis: they are Psycho Jacks.

Terrifying.

In any case, for “In Her Likeness”, I helped Jonas to use his newly-repaired toy factory to make a doll in the likeness of his late daughter, by collecting a few materials: 30 of the Locks of Doll’s Hair that I’d previously obtained from the Sophilia Dolls, 10 Heartstoppers that I’d previously obtained from Sophilia, 10 Pumpkin Taffies from Glutton Ghouls or Sophilia Dolls, & 10 Maple Pops from Glutton Ghouls or Nightmares.

The result is a “Perfect” Sophilia Doll, & in “A Father’s Love”, I bring it to Sophilia. Sophilia is already satisfied with her first doll, & wishes that her father would stop spending so much time to make more, instead spending that time with her. Delivering Annabelia to Jonas, he realises why Sophilia kept the old doll all this time, & wants to atone for his parenting mistakes in “Reparations”.

“Reparations” then splits into three subquests, which require 50 each of Psycho Jack, Sophilia Doll, & Nightmare kills.

[T]o think that they have become so corrupted and frightful. The thought of breaking my own creations is distressing, but it eases my heart to see that any image of my daughter so desecrated is reduced to scrap.

The rewards are the Toymaker Cap, Toymaker Cape, & Toymaker Hammer, respectively.

Looking goooood. 😎

Jonas also gives me a Toy Workshop Key so that I can go in & out of his workshop whenever I want. In “The Past in the Present”, Old Man Tom sees the key & is — reluctantly — convinced that I actually did meet with Jonas in that Haunted House. Old Man Tom reveals that his full name is Thomlin Prendergast — he’s Jonas’s younger brother!! Everyone in there was a ghost, this whole time!!! No way… I never would’ve guessed…

Anywho, the ultimate reward for unlocking the mystery of the Prendergasts is a set of five Sophilia’s Necklaces:

Unfortunately, these are almost useless to me, given that Super Transformation is a +60 (W/M)DEF buff…

But that’s not all. In The File Thickens parts 1 & 2, Falcon reveals that the so-called “Mr. Bones” is actually just a name that he made up to make Jonas Prendergast sound scary. Falcon has now given up on trying to enter the mansion at all, & instead sends me in to do the dirty work. I get Jonas to fill out a Paranormal Subject Questionnaire… Great. I can’t imagine what such a questionnaire could possibly contain…

There is just one more thing, though. You see, although the exact cause of Sophilia’s demise is left unspecified, we do learn from Edmunds in “Eat This!” that she died whilst somewhere inside a fireplace, or inside the chimney system. For this reason, there are now locked grates protecting each fireplace from unauthorised entry. If I can bring 20 of each type of Gummy Slime — Red, Green, Purple, & Orange — to Edmunds, then he’ll give me the Omni Key, a kind of skeleton key for the fireplaces:

So… I did that! It’s a bit tedious, so as it turns out, this quest takes probably at least as much time as all other Haunted House quests combined. Although they are, as previously mentioned, quite painful to fight, Glutton Ghouls are nevertheless a good source of the Gummy Slimes, as they drop all four colours.

Omni Key in hand, I unlocked my way into the Ghost Chimney — the main artery of the Haunted House’s chimney system:

As you can see, there are branches off of this main artery — indeed, roughly a dozen or so of them. But in order to make your way up the artery, you have to use those shitty little platforms. The platforms are devilishly spaced, & even with how relatively confident in my JQing abilities I am now, I had an extremely hard time making much progress at all.

In fact, it’s so difficult to go up, & so easy to fall quite far down, that I used probably a few dozen Return Scrolls — Nearest Town in the process of climbing this thing. I entered through the fireplace at the top of the Foyer, & if I fell significantly below that entry point, I just used a return scroll. I could then painstakingly walk back into the Haunted House, all the way up the Foyer’s four floors again, & back into the abominable Ghost Chimney. Seriously, that was faster than just climbing up the JQ. It is so difficult.

Eventually, however, I did make it up to the top!:

Exhausted, I entered that top portal, & wound up in… Jonas’s workshop. Again. 🤦🏽‍♀️🤦🏽‍♀️🤦🏽‍♀️ So yeah, I maybe slightly improved my JQing skills… but that’s about it. Ouf.

As I was wandering around the Haunted House in this way, I did come across an uncanny Barren Room…:

Apparently, there are four of these things! Creepy…

So there it is! My first experience with the Haunted House quests!! I had heaps of fun doing these; although none of the quests involve doing anything particularly interesting on their own, the way that each quest — with the slight exception of “Eat This!” — is relatively lightweight, & the way that the quests are all webbed together, is pretty neat. Plus, you get the Toymaker set!

You might have noticed that these quests are quite plot-driven. And the Haunted Mansion quests (see the “The adventures of GiIf & d00r in the Haunted Mansion” section above), even more so. In addition to being plot-driven, these quests are characteristically vague with the actual premise of said plots. There’s a lot of spooky ghost action, which allows for characters to be clearly ghosts, not-so-clearly ghosts, living persons, or even an ambiguous compromise between these things. It also allows for a lot of objects & events to be “just kinda ✨magical✨” — that’s just how ghosts work, I guess!?!

The basic premises of both the Haunted House & the Haunted Mansion quests are, very broadly, similar: Jonas Prendergast is a wildly successful toymaker who lives in an opulent mansion; his daughter Sophilia tragically dies at a young age; Jonas is driven mad in an attempt to create the “perfect” doll of his daughter’s likeness; Ludmilla is the incredibly stuck-up remarried wife of Jonas who doesn’t care for Sophilia & doesn’t treat anyone with respect; and the player later learns that everyone with whom they were interacting was actually a ghost this whole time.

Still, there are a great number of differences, including several major plot points & characters. When combined with the wishy-washy nature of the premises, & the somewhat convoluted nature of the plots, the result is that I’m personally not so great at keeping track of what happens… what’s what… who’s who, & when… It hurts my tiny dum-dum brain. So I think that’s about enough of that…

Ex mātrimōniō

In the “A grasp nullary” section of the previous entry in this diary, we saw tara get married to some random nerd sair named Lvl1Crook. Amongst other things, this opened up a few Amorian quests for yr grrl tara.

I actually started out with “Cupid’s Courier”, in spite of what I said about this quest back in the “tara tara!” section of pt. xc of this diary:

Look at that. Angelique deadass tossed two Watermelons at me and laughed in my face. Well, that’s the last time that I do her a favour…

Unfortunately, this quest has continued to sit around in the “Available” tab of my quest journal, despite my having completed it. I actually like using my quest journal, so it’s a bit irritating to have spurious entries in it. I figured that this was probably just one of those quests (like e.g. “The Final Stage of the Sealing Ritual”), but just in case doing it a second time would actually fix it, I did exactly that. Aaaand it is, indeed, just one of those quests. Bummer.

But let’s move on to the new stuff, shall we? Like “Red Dahlia”, featuring Ben, a farmer of Amoria. Ben promised his wife a Red Dahlia, but his left arm is in an orthopædic cast, & he hasn’t the strength to fight. So, it’s my job to gather the materials for him to… make one? Namely: three Diamond Ores & 30 Stiff Feathers. I thought dahlias were flowers, though? You can’t just smash some diamond & bird feathers together to make a whole plant! That’s not how anything works!!

The genus Dahlia is native to Mexico, & is known for the beautiful flowers of its specimens — indeed, the national flower of Mexico is D. pinnata (a.k.a. D. rosea)! Given that D. pinnata is also the type species of the genus, & given that its flowers can(!) be red, it might be the red dahlia of “Red Dahlia”. Then again, the English Wikipedia article Dahlia recounts an anecdote of an outstanding red dahlia that is almost as poorly-sourced (🥲) as it is stirring:

In 1872, J. T. van der Berg of Utrecht received a shipment of seeds and plants from a friend in Mexico. The entire shipment was badly rotted and appeared to be ruined, but van der Berg examined it carefully and found a small piece of root that seemed alive. He planted and carefully tended it; it grew into a plant that he identified as a dahlia. He made cuttings from the plant during the winter of 1872–1873. This was an entirely different type of flower, with rich, red color and a high degree of doubling. In 1874, van der Berg catalogued it for sale, calling it Dahlia juarezii^[3] to honor Mexican president Benito Pablo Juárez, who had died the year before, and described it as “…equal to the beautiful color of the red poppy. Its form is very outstanding, and different in every respect from all known dahlia flowers”.

This plant has perhaps had a greater influence on the popularity of the modern dahlia than any other. Called « les étoiles du diable » (“stars of the devil”) in France and “cactus dahlia” elsewhere, the edges of its petals rolled backwards, rather than forward, and this new form revolutionized the dahlia world. It was thought to be a distinct mutation, since no other plant that resembled it could be found in the wild. Today, it is assumed that D. juarezii had, at one time, existed in Mexico, and subsequently disappeared. Nurserymen in Europe crossbred this plant with dahlias discovered earlier; the results became the progenitors of all modern dahlia hybrids today.

Resurrecting a beautiful plant species from extinction! How romantic! And for my efforts, Ben gave me a single eye accessory for WACC 60% scroll…:

Gee, thanx…

Luckily, the other new quest that I got to do — Circle of Trust — was much more rewarding. As it turned out, I already had a number of Ring Fragments from the Tweeters that I killed over the course of my Lūdus Lake questing. So I was able to — with Moony’s help — repair Gary and Shatima’s wedding rings, & redeem my Gelt Chocolate right away:

If you’ve never seen Gelt Chocolates before, they’re basically like Onyx Apples on steroids, giving +120 WATK for 10 minutes, instead of +100. This makes them the strongest WATK buff in the game!! But they’re also more special than Onyx Apples, as you’d expect: they’re untradeable, & quite rare. As far as I know, this non-repeatable questline is the only way to get one, other than rare drops in CWKPQ’s bonus stage.

Gelt chocolate (more likely chocolate gelt) refers to chocolate Ḥanukkah gelt; the in-game item description even mentions the “Festival of Lights”, another name for Ḥanukkah^[4]. Sometime during the early 20^th century, U.S. chocolatiers began selling coin-shaped chocolates individually wrapped in shiny tin foil, as an edible replacement for gifts of actual money to children. The origin of monetary gifts for Ḥanukkah appears to be slightly obscure, apparently stemming from tips given to certain workers towards the end of the year.^[5]

The forefronting of Ḥanukkah as a major Jewish holiday in the U.S. — in spite of its relatively little religious significance — is interesting in its own right, & coincides (in the early 20^th c.) not only with the innovation of chocolate gelt, but also with the American practice of gift-giving during Ḥanukkah, which developed by analogy with Christmas (another end-of-year holiday). Although chocolate Ḥanukkah gelt has since spread outside of the U.S., chocolate gelt & gift-giving are historically linked to a distinctly American commercialisation of Ḥanukkah that occurred during the first half of the 20^th c., alongside a similar commercialisation of Christmas — a source of some disappointment for rabbis & Christian ministers alike.^[6] It is thus unsurprising that the innovation of chocolate gelt was due to American commercial chocolate manufacturers.

The word gelt /gɛlt/ “money; tribute, tax” had already entered the English language by the 16^th c., via MHG gelt /gɛlt/ (descending to SHG Geld /gɛlt/ “money”). Ultimately OHG gelt /gɛlt/ “payment, money”, from PWGmc *geld “payment”, from PGmc *geldą “payout, reward, gift, money”, from PIE *gʰeldʰ- “[to] repay, reward, pay”. But PWGmc *geld is also the source of Yiddish ⁧געלט⁩ ⟨gelt⟩ /gɛlt/ “money”, whereby the word re-entered English in a specifically Jewish context. We also have:

• *geld (& *geldą > … > ON gildi /ˈgiːldi/ “recompense; repute, esteem; feast, banquet; guild”)
• > OldE ġild (also ġield, ġyld, ġeld) /jild, -ie̯-, -y-, -e-/ “payment, compensation; tax, tribute; guild; pagan god, idol”
• > MidE
  • ȝeld (also yelde, ȝilde, etc.) /jeld, -i-/ “tax, rent, fee; reward, recompense; agricultural yield”,
  • gild (also gilde, guild, guilde, etc.) “guild”
• > ModE
  • yield /jild/,
  • guild /gɪld/.

So… there you have it. English gelt, yield, & guild were originally all the same word.

More seriously, the association between “Amoria & Masteria” & “chocolate Ḥanukkah gelt” makes some very vague sense insofar as these two in-game regions originated in GMS, which is operated by Necksawn America, which is headquartered in the U.S. state of California. But chocolate Ḥanukkah gelt has no other relation to Amoria nor Masteria, & moreover, is really only relevant during December^[7]…? Looking through old GMS data, Gelt Chocolate was first added in v32, which is also the version in which Amoria was added. Note that the release of Masteria would have to wait all the way until v40. v32 was released on 2006-12-06,^[8] nine days before the first day of Ḥanukkah!

I did quite a bit of digging — like, probably too much digging — for the v32 patch notes. I found a few leads that almost got me there; for example, Alexa began crawling the GMS news pages in early December of 2006, just in time for this update! However, any archived pages prior to about 2007-05-28 or so get you this little gem:

Remember when Internet Explorer™ was actually a good browser? Me neither. But I do, unfortunately, remember when it was the only widely-used one… 2006 was a different time. I tried spoofing my user agent string to be IE 6 or 7 on Windows™ XP, but to no avail. 😢 Nonetheless, huge shoutout to ruffle for making these archives of old websites considerably more functional than they otherwise would be; apparently, there was a lot of Flash™ shit going on in the old GMS website.

Neoseeker was another promising lead; for example, they have the patch notes for GMS v9. But the archival is pretty sparse, & v32 ain’t in there. southperry.net posts Negzaan’s announcements on their forum, but they only started doing so somewhere around mid-2008. And so on…

Finally, I remembered to check the Hidden Street forum! As it happens, GMS v32 is the earliest GMS update that was recorded on there, so I really lucked out! Here’s the full text of Nehcsaun’s announcement:

A-ha! As expected, this version came not only with Amoria, but also with a festive season event. In particular, we see reference not only to Happyville (which is Christmas-themed), but also to Shalom Temple. As a bit of a humorous (to me, at least) sidenote, looking through the Hidden Street thread reveals that GMS players were not too happy about this update at the time: other versions like KMS, JMS, CMS, etc. were way ahead of them in terms of game content, & all they got were some stinky weddings that you have to pay a fat little chunk of IRL money for! Of course, there are no direct references to particular quests, nor to Gelt Chocolates, so we’ll have to dig a little more.

I took a look in the WZ data. As I mentioned previously, Masteria didn’t exist yet, so ostensibly, the only source of Gelt Chocolate would be Circle of Trust (QIDs 8867, 8868, & 8871). The problem is, in GMS v32, the QIDs below 9000 only go up to 8843, meaning that Circle of Trust just didn’t exist yet. So was Gelt Chocolate in the game data only, & inaccessible in-game? Not so fast. QID 8843 is “Festival of Lights — Blessing the Festival reward”, associated with the NPC Hannah. She has this to say:

• No one knows what’s inside this Hanukkah Present Box. Would you like to find out for yourself?

• I shall open the box for you. You will be blessed with an item that should fit your need.

• Here it is. Hope you like it.

• The item(s) you have received shall help thee on thy journey down the road. Happy Hanukkah, and Happy New Year!

Hannah can be found in Shalom Temple, where the NPC Simon can also be found.

Thankfully, MapleTip has record of this quest, & gives the big fat list of possible rewards, almost all of which are pretty much garbage. I mean, you can get 100 Arrows for Bow… W—wow…! But you can also get five(!!) Gelt Chocolates, which is pretty crazy.

“Festival of Lights — Building the Altar” (QID 8829) is the first quest in the Festival of Lights questline:

Excellent. I’d like to build an altar for the festival, so we’ll need 25 Altar Pieces. Some others attempted to help me, but their axes got stuck in the wood. Fighting those monsters should give you the needed pieces. Thank you and hurry back!

Once you get those Altar Pieces from the Axe Stumps, the quest ends by awarding two Gelt Chocolates, & giving this text:

Ok, I’ll start building the Altar. In the meantime, I’ll need you to gather 9 Sabbath Candles so we can officially start the festival. They can be found on those strange horned mushroom creatures.

This leads into “Festival of Lights — Blessing the Festival” (QID 8830), & once you get those Sabbath Candles from the Horny Mushrooms, you’re rewarded with the aforementioned Hanukkah Present (which Hannah erroneously refers to as a *“Hanukkah Present Box”).

By just the next version (v33; 2007-01-11)^[8][9], we do get Circle of Trust, & it does award one (1) Gelt Chocolate. CWKPQ is not added until v66 (2009-03-05)^[8]. Nevertheless, based on forum posts from around that time, it appears that CWKPQ’s bonus stage has always given out Gelt Chocolates.

Thus, it seems that Gelt Chocolates were indeed added to the game specifically for a one-time Ḥanukkah–New Year event, but were later re-used for miscellaneous rewards in other GMS-specific content.

Alrighty, then. Now that we know absolutely way more about Gelt Chocolate than literally anyone has ever wanted to know about it, it’s time to move on. Amoria is over!!

Top secret military operation: “Nurture cute baby bird”

In the “The mystery of the Prendergast Mansion” section above, I mentioned that I’d been wanting to do the Haunted House quests ever since I joined MapleLegends, but never had the chance to — until now. Well, the story with Clock Tower Monster Wipe-Out is similar: I’ve previously professed my love for area boss questlines, but Clock Tower Monster Wipe-Out was always mysteriously missing from MapleLegends. Timer has always been there in the Whirlpool of Time, but its questline just… wasn’t. Until a recent update! Hecc ya!!

This quest appears to be entirely unchanged from its original GMS debut. This primarily means that the dialogue is slightly embarrassing (although there’s certainly worse elsewhere), & was probably translated from Korean somewhat hastily. That being said, it’s still super fun, & pretty damn good, even by area boss questline standards! So let’s fuccing do it??

For “Operation Strategy 1 : Collect All the Eggs”, Ghosthunter Bob wants ten each of Tick-Tock’s Eggs, Chronos’s Eggs^[10], & Timer’s Eggs.^[11] I haven’t found any eggs quite yet, but I did find this adorable li’l wind-up worm:

After fighting some Tick-Tocks at the Whirlpool of Time, I found my first Tick-Tock’s Egg:

Wow, it’s already got the pair of bells on top! Nature is fascinating! But really, these function less like “eggs” than you might expect. Using one reveals its contents:

Sometimes, we get a Springy Worm. Sometimes, we get a Chick Cookie, which is an untradeable little cookie in the shape of a baby bird that heals 1k HP & 1k MP. And other times, we get a Timer’s Egg. Indeed, opening up Tick-Tock’s Eggs & Chronos’s Eggs is the only way to get Timer’s Eggs. Bob explains that this is because Timers always lay their eggs in other birds’ nests.

Once I pulled ten Timer’s Eggs out, & collected ten more Tick-Tock’s Eggs, I could move on to the Chronoses in The Path of Time ⟨2⟩:

For “Operation Strategy 2 : Defeat the Monsters”, I destroyed 50 Master Chronoses…:

…40 Platoon Chronoses…:

…30 Chronoses, & 20 Tick-Tocks.

That left only big mamma bird: Timer.

Now that the birbs had been wiped out, it was time to… replace them, I guess? But with tamed birbs. Out with the big scary Timer, in with the docile, domesticated Timer. For “Operation Strategy 3: A Cute Baby Bird”, Bob gave me a Timer’s Egg of my very own, to raise as my plastic avian child:

The mechanic here is substantially identical to the mechanic used for putting together the Rag Doll in the Haunted Mansion quests (see the “Four-way tragedy” section above). Double-clicking the item opens a dedicated little window that displays the progress of the item, & in order to make my little Timer Jr. grow up to be a healthy adult Timer Jr., I would have to drag Springy Worms onto it:

Wow! They grow up so fast! 🥹

ˈbɔsɪŋ

We’re really on a roll here when it comes to Questing With tara™, but I’m afraid that I must interrupt this episode for a commercial break li’l bit o’ bossin’. I did a handful of Zakky-wakky runs with Funk & co., including marksman & bishop xBowtjuhNL (PriestjuhNL, BuccjuhNL), paladin Palladino (Pengebingen), hero Machoke, & fellow buccaneer Camylla!:

We tried to get some Zaqqūm Helmets for Palladino’s bishop Pengebingen, but we got very unlucky with our first run, & zero helms dropped. So we did the second run, and…

What?? I mean, I’ve seen Zak not drop any helms before, but not twice in a row! Welp. Maybe next time…

I continued Zakking with xBowtjuhNL, including this pair of runs featuring bishop misandrist (xRook), shadowers Harlez (Gock, Murhata, VigiI) & TheBigThief, and fellow buckin’ near Government:

This was pretty funny to watch. You see, Harlez doesn’t actually speak a lick of Portuguese. But, thanks to her being a member of the Favela guild, TheBigThief understandably assumed that she did. Being a Lusophone himself — & not knowing much English — TheBigThief immediately tried to converse directly with Harlez in Brazilian Portuguese. Although Harlez later made it clear that she didn’t speak the language, she initially tried just using Google Translate™ to maintain a conversation, as you can see in the image above.

The Portuguese that you see above is written very formally & standardly, as you’d expect from the generic output of a machine translator. The result is that not only does it not look like the Portuguese that you’d normally see in MapleLegends, but it’s also reasonably intelligible to even a Hispanophone, to some degree. Consider a direct comparison (keeping in mind that this is written language, so phonetic differences are not generally adequately represented):

This somewhat artificial example is unusually neat, but still… It does highlight how formal written registers of languages can bring them closer to their siblings & cousins.

Speaking of Google Translate™, I slapped some big tree with Lvl1Crook & some randos:

What does that have to do with Google Translate™? I dunno, but I’m sure that you can figure something out.

And finally, I partook of a few cheeky Bigfeet in a quartet alongside Lvl1Crook, Taima, & priestlet AppleBasket (LawdHeComin, OrangeFungus, Mushers, Snel)!:

Bigfeatened. 😎

Health point challenges

⚠️WARNING:⚠️ Here be spoilers!! Normally, I’m covering ancient MapleStory content in my diary, so spoiler warnings are frequently unnecessary. Not in this section (“Health point challenges”), though! This is relatively new, MapleLegends-exclusive content.

Okay, phewf. Time for more Questing With tara™ now, right? Right. But these quests are… a little different.

You see, MapleLegends’s patented HP Challenges™ have finally been released… sort of. The first two tiers of the system are pretty much fully functional at this point, whereas the others are coming Soon™. I spoke with the HP Challenges NPC, Rosa, to ask about the MAXHP rewards that I’d be able to redeem by completing the Challenges:

Hmmm… So basically nothing — at least, not until tiers ≥5 are released, and I complete all tiers up to & including that point.

People often assume that this underwhelming calculus is the result of my having HP washed previously. That certainly is a factor here, but not very much so; I never had more than 40 base INT at any given time, & I stopped washing entirely at around level ≈50 or so. The real kicker here is simply natural MAXHP gain upon levelup. In this screenshot, I’m level 156, so I just naturally have way more MAXHP than the Challenges system expects me to have for the lower tiers 1〜4, which are geared towards levels 100〜130 (56〜26 levels below me), respectively.

Nonetheless, I’m largely unfazed by this structure. In the long run, assuming that the later tiers actually do get released, I will end up with considerable MAXHP rewards. Moreover, I’m not really in it for the MAXHP, in this case — although it did provide the motivation to stop washing so long ago. What I really want are those ✨juicy quests✨.

The original beta testing version of the HP Challenges didn’t involve any quests in the conventional sense. This released version, however, adds onto the old version a fully-fledged questline for each tier, complete with many quests, NPC dialogue, varied tasks, characters & plot, and all that jazz. Super cool!!!

We still have the other non-questlike content from the beta versions of the Challenges system, but we’ll get to that later. For now, we’ll start with the good stuff.

Full transcript

As I did these Challenge quests — including the non-questlike parts, for that matter — for the first time, I kept a full transcript, for reference. This made questing take easily an order of magnitude longer, because I insisted on writing absolutely everything down verbatim. It is painfully fastidious, & if for some reason you’d like to make use of it yourself, you can find it here:

[Challenge: Tier 1] Rosa and the Diary of Challenges transcript

The Markdown source can be found here: hp-challenges-tier-i.md.

I tried to reproduce the errors in the in-game text faithfully, so if you see typos, they are probably(!) in the original.

The formatting is basically ad hoc, & it’s all in Markdown. I used heading elements (#, ##, ###, etc.) to create the necessary hierarchy of quests, subquests, etc., which worked okay enough in this case, although I am aware of its limitations.

NPC dialogue & item descriptions use MapleStory’s own bespoke textual markup system. This is fairly accurate to the actual way that the data are represented programmatically in the game (read: in the WZ data). The main exception is that I use literal line feeds (U+000a) to represent what would actually be CRLF (U+000d U+000a), & would normally be represented in handwritten code via C-style escape sequences: \r\n.

MapleStory’s little markup language isn’t terribly complex, so here’s a little cheatsheet that you can use for reference:

If the rendering on my website somehow doesn’t work, you prefer not to read the raw Markdown, & you prefer not to render it yourself, then you can just use Codeberg’s rendering:

[Challenge: Tier 1] Rosa and the Diary of Challenges transcript (rendered to HTML by Codeberg)

Although it is quite fastidious, there are nonetheless some things missing from my transcription: branches of dialogue that I didn’t manage to find, text from the quest journal entries, etc. Maybe I’ll be able to add some of these things on a second pass, on another character…

Part I: The breaking of Arel

Our quest begins with Vikin, who is happy to see me again, so long after I helped him recover that “Pirate’s Map”. Vikin needs another favour, but this time, not to recover a map, but to recover his booty.

No, not that kind of booty. Gold. Gold bars, gold coins… Vikin keeps his booty locked up in a secret place, but JM From tha Streetz nevertheless reported that some of it may’ve been stolen. So I headed to Kerning to check it out.

JM claimed that he saw someone furtively sprint into the sewers & accidentally spill a few coins on the way in. JM suggested that if this mysterious individual were so careless in their haste, then they probably left behind at least some physical evidence of what they were doing. So, I headed into the sewers to check it out.

Unfortunately, JM didn’t really give any concrete instructions other than “check near the sewer entrance”:

That’s all I know, seriously. Don’t trust me? Then go ahead and check in the sewers. The coins I found were here around the entrance. If you’re luck[y,] you might find some more inside.

So… I tried that. I didn’t see anything that looked suspicious, & when I tried moving my mouse pointer over either side of the sewer entrance, there was still nothing of interest. I tried killing the Ligators & Jr. Neckis of The Swamp of Despair I (the only sewer map near the entrance) — after all, the coins might’ve looked tasty to eat. Perhaps they were just chocolate gelt… Heck, I even tried beating up the Witchgrass plants.

I tried poking around a little more, & thought for a brief second that I might’ve figured it out:

I got this message as soon as I entered Swamp of the Jr.Necki, but soon realised that it was for a different part of this Challenge tier, which we’ll get to later.

I had now spent quite a while on this very first part of the questline that wasn’t just reading dialogue. Taima had already done this part, so I asked for a hint. She said that there was indeed a hidden NPC that you had to find by moving your mouse pointer all over, but didn’t tell me where it was, in order to let me figure it out on my own.

It took me an embarrassingly long time, & I got quite close to just asking Taima to spoil it for me, but I did eventually find it myself:

Yep! It’s completely unmarked. Not even a little one-pixel sparkly thingy. The good news is that this questline gets the poorly-designed stuff out of the way first, & the rest is actually fun.

In the hidden spot, I found some more of the gold coins:

I brought the coins back to Vikin, who confirmed that they had indeed been shaken loose from his booty. The only logical conclusion, then, was that someone had found his ultra-top-secret hermetically sealed booty hole, which was guarded by none other than Dances with Balrog. When I visited DwB about it, however, he was shocked to learn that I even knew that the booty was stored in the Warriors’ Sanctuary, & he insisted that not just anyone was allowed to plunder Vikin’s booty.

Indeed, even though my intent was merely to investigate, DwB would only be willing to trust someone with the pure soul of a true warrior. I’m a pirate, not a warrior, so it was game over for me.

Then, Blackbull noticed me exiting the Sanctuary with a dejected look on my face. When I explained my situation, Blackbull suggested that I was strong enough to obtain a Wild Amulet^[11] of my very own. Only a true warrior could possibly sport such a sacred amulet, so it would surely be enough to convince DwB.

The main component of a Wild Amulet is a Fierce Fang, which can only be obtained from Red Drakes. So I headed to Dangerous Valley II:

After some farming, I found a Fierce Fang of my very own:

But taking this Fierce Fang & merely stringing it onto a necklace wasn’t going to cut it. First, the fang needed to be empowered, & there was only one person with the alchemical knowledge to do such a thing: Sophia. So I made my way to Sophia’s general store.

I was pleased to find that Sophia was willing to teach me the necessary alchemical ritual! She read the following passage from the tome of the Wild Amulet:

“First, you must pour the quickness of a wild hog, to reach superior speeds while pursuing.

“Second, you must pour the agility of a nimble monkey, to hone your reflexes while dodging.

“Third, you must pour the keen eye of a sharp eagle, to improve your aim while hunting.

“Fourth, you must pour the strength of a hunter, to increase the limit of your might.

“And last, you must pour the wisdom of the ancient gods, to gain the instinct of the natives, and thus seal the energy within.”

Thankfully, Sophia already sells each one of these five components in her general store:

1. Speed Potion.
2. Dexterity Potion.
3. Sniper Potion.
4. Warrior Potion.
5. Magic Potion.

It was then time to test my understanding, & to make an Empowered Fang:

With the fang now empowered, I thanked Sophia & went back to Blackbull to have the necklace made. With the Wild Amulet around my neck, DwB now recognised me as a true warrior, & allowed me to inspect the Treasure Room:

DwB was astonished to see that Vikin’s booty had indeed been stolen, right from under his nose. He was certain that no one could possibly have passed through the Warriors’ Sanctuary whilst evading his ever-vigilant gaze. After some investigation, I found that there was a strange goopy residue left behind where the booty had once been. DwB suggested that I take it to Sophia, so that she might identify the substance.

Sophia, however, was stumped. It was nothing like any of the chemicals that she’d worked with before. She suggested that if anyone had information related to this goop, it would be Jay of Henesys. When I arrived to meet with Jay, our conversation wasn’t off to a great start:

Jay: Oh, hello. Do we know each other? You don’t look like the scholarly-type.

tarandus: I actually am! I love reading ancient books. I finished my last tome yesterday!

Jay: You… definitely aren’t. But it makes me wonder, why would you lie to me? What brings you here… are you looking for information of some kind?

Rude!! Hello??

Whatever. After a bit more chatting, I was able to convince Jay to let me into his library:

Cool!! I’ve never been in here before!

Jay isn’t an alchemical expert like Sophia, but his library is quite wide-ranging, so we browsed it, hoping to find something relevant:

This tome is titled “Invisibility and You”! Maybe that’s it!

…The pages are all empty…

Eventually, Jay found this recipe in a book:

Purple Translucent Potion

A potion that grants transparency to whoever consumes it. The bigger the consumed quantity, the stronger the effect. A full flask will grant temporary invisibility.

Warning: prolonged exposition to the air will turn the liquid potion into a sticky gel waste.

━━ Ingredients: ━━

• 2 Violet Sagecap mushroom caps,
• 1 Giant Fern frond,
• 1 root of Black Starshroom,
• 3 minced Woodmic cloves,
• a pinch of Skydust salt.

A-ha! A “sticky gel waste”, you say? My natural inclination was to attempt to reproduce this potion by following the provided recipe, but… no one seemed to know what the ingredients list was about. Harvesting Violet Sagecaps is doable, as they grow all the way up at the Top of the Tree That Grew…:

…But what about the others? “Black Starshroom”? “Woodmic”?? “Skydust”??? Neither I, nor Jay, nor even Sophia had ever heard of such things. Luckily, the recipe is attributed to an author: Wing the Fairy of Ellinia.

I met with Wing, & before I could even mention the recipe, he already knew what I was going to ask. He could see that I’d harvested a Violet Sagecap — a fungus with no common uses. Wing explained that the ingredients of the recipe — excepting the Violet Sagecap — were given fake names to prevent anyone from successfully using the recipe without first consulting him personally. That way, not only did he have the power to withhold the necessary information, but even if he did give the real recipe, he’d have a list of people who knew about it.

Wing remarked that I’d be the first person to be put on said list. But that didn’t add up: either we had the wrong alchemical recipe, or Wing was the thief whom I was after… or someone else figured out how to make this stuff. Wing admitted that there was one other person who knew about this recipe: his childhood friend Arel, with whom he’d originally invented the potion. But Wing was adamant that not only was it beyond unlikely that someone had independently rediscovered the potion, but also that Arel was perfectly upstanding, & would never resort to such theft. Although this was starting to sound a bit self-incriminating, Wing did have an alibi, so I decided to take his word for it.

The actual remainder — apart from the Violet Sagecap — of the recipe was:

•   1 Lemon.
• 100 Curse Eye Tails.
• 200 Malady’s Experimental Frogs.
• 100 Clang Claws.
• 200 Tortie Shells.

So I headed to the nearby Forest of Evil II, to hunt some Maladies…:

…And up north, to TDFUNIX^[14], to farm some Curse Eyes…:

…Over to the Hot Sands of Florina Beach, to grind some Torties & Clangs…:

…And finally, took out that mouldy Lemon that’s been sitting in my inventory since… the last time that I scrolled up that far. It’s fine. I’m sure it’s still good.

Of course, I’m no alchemist. With this whole heaping pile of ingredients in my arms, I slowly made my way back to visit Sophia again. When I got there, though, Sophia had other ideas: she wanted Jay’s copy of 101 Recipes and Potions with Ancient Trees Sap. So I dropped my pile of several hundred random monster body parts (& one Lemon of questionable quality) onto the floor of Sophia’s general store, & went back to Jay, to check a book out from his library. Jay found the book for me, & I returned to a contented Sophia, who was now willing to help me prepare this “Purple Translucent Potion”…

Much like with the empowering of the Fierce Fang, Sophia tested my understanding of the recipe in question:

(Let’s prepare the Curse Eye Tails…)

• ☛ Mince them.
• ☛ Peel them.
• ☛ Add salt to them.
• ☛ Cut them in slices.

Once we were finished brewing, it was time for me to put this alchemy to the test. I took a swig of the potion, & entered the Warriors’ Sanctuary as silently as I could.

DwB didn’t seem to notice, & so I was able to enter the Treasure Room unimpeded. To prove that I’d really pulled a fast one on him, I took one of the coins out of the Treasure Room. DwB was shook. How could I have done it?? Recklessly ingesting bizarre & mysterious alchemical products, that’s how.

Now that I’d confirmed that the Translucent Potion^[15] was capable of enabling this crime, & confirmed that its residue was identical to the residue left behind by our perpetrator, I had one strong lead: Wing’s childhood friend Arel. Wing continued to insist that Arel would never do something like this, but also gave some useful description of Arel: tanned skin; hazel eyes; dark brown hair, always dishevelled; wears gold jewellery; and a native of Sleepywood who went to Ellinia to study.

When I reported all of this to Vikin, he had a rather belated confession to make: he & DwB were not the only ones aware of Treasure Room. Jr. Balrog & his cronies had such fierce control of the air & waters surrounding the Victorian Archipelago, that it made practical sense for Vikin to offer some of his booty to the Balrogs, in exchange for a guarantee of safe passage to & from Victoria.

Of course, the figure that JM saw was certainly much too small to be a Balrog. Nevertheless, considering that Arel was a Sleepywood native, considering that he was found running through the Kerning sewers towards Sleepywood, & considering that Jr. Balrog’s Cursed Sanctuary is buried deep in the heart of the Sleepywood Dungeon’s Ant Tunnel, I knew my next destination.

I asked around at Sleepywood, questioning the locals about Arel, but no one seemed to have any useful info. Amongst them was Sabitrama, who thought that I looked quite weary, & could perhaps use a relaxing rest at the renowned Sleepywood Hotel. Of course! I confirmed with the receptionist at the that Arel had passed through, briefly staying at the hotel, but also that he was gone now.

As I entered ATI, I spotted a figure whom I quickly identified with Arel:

Noticing that I’d spotted him, Arel made twofold his pace. So I gave chase:

Finally, at TTLLII, Arel, in his great haste, tripped over some of the jaggèd blue rocks:

The look on Arel’s face was one of remorse. He explained that he never wanted to steal anything, but that he had no choice, if he wanted to survive to see his own family & community intact. Jr. Balrog had so cruelly recruited Arel’s expertise — offering in exchange a better life, wealth, & access to incredible scholarly knowledge — in order to steal the entirety of Vikin’s booty for itself. When Arel refused to assent to such a devil’s bargain, the Balrog played a heinous trick: to manipulate the minds of Arel’s family, friends, & the other Sleepywood locals, to cause them to forget about Arel’s existence entirely. Jr. Balrog threatened Arel, claiming that only Arel’s collaboration could reverse the memory curse, & that if he still refused, the Balrog would only intensify the curse to cover the whole of the Victorian Archipelago.

And so it was just as I’d suspected after Vikin admitted to bargaining with the Balrogs: Jr. Balrog was the real culprit here. Arel didn’t have the booty; he’d already handed it over to Jr. Balrog. So I promised Arel to take care of the beast, & headed towards its Cursed Sanctuary.

Just as I moved to enter the Sanctuary from its entrance, however, a strange magical force sent me all the way back to Another Entrance. When I appeared there, the Insignificant Being greeted me. I was already quite familiar with the Insignificant Being from A Spell That Seals Up a Critical Danger, & this time, I had to ask him to help me do the opposite: to take down the very spell that seals up this critical danger, so that I could get in there & beat the piss out of Jr. Balrog.

Of course, much like last time, doing such a thing would require no small quantity of elbow grease. To weaken the barrier, I’d have to slay a thousand each of Ice Drakes & Dark Drakes. Moreover, once the barrier was weakened, a counterspell would be required, in order to shatter it altogether. For the counterspell, I would need to gather up ten Thunder Catalysts from Tauromacis & Taurospears:

Since Taima was also doing this questline, I teamed up with her to farm Sanctuary Entrance IV:

These Thunder Catalysts are no joke. Although we only needed ten each, their droprate was not particularly high, especially given that Tauros spawn at a quite leisurely rate even in Sanctuary Entrance IV, which is unfortunately the best place to farm them.

And we likewise teamed up to grind Ice & Dark Drakes — first at Drake’s Nest, & later at Cold Cradle, when we realised that Drake’s Nest had a disproportionately large number of the Dark Drakes:

Phewf! That was a lot of Drakes!!

With the sealing spell weakened, the Insignificant Being put together a counterspell that demolished it. Now able to enter the Cursed Sanctuary, I found myself faced by a vexed Jr. Balrog who was anticipating my arrival. So I gave him a taste of my fists!!:

With the beast incapacitated, I scooped up Vikin’s booty, & scrambled all the way back to the Treasure Room. As I departed, Jr. Balrog spat curses in my direction:

Aaargh… You…! Who are you? How dare your steal what’s mine? Come back here! Curse you!! Damned human!

DwB, & especially Vikin, were of course very pleased to see the booty returned to its rightful place — although it could perhaps be argued whether the hands of a pirate is the “rightful place” of some booty that they plundered to begin with.

In any case, although we got a kind of happy ending, there are still some loose ends & things that just don’t add up. What other nefarious things could the Balrogs be planning? And more importantly, how did Jr. Balrog come into such incredible powers of massively mind-altering magic, so far beyond the powers ordinarily attributed to Balrogs??

My little review of The breaking of Arel

This mandatory questline for tier 1 of MapleLegends’s HP Challenge system is entitled Vikin’s Treasure (here, I rename it to a more spoilery name: “The breaking of Arel”), & is totally custom content made for MapleLegends specifically. As far as I can tell, these mandatory questlines are what make the HP Challenges really worth doing, even beyond the promised MAXHP rewards: although the system’s raison d’être is making MAXHP balanced, the ultimate reality is that it is game content.

I can sum up my overall opinion pretty succinctly: it’s very good.

It goes without saying that I have some personal bias here, as I’m a huge sucker for questing. But even further, The breaking of Arel is just very high-quality questing that stands up well to the so many other quests in MapleLegends, which are generally designed by Wizet–Necksawn. We have:

• Quests that are varied & don’t get terribly bogged down in grinding a single quest for overextended periods of time. The breaking of Arel has various flavours of combat quests: collecting ordinary ETC items, grinding kills, collecting rarer quest-only items, & even a boss fight at the end! And there are also non-combat-oriented components to mix it up: Sophia’s alchemical quests, as well as some reasonably interesting dialogue that occasionally asks the player to “choose the good options”.

• An interesting plot that starts very small, with “someone stole my treasure 😢”, & then unfolds itself into a satisfying story that still leaves room for the later parts — this is just tier 1, after all!

• Something that fits snugly within its medium of “MapleStory quests”, & makes profitable use of the material that’s available to it. We see the appearance & reappearance of familiar faces from Victorian quests: Sophia, Vikin, JM From tha Streetz, Wing the Fairy, Insignificant Being, etc., & we see them in a new — or in the case of the Insignificant Being, a familiar but refreshed — context. Sophia is perhaps a lot cooler than Maya and the Weird Medicine & “Sophia’s Snack” would clue you in on. And of course, the classic villain of Victoria — & of MapleStory in general — makes its appearance as the antagonist of this part: Jr. Balrog.

  It’s worth noting how much The breaking of Arel benefits from taking place entirely within the Victorian Archipelago, & I’ll definitely be doing tier 1 on my Viclocked dagger spearwoman d34r at some point!!

That being said, there are still a few flaws, & some room for improvement, like with any questline. The obvious flaw is the first quest, where the invisible NPC must be located entirely through the arbitrary use of the mouse pointer. A tiny visual indicator, like the one used for the second 2nas quest, could go a long way, for example.

I’m also a big sucker for written dialogue — as careful readers of my past few diary entries may have noticed… — and so I spotted some room for improvement there, as well. Quite frankly, the bar is usually set pretty low here, partly — but certainly not entirely — as a result of so much dialogue in GMS (upon which MapleLegends is based) being sloppily translated or half-translated into English from the original (usually Korean, or Japanese). By comparison, The breaking of Arel easily holds up. Still, there are a few small things that I’d like to see:

• There are a few typos that may be fixed, of course.

• Some of The breaking of Arel’s dialogue has readability issues arising from poor formatting. The most glaring one is failing to distinguish between soft & hard word wrapping: MapleStory NPC dialogue automatically soft wraps when it runs out of horizontal room, but this effect is visually identical to there being a \r\n sequence in the dialogue string itself. This is not to say that a single \r\n is useless; in fact, it has plenty of uses, & The breaking of Arel makes several such uses. But this is to say that paragraphs in particular must not be separated in this way; they should instead be separated by \r\n\r\n. There are also some issues with consistency & style, but the paragraph problem is the main one here.

• The breaking of Arel certainly doesn’t do a terrible job of this, but I do like to see a little more distinction in attitudes & styles of speech, when it comes to the various NPCs throughout various quests — & throughout various mental states. Certainly one of the greatest powers of reading (or hearing) dialogue from a variety of characters is simply the colourfulness & variety that it brings.

• I quoted this little snippet of dead-end NPC dialogue above:

  This tome is titled “Invisibility and You”! Maybe that’s it!

  …The pages are all empty…

  Of course, this is a bit of a silly little gag, but that’s the point. The injection of this kind of humour breathes a surprising amount of life into the content of a game like MapleStory; everything is cartoonish to begin with, some things & game designs are silly, characters may behave irrationally, & any of these things can sometimes be a source of humour. Moreover, this can still be tastefully applied even in quests that are otherwise tragic or serious. So, you know, I’d like to see more of that… personally…

Everything else

What a strong start. Now it’s time for the disappointing part: the entire rest of tier 1!

The Test of Strength

The other mandatory quest(line) isn’t much of a quest, & definitely isn’t a questline. Instead, it consisted entirely of Kyrin telling me to deal 50M damage with Energy Blast, saying “go get ’em, kiddo”, & paddling me on the bum. Well… minus the part where she touched my butt. That would be weird. Oh, and there’s also the part where I had to Energy Blast a bunch of Rāvaṇa summons until the quest was complete.

This quest is obviously super boring, so there appears to be… not much to say about it. That being said, although it is boring, it may or may not be tedious — or even impossible — to complete. Consider that this is a tier 1 Challenge, meaning that it is specifically designed under the assumption that the PC is in the level range 100〜109 — with the caveat that they might actually be level ≥110, like in my case. SP builds for marauders are not trivial, & choosing whether to prioritise EB — or even whether to bother setting aside SP for it at all — is part of the experience of playing a marauder/buccaneer, just like SP builds are part of the experience of playing almost any class in MapleStory. Personally, if I were to play tara over again, I would go the Shockwave route, because I think it’s actually a pretty underrated skill — indeed, I still use it fairly frequently, even as a level one hundred & fifty-something buccaneer who only levelled it up to 18 (out of a maximum level of 30)!

Kyrin, then, seems to be insisting otherwise, for some reason: the level 100〜109 marauder is expected to have already invested significant SP into this skill, & if they haven’t, then their “strength” mysteriously cannot be accurately “tested”. Worse, I’ve heard second-hand reports of HP Challenge quests requiring Magic Crash, despite white knight & paladin being notoriously SP-starved classes. And so on…

And, naturally, it gets even worse: these quests can completely exclude entire heaps of character builds from this entire category of game content, especially because the full completion of one entire Challenge tier is a hard requirement for the commencement of any part of the next tier. As much as I’d absolutely love to, I won’t be insulting the reader’s intelligence by presenting a huge list of such character builds. Instead, it’s much easier to simply ask: why? The “Tests of Strength” are boring, in many cases don’t even warrant the player doing anything other than playing their character normally, contribute nothing to this game content, & are weirdly incredibly specific in their requirements, for no ostensible reason.

Grab-bag of optionals

Then, we have the optional Challenges. These are, of course, not totally optional, but rather, are a “do any 𝑛 or more of these” sort of affair.

These Challenges have plenty of potential if the author is willing to put in the creativity, & they’re also forgiving enough, thanks to the player’s ability to choose which ones they want to complete. The result is basically what you’d expect: a mixed bag. For tier 1, there are eight of them, & at least three must be completed. I’ll leave a brief comment on each:

•  5 Crimson Balrog kills.

  This is the “bossing” optional for this tier. For tier 1, it makes sense, although of course it’s a bit annoying to find these Crogs.

•  5 weddings attended.

  Cute idea, & a great way to tie Amoria into this tier. Unfortunately, I don’t really expect that anyone will actually do this one — five is just kind of a lot, so you’re likely better off doing some other optional.

• 20 Owls of Minerva used.

  What? Oh, I get it. HP Challenges replace HP washing for some people, so you need a new way to get said people to still vote on GTop100. Sure… Come to think of it, the “5 weddings attended” optional might also have something to do with this…

• 10 unique Victoria Island maps visited.

  This one is actually kinda fun! I’ll talk about it in some more detail.

• 10 100% scrolls passed.

  What?

  

•  3 Victoria Island JQs completed.

  I actually like this one, wow. I used to be so bad at JQs that I had no choice but to despise them. Now, I can actually do some of them! Especially the Kerning & Sleepywood ones!! High-quality JQing.

• 10 equips crafted with Chrishrama.

  I understand. Chrish needs some love.

•  3 sets of Omok or Match Cards crafted.

  Huh? Anyway, it is a little sad that I actually do craft these, & they don’t count towards the relevant HP Challenge unless you actually have the Challenge active at the time that you perform the craft — which wasn’t previously possible…

The “10 unique Victoria Island maps visited” optional is the one that I accidentally triggered when I entered the Swamp of the Jr.Necki above. I do like the idea of a travel optional, especially one that encourages exploration & going to maps that are rarely used, or even rarely set foot in. I found nine out of the ten “unique” maps by literally just running across every Victoria Island map that I could think of. Then, I got a little lazy & cheated to find the last one, because I forgor that the road to Mushmom’s house was a thing. Can you blame me? Mushmom has been hunted to extinction, after all…

When I visited some quite unique maps & didn’t get credit for them, I started to think that maybe the list could have been longer. For example, The Pig Beach doesn’t count, despite it counting for a beta version of the Challenge. My precious baby CFD isn’t in there, either. TDFUNIV isn’t in there. Hut in the Swamp isn’t in there. Mushroom Garden isn’t in there. And so on… Perhaps, instead of requiring ten out of ten, 18 out of 20 could be required…?

Ellinia’s guarded forest

In any case, another one of the HP Challenge tier 1 optionals that I liked was the “3 Victoria Island JQs completed” one. This just so happened to coincide with a certain pair of quests that I intentionally skipped so long ago on tara: “Sabitrama and the Diet Medicine” & “Sabitrama’s Anti-Aging Medicine”. tara? Skipping quests?? On purpose???

Right, so, here’s the deal: Sabitrama is Sleepywood’s local snake oil salesman. He wants some ingredients to make some weight loss pills & some anti-ageing pills, but there’s a catch: you have to climb The Forest of Patience in Ellinia to get them. Shane (Roger’s name when he’s not doing maple leaf cosplay) will let anyone into The Forest of Patience for a fee, but the way up is perilous. I was always pretty shite at these JQs, & they’re considered the most difficult Victorian JQs.

So, it looks like it’s time to finally clear these JQs for the first time ever in my entire Maple career. Then, I’ll only need one more Victorian JQ completion to finish that optional!

Spikes. Logs. Monkeys.

As it turns out, these JQs might be good MapleStory Physical Fitness Test (“Fitness”) practice. Unfortunately, my Fitness skills are still not that great, even after so many Fitnesses on my Viclocked clericlet d33r… 🥲

But I did it! I got to a pile of flowers!!:

I was relieved to find that this JQ actually… wasn’t terribly difficult? Then I realised that the real difficulty is, of course, in the second quest.

As you can see in the example above, the “Anti-Aging Medicine” JQ starts out with a bunch of really tight jumps. There aren’t even any moving parts nor obstacles yet; just pure jumping. These jumps are very much not trivial, however, & I definitely fell all the way down at least a few times. Nevertheless, unlike the last time that I was here (on my Viclocked dagger spearwoman d34r), I was eventually able to nail all the jumps! That actually makes these jumps easier than the ones in the Ghost Chimney (see the “The mystery of the Prendergast Mansion” section above)…!

That’s not all, though. Once I got past the austerity of the static jumping bits, things started moving around & threatening to yeet me:

Oh no… I’m having Hinamatsuri flashbacks (see the “Hinamatsuri” section of the previous entry)…

And finally, towards the end, there are the Lupin diagonals:

There are three of these bad boys in step 5, & they’re kinda nasty, especially when you first see them. Going up three or four diagonal jumps takes a while, so I think it’s mostly a matter of timing. If you’re a pro speedrunner, or you’re like me and just fuck up, you might take a hit from a banana, & hopefully shake it off.

I don’t mind timing them, though, so I was able to make it through the Lupin diagonals, thus arriving at a pile of herbs!!!:

I did it!! I beat JQing!!! I’m an epic JQer now!!!! 😎😎😎

But I still needed that one (1) more Victorian JQ completion for my HP Challenges, so I did Kerning’s B2 JQ once again…:

JQ’d.

Puckishness

Holy moly guacamole. That was a lot of fuccing Questing With tara™!! Hecc ya!!! To wrap things up, how about we see tara level up one more time…?

I did some duo Rāvanning & Papping with Taima. As it turns out, I can’t really pin Rāv to the left and take care of his summons at the same time…

But so many gobby goblins are no match for 🦕tarasaur🦕! rawr xD

Plus, this was how I completed my “Test of Strength”…

And, after this fateful Pappening, I was at 99.9% EXP?? So I headed out to the Warped Passage, to level up by killing a few newly-buffed Gatekeepers!!:

Footnotes for “In quest impressed, & all the rest”

1. [↑] See footnote #2 of the aforementioned diary section, for more on the name of this monster species.

2. [↑] See footnote #4 of the “Down the rabbit-hole” section of pt. cvii of this diary, for more on the presence of musical boxes in MapleStory.

3. [↑] Not a recognised specific name these days, as far as I know.

4. [↑] Hebrew ⁧חֲנֻכָּה⁩ (⟨Ḥanukkah⟩; Modern Hebrew /χä.nuˈkä/) or ⁧חנוכה⁩ (⟨Ḥnwkh⟩, ⟨Ḥ[a]nuk[a]h⟩; /ˈχä.nu.kä, χä.nuˈkä/). English /ˈhɑ.nə.kə, ˈhʌ.nʊ-, ˈx-/. Also commonly transliterated into English as ⟨Chanukah⟩, amongst many others.

5. [↑] Due to Gina Glasman, as quoted in Deena Prichep; “Hanukkah History: Those Chocolate Coins Were Once Real Tips”; NPR: The Salt; 2014-12-13; archived.

   Most sources seem to credit Loft’s Candies™ for the innovation of chocolate gelt sometime during the 1920s. The exact origin is not super relevant here, however, so I’m relegating this to a footnote.

6. [↑] Due to Rabbi Menachem Creditor, as quoted in Rachel E. Greenspan; “The Real Reason American Jews Give Gifts During Hanukkah”; TIME; 2019-12-20; archived.

7. [↑] By pure coincidence, I happen to be writing this exactly on the first day of Ḥanukkah! By the time that this is actually published, though, it’ll probably be like, March or something. 🙄

8. [↑] See:

   • BBD; “What Came When in GMS”; BasilMarket; archived 2020-11-12.
   • JohN247; “Returning to Maple? Here’s what you missed.”; GameFAQs Message Boards; 2008-07-07; archived.
   • DigitalTQ; “MapleStory Patch History — What Came When — Timeline Before Big Bang”; DigitalTQ blog; 2022-05-17; archived.

9. [↑] This version fleshed out Amoria somewhat, adding some NPCs, quests like Circle of Trust, as well as a single map that has monsters: Purple Plains 1. APQ would not be added until v39 (2007-06-13), however.^[8]

10. [↑] Also erroneously spelt in-game as *Cronos' Egg.

11. [↑] Whenever MapleLegends updates their Library again (read: the year 2999), these will stop producing HTTP 404s.

12. [↑] I do question the choice of atualizado here, as it appears to mostly be used in Portuguese to mean “updated; modernised”. Maybe completo “completed” makes more sense…?

13. [↑] Sic. The correct spelling is Torr.

14. [↑] I’ve always pronounced this as /ˌti.diˈfʌ.nɪks/ in my head. Is that weird…?

15. [↑] The actual item is called Translucent Potion, & is often referred to as such, although the recipe calls it Purple Translucent Potion.

(…cnvpstdf…)