Skip to the entry’s beginning

rangifer’s diary: pt. cxxxii

Appendix A: Classicality in depth

Throughout On odd jobs, I make the following essentially tripartite distinction between jobs:

Classically odd

The strictest form of oddness.

Nonclassically odd

Jobs that are “odd enough” for our purposes (whatever those purposes may be), but that aren’t classically odd.

Nonodd

Jobs that aren’t odd at all — “normal” jobs, “standard” jobs, or what have you.

On odd jobs isn’t aimed at rigorous internal ontological discussion, but is instead dedicated to the basic philosophy of odd jobs, to dispelling misconceptions about odd jobs (& jobs in general), & ultimately to an argument for why odd jobs are for everyone — as the commonly-conceived gameplay-style barrier between odd & nonodd is in fact entirely illusory. For these reasons, the above tripartite scheme furnishes more than enough analytical power for the purpose of On odd jobs.

However, we’ll have to get significantly more precise if we want to treat these kinds of demarcations in depth.

Deprivation ontology

We often think of odd jobs in terms of self-imposed restriction. This is — not to put too fine a point on it — entirely backwards. The PC’s job is no more than a result of the choices that its player freely makes; that is, it’s the player’s relative unrestriction that allows her to define her PC’s job, simpliciter. This is equally true for the odd job as it is for the nonodd job — both are “self-restricted” in precisely the same sense.

Nonetheless, as explored in the main essay, “oddness” is defined in opposition: odd jobs are odd because they’re not normal. Thus, if we want a strictest(!) definition of “oddness” that’s also reasonably rigorous, then we need to evacuate it of the not-odd.

To this end, we first need to know what elements constitute a job (odd or not):

• Class.
• AP allocation (= base stats).
• Weaponry & ammunition.
• SP allocation.

Class

For reasons explained throughout On odd jobs (see especially: “Late grade-advancement restriction”), certain class choices are already odd on their own. In particular, perma-0^th-grade jobs (= permabeginners) & perma-1^st-grade jobs (= permafirsts) are always odd, even under the strictest possible definition.

However, many odd jobs seem to be defined in terms of class without being odd because of class alone. This is where a more involved ontology starts to become necessary.

AP allocation

Like with class, it seems as if some jobs defined in terms of AP allocation can be odd because of that allocation alone — but most cannot.

We therefore introduce the tripartite AP régime: there are primary stats, secondary stats, & tertiary stats. These derive from how damage dealt by PCs is calculated in all versions of pre-BB MapleStory: the primary stats have mastery & a nonunit coëfficient (which I’ve called a PSM = primary stat multiplier) applied to them, the secondary stats have no coëfficient (or equivalently, a unit coëfficient), & the tertiary stats don’t factor into the calculation at all (in effect, a tertiary stat is just any stat that’s neither primary nor secondary).

We define these categories in relation to class — even though they’re arguably more related to weapon type — for the following reasons:

• At least one weapon type (viz. dagger) changes its tripartite régime entirely, depending on its wielder’s class!
• Magical attackers simply do not observe weapon type in this way. We’ll return to this point.
• The PC must commit to a class (at most, they can grade-advance), whereas weapon type can more-or-less be changed on the fly.
• Class is simply more deeply linked to job, to the point that the class vs. job distinction almost(!) ceases to be useful if all odd (sēnsū lātō) jobs are eliminated from consideration.
• Primary stats are partly defined in terms of mastery. The only way to gain mastery (i.e. to have more than 10% mastery with a given weapon type) is through various skills, all of which are restricted to classes of 2^nd or higher grade, & each of which only applies to a single weapon type (or sometimes to a pair of weapon types that differ only in handedness). Thus, the “primaryness” of a stat cannot be fully realised in relation to weapon type alone; it must also be combined with a class that has a relevant mastery skill.

With that, we get the following régime:

		primary	secondary	tertiary
beginner		[not applicable]
warrior		STR	DEX	INT, LUK, maxHP, maxMP
mage		INT	LUK	STR, DEX, maxHP, maxMP
archer		DEX	STR	INT, LUK, maxHP, maxMP
thief		LUK	STR, DEX	INT, maxHP, maxMP
pirate	pirate	[not applicable]
	brawler	STR	DEX	INT, LUK, maxHP, maxMP
	gunslinger	DEX	STR

The first thing that we notice is that we cannot define the categories for certain classes. With the swordman, magician, bowman, & rogue, we’re able to sneakily backport the categories from the throughclasses that they branch off into — but only because the categories just so happen to be the same for all branches of the given archetype.^[1] With the beginner, this method is clearly impossible; but the 1^st-grade pirate has essentially the same problem as well. The fact that this “problem” only exists at grades below the 2^nd doesn’t just coïncide with, but in fact corresponds to, what makes perma-0^th-grade & perma-1^st-grade jobs strictly odd.

Properly speaking, the status of LUK as a secondary stat for the mage typically doesn’t apply on the basis of damage calculation alone (at least in post-alpha versions of the game). Nonetheless, we can feel confident in the mage’s entry within the above table for the following reasons:

• Heal, the only magical attack with damage calculated analogously to that of a physical attack, has LUK as clearly a secondary stat.^[2]
• LUK very obviously patterns as a secondary stat in terms of stat requirements on mage equipment, & stats bestowed by said equipment.
• LUK has equal standing with INT for the purpose of MAcc; but no other stats are involved (indeed, these two stats are the only sources of MAcc in the entire game).
• Making LUK a tertiary stat would, bizarrely, make the mage the only archetype that allows for no secondary stats in spite of having well-defined primary & tertiary stats.

The reader might object to STR being a secondary stat for the assassin throughclass. But in fact, the claw-wielding assassin’s damage output scales up directly with STR, with the sole exception (amongst attacks that take the assassin’s stats into account) being L7/TT.

Odd AP

In any case, strictly defining oddness at the intersection of class & AP allocation is now easy: the AP allocation is odd iff it doesn’t allocate any AP into any primary stats (except as absolutely necessary to take the 1^st grade advancement). Eliminating primary stats “evacuates” the possibility of nonoddness.

Under this definition, permabeginners & permapirates cannot have their stats factor into their oddness, even if we consider narrower jobs defined partly by stats (e.g. DEXginner). But this is a nonissue, since permabeginners & permafirsts are already strictly odd anyway.

The reader might’ve noticed that this definition doesn’t require a distinction between secondary & tertiary stats — both are equally nonprimary. But even apart from the fact that the secondary–tertiary distinction is important to defining certain groups of odd jobs (e.g. the secondary-statted or the blood combatant), we’ll later see that this distinction also factors into a different — but closely related — conception of oddness.

Moreover, the fact that maxHP & maxMP are always tertiary whenever this category is well-defined leads us to oddness at the level of AP allocation itself:

• A job is statless iff it’s STRless & DEXless & INTless & LUKless. All statless jobs are strictly odd for our purposes.
• A job is APless iff it doesn’t allocate any of its AP. All APless jobs are statless, but not all statless jobs are APless.

Weaponry & ammunition

In some ways, weaponry is here treated similarly to AP allocation. For a given throughclass, there are canonical weapon types — analogous to primary stats. In many cases, this is simple:

• The fighter has swords (of both handednesses) & axes (of both handednesses).
• The page has swords (of both handednesses) & blunt weapons (of both handednesses).
• The spearman has polearms & spears.
• The hunter has bows.
• The crossbowman has crossbows.
• The assassin has claws.
• The bandit has daggers.
• The brawler has knuckles.
• The gunslinger has guns.

But this omits the mages. We’re surely tempted to simply assign wands & staves to all three mage throughclasses; but the situation here is different, because mages are not canonically physical attackers. Inasmuch as “wands & staves” is the only possibility, we can eliminate all possibilities:

• The useful effects produced by a magical attack (including, but not limited to, damage) don’t observe weapon type in any way.
• Although we’re tempted to privilege wands/staves for TMA reasons, the fact is that weapon type is merely incidental to TMA. This is even evidenced in practice by many (& in some cases, most) nonodd mages wielding weapons of choice that are neither wands nor staves, simply for their stats: Green Paint Brush, Black Umbrella, &c..
• Although some magical attacks do observe weapon type to determine whether they’re usable at all, this is only on a handedness (one-handed vs. two-handed) basis, meaning that all one-handed weapons are lumped in with wands & staves.
• The sole skill that actually discriminates between wands/staves & other weapon types is Spell Booster. However, an entire mage throughclass (viz. cleric) never gets access to this skill! Moreover, this is a 3^rd-grade skill, & therefore both (1.) long postdates combat-style specialisation (i.e. throughclass choice), & (2.) doesn’t exist in many versions of the game.

Thus, for this purpose, we’re forced to lump the mages in with the permabeginners & the permafirsts as “not applicable”.

Methods of disarmament

To evacuate, via weaponry, the possibility of nonoddness, we have three options from which to choose:

• Evacuate the job of all canonical weapon types.
• Evacuate the job’s canonical weapon type(s) of all its ammunition (including of virtual ammo, e.g. Soul Arrow).
• Evacuate the job of weaponry entirely.

Of course, the second option only works for canonical weapon types that’re ranged, & thus can actually use ammo: bows, crossbows, claws, & guns. The third option only works for pirates, as pirate is the only archetype^[1] capable of attacking whilst unarmed.

SP allocation

And finally, there are other ways of undoing specialisation, apart from avoiding the 2^nd grade advancement. SPless jobs eliminate skills almost entirely.^[3] A more subtle variant only eschews skills that’re attacks; that is, the job is basic-attack-only. For this to be well-defined, we need an operational definition of “attack”.

Fortunately, the game itself provides a direct indicator of whether something’s an attack: a skill is an attack iff the skill itself puts its user into the alert state.^[4] Simple as that.

The attack

Given that the lexeme attack is already quite overloaded in MapleStory terminology, the reader might question my nomenclature here. However, this is simply the word used by the game itself, in its English-localised incarnations:

In any case, the attack or nonattack status of a skill has various game-mechanical implications beyond its simple definition. These are worth noting, so as to flesh out the notion of what “being an attack” really means — since operational definitions confer little insight:

1. If it can directly displace a monster’s position (like e.g. Rush), then it’s an attack.
2. If it directly deals damage, then it’s an attack. However, the key word here is “directly”; if a skill merely produces a temporary effect that, in turn, can result in damage being dealt independently of the player’s activity, then it’s not necessarily an attack.

3. If it cannot interact with the “aggro” mechanic, then it’s not an attack.

   Note that aggro is often rather complicated, so that successfully landing an attack on a monster doesn’t necessarily change its aggro-target in every possible situation. It suffices to simply interact with the monster’s state machine at all, thus producing an observable effect at least some of the time.

4. If it cannot be used unarmed, then it’s an attack.

5. Nonlethal damage is irrelevant for the purpose of determining whether something is an attack.

   Something is nonlethal iff it cannot last-hit^[5] monsters.

Examples for concreteness

Situationally an attack (i.e. these may or may not cause the PC to enter the alert state, depending on how the PC chooses to use them):

• When unarmed, Heal cannot target monsters, & doesn’t put its user into the alert state. When armed, the reverse is true. Heal is therefore an attack iff its user is armed.^[2]

Nonattacks (i.e. none of these cause the PC to enter the alert state):

• Shadow Web doesn’t satisfy (a.); it cannot displace monsters directly, but can — at best — merely cause monsters to cease walking or flying (& even then, this is a result of a debuff, not of the skill-usage itself). It does satisfy (c.).

  It doesn’t satisfy (b.), because although the SW debuff does deal damage, that damage is a result of the debuff, & not of the skill connecting with its target. The main giveaway here (without getting into implementation details) is that SW’s sole source of damage is damage-over-time (DoT), which thus requires its various damage instances to be independent of the SW-user’s activity (e.g. the user may deal damage even when entirely offline). Another big giveaway is that SW can in no way (directly or indirectly) deal damage to bosses (as a result of bosses being immune to most debuffs, SW’s included). Also note that SW damage is nonlethal.

• Ninja Ambush isn’t an attack, & behaves very similarly to SW. Neither Ninja Ambush nor SW can proc Venomous Star nor Venomous Stab, as a direct result of failing to satisfy (b.).

• Power Guard is perhaps the only really ambiguous case (see Energy Charge below, for why that seemingly-similar case is unambiguous); nonetheless, it deserves to be placed with the other nonattacks. Clearly, PG itself can be used without entering the alert state. Moreover, when it does deal damage, that damage is dealt indirectly (thus failing to satisfy (b.)) insofar as it’s a result of a temporary effect combined with the incoming attack of a monster (that is, not a PC’s attack). This interpretation is bolstered by the fact that PG’s damage is dealt as an aliquot portion of the damage incoming to the PC, thus divorcing its damage from the PC almost entirely; moreover, it has no effect when the PC is “MISS”ed. When the PC does eventually enter the alert state (if they do at all), it’s a result of being struck, & not a result of attacking.

• Summons — Octopus, Silver Hawk, Elquines, &c. — don’t target monsters at all. They therefore cannot possibly satisfy (a.) nor (b.). Summons can also be used unarmed (even by nonpirates!), & therefore cannot satisfy (d.).

  The fact that a summoned creature doesn’t target monsters is evidenced by the fact that it has full effect — up to & including damaging monsters that might spawn later on during the summoned creature’s lifetime — even when used on a map with no live monsters in it. If monsters are targeted, then, they must be targeted by the summoned creature itself, rather than by the PC.

Attacks (i.e. all of these cause the PC to enter the alert state):

• Aerial Strike causes the corsair to enter the alert state, & also easily satisfies (b.).
• Poison Mist is — perhaps unexpectedly — an attack, & it even satisfies (b.). Although the mist effect itself is merely a map effect that applies a debuff — & therefore doesn’t imply that Poison Mist is an attack — the skill also acts as an ordinary poison-typed magical attack in exactly the same way that e.g. Paralyse does. This aspect of Poison Mist is just so infrequently used intentionally that we tend to forget that it’s there. Poison Mist also satisfies (d.).
• Monster Magnet satisfies (a.), in spite of not really satisfying (b.) (but note that the latter isn’t strictly true, mechanically speaking, because MM can deal 0-damage lines that appear as “MISS”es). Furthermore, MM doesn’t satisfy (c.). Bizarrely, MM is the only attack usable by unarmed nonpirates.
• Although not a skill at all, the basic-attack — true to its name — is an attack. It also satisfies (b.) & (d.) (albeit the latter only for some archetypes, which is good enough).
• Attacks that’re designed to be compatible with knuckles — Flash Fist, Somersault Kick, Demolition, &c. — are usable unarmed. But they’re still attacks for otherwise typical reasons, & the fact that they can be used unarmed is a game-mechanical feature specific to the pirate archetype.
• Energy Charge appears, to someone who’s never played marauder before, very similar to the Power Guard case discussed above. However, there are several crucial differences:
  • Whereas PG itself can be activated without entering the alert state, EC cannot: to activate EC, it must be charged, which requires attacking. The fact that there’s an identifiable single attack that takes EC directly from <100% charge to 100% charge, then, means that EC’s activation per sē is the direct result of an attack. If EC were considered a nonattack, then it would be the only nonattack that cannot deal damage when used by a pacifist PC.
  • Dealing damage with EC causes the PC to enter the alert state even if the PC wasn’t damaged (i.e. it was a “MISS”). Because no damage whatever is reflected (unlike in the PG case), the PC must be taking the initiative by attacking.
  • EC’s damage is unlike PG’s in that it’s calculated completely ordinarily, like any other attack (e.g. Somersault Kick).

An automatic representation schema

With all this in mind, we can codify “job” in a way that allows us to automatically determine whether it’s “odd” in this specific sense of being “evacuated of the not-odd”.

$J$, the type of automatic “jobs”, is a product type

$$J\stackrel{\scriptscriptstyle\mathrm{def}}{=}\left(𝟙+\mathrm{\beta }+F+\mathrm{\Theta }\right)\times \left(A\to C\right)\times S\times W\times M$$

where:

$x+y$

is the coproduct of $x$ and $y$.

$𝟙$

is the unit type representing nothing in particular.

$\mathrm{\beta }$

is the unit type representing the beginner archetype.

$F$

is the type of first-grade classes.

$\mathrm{\Theta }$

is the type of throughclasses.

$A\to C$

is the type of AP constraints.

$S$

is a type, an instance of which, $s:S$, is a set of skills into which SP may be allocated (or, equivalently for our purposes: a set of skills that may be used).

$W$

is a type, an instance of which, $w:W$, is an inhabited set of weapon types.

$M$

is the ammoness of the job: $0$ if ammoless; $1$ otherwise. Shortened from munitions.

We can then codify $J$’s component parts, using doubled square brackets $⟦\cdots ⟧$ to reference game-mechanical objects.

$$\begin{array}{ccc}F& \stackrel{\scriptscriptstyle\mathrm{def}}{=}& \left\{⟦\mathrm{swordman}⟧,⟦\mathrm{magician}⟧,⟦\mathrm{bowman}⟧,⟦\mathrm{rogue}⟧,⟦{\mathrm{pirate}}_F⟧\right\}\text{.}\\ \mathrm{\Theta }& \stackrel{\scriptscriptstyle\mathrm{def}}{=}& \left\{⟦{\mathrm{fighter}}_{\mathrm{\Theta }}⟧,⟦{\mathrm{page}}_{\mathrm{\Theta }}⟧,⟦{\mathrm{spearman}}_{\mathrm{\Theta }}⟧,⟦\mathrm{fire/poison}⟧,⟦\mathrm{ice/lightning}⟧,\cdots \right\}\text{.}\end{array}$$

$A\to C$ is a type of total function, where:

$$A\stackrel{\scriptscriptstyle\mathrm{def}}{=}\left\{⟦\mathrm{STR}⟧,⟦\mathrm{DEX}⟧,⟦\mathrm{INT}⟧,⟦\mathrm{LUK}⟧,⟦\mathrm{maxHP}⟧,⟦\mathrm{maxMP}⟧\right\}$$

and

$$C\stackrel{\scriptscriptstyle\mathrm{def}}{=}\left\{𝟘,⧺,?\right\}$$

where:

$𝟘$

is statlessness, i.e. no more of the stat than absolutely necessary for the job’s archetype.

$⧺$

is statfulness, i.e. strictly more of the stat than necessary for the job’s archetype.

$?$

is unconstraint, i.e. the stat may have any value whatever.

Some possible instances of $S$ include:

$\varnothing$

The empty set. (Tacitly includes all 0^th-grade skills.)

$B$

The set of all skills that aren’t attacks. Shortened from basic-attack-only.

$U$

The universe of all skills. (Alternatively shortened from unrestricted SP.)

And finally, for weaponry, we have:

$$W\stackrel{\scriptscriptstyle\mathrm{def}}{=}𝒫\left\{⟦\mathrm{𝑢𝑛𝑎𝑟𝑚𝑒𝑑}⟧,⟦\mathrm{1H‐sword}⟧,⟦\mathrm{1H‐axe}⟧,⟦\mathrm{1H‐BW}⟧,⟦\mathrm{dagger}⟧,⟦\mathrm{wand}⟧,\cdots \right\}\setminus \left\{\varnothing \right\}$$

where $𝒫$ is the powerset operator.

Additional definitions

The primary stat function $p$ is total, because it’s defined on throughclasses:

$$p\left(\mathrm{\theta }\right):\mathrm{\Theta }\to A\stackrel{\scriptscriptstyle\mathrm{def}}{=}\{\begin{array}{cc}⟦\mathrm{STR}⟧\text{,}& if\mathrm{\theta }=⟦{\mathrm{fighter}}_{\mathrm{\Theta }}⟧\text{;}\\ ⟦\mathrm{STR}⟧\text{,}& if\mathrm{\theta }=⟦{\mathrm{page}}_{\mathrm{\Theta }}⟧\text{;}\\ ⟦\mathrm{STR}⟧\text{,}& if\mathrm{\theta }=⟦{\mathrm{spearman}}_{\mathrm{\Theta }}⟧\text{;}\\ ⟦\mathrm{INT}⟧\text{,}& if\mathrm{\theta }=⟦\mathrm{ice/lightning}⟧\text{;}\\ ⋮& ⋮\\ ⟦\mathrm{DEX}⟧\text{,}& if\mathrm{\theta }=⟦{\mathrm{gunslinger}}_{\mathrm{\Theta }}⟧\text{.}\end{array}$$

The canonical weapon function $\mathrm{\gamma }$ (shortened from gear), however, is partial, because it’s undefined for throughclasses of the mage archetype:

$$\mathrm{\gamma }\left(\mathrm{\theta }\right):\mathrm{\Theta }\rightharpoonup W\stackrel{\scriptscriptstyle\mathrm{def}}{=}\{\begin{array}{cc}\left\{⟦\mathrm{1H‐sword}⟧,⟦\mathrm{1H‐axe}⟧,⟦\mathrm{2H‐sword}⟧,⟦\mathrm{2H‐axe}⟧\right\}\text{,}& if\mathrm{\theta }=⟦{\mathrm{fighter}}_{\mathrm{\Theta }}⟧\text{;}\\ \left\{⟦\mathrm{1H‐sword}⟧,⟦\mathrm{1H‐BW}⟧,⟦\mathrm{2H‐sword}⟧,⟦\mathrm{2H‐BW}⟧\right\}\text{,}& if\mathrm{\theta }=⟦{\mathrm{page}}_{\mathrm{\Theta }}⟧\text{;}\\ \left\{⟦\mathrm{spear}⟧,⟦\mathrm{polearm}⟧\right\}\text{,}& if\mathrm{\theta }=⟦{\mathrm{spearman}}_{\mathrm{\Theta }}⟧\text{;}\\ \left\{⟦\mathrm{bow}⟧\right\}\text{,}& if\mathrm{\theta }=⟦{\mathrm{hunter}}_{\mathrm{\Theta }}⟧\text{;}\\ ⋮& ⋮\\ \left\{⟦\mathrm{knuckle}⟧\right\}\text{,}& if\mathrm{\theta }=⟦{\mathrm{brawler}}_{\mathrm{\Theta }}⟧\text{;}\\ \left\{⟦\mathrm{gun}⟧\right\}\text{,}& if\mathrm{\theta }=⟦{\mathrm{gunslinger}}_{\mathrm{\Theta }}⟧\text{;}\\ \perp \text{,}& otherwise\text{.}\end{array}$$

The formula

An automatic representation $j:J$ is automatically odd iff one or more of the following obtain:

• ${\pi }_1\left(j\right):\mathrm{\beta }$.
• ${\pi }_1\left(j\right):F$.
• $p\left({\pi }_1\left(j\right)\right)$ is well-defined, and ${\pi }_2\left(j\right)\left(p\left({\pi }_1\left(j\right)\right)\right)=𝟘$.
• $\mathrm{\gamma }\left({\pi }_1\left(j\right)\right)$ is well-defined, and $\mathrm{\gamma }\left({\pi }_1\left(j\right)\right)\cap {\pi }_4\left(j\right)=\varnothing$.
• $\mathrm{\gamma }\left({\pi }_1\left(j\right)\right)$ is well-defined, and $\mathrm{\gamma }\left({\pi }_1\left(j\right)\right)\subseteq \left\{⟦\mathrm{bow}⟧,⟦\mathrm{crossbow}⟧,⟦\mathrm{claw}⟧,⟦\mathrm{gun}⟧\right\}$, and ${\pi }_5\left(j\right)=0$.
• ${\pi }_3\left(j\right)\subseteq B$.

Prose version of the above list

• $j$ is a permabeginner.
• $j$ is a permafirst.
• $j$ has a primary stat (i.e. $j$ has a well-defined throughclass), & $j$ is entirely missing — to the extent game-mechanically permitted — that stat.
• $j$ has a set of canonical weapon types (i.e. $j$ has a well-defined throughclass), & $j$ cannot use any weapons of those canonical types.
• $j$ has a set of canonical weapon types, & all of that set’s elements are “ranged”, & $j$ is ammoless.
• $j$ hasn’t any skills that’re attacks.

Post-Cygnus

Extending this automatic representation schema to post-Cygnus classes

As given, $J$ is based on post-pirate, pre-Cygnus classes (e.g. GMS v62). Because this doesn’t encompass all of pre-BB MapleStory, the list below indicates — for completeness — how the extension to immediately-before-BB MapleStory can be effected:

Cygnus classes in general

The Cygnus throughclasses have only three grades (rather than four). They also break the easier, old definition of throughclass (as used here & elsewhere) because, although a Cygnus class is — like all pre-Cygnus classes — unspecialised until 2^nd grade, there’s only one possible class that can be advanced to if the PC is a 1^st-grade Cygnus. That is to say, once the Cygnus takes their 1^st grade advancement, their further advancement into a throughclass involves no choice.

However, putting aside these technical details, Cygnus classes behave just like pre-Cygnus classes for the purpose of automatic representation. In particular, perma-0^th-grade Cygnus & perma-1^st-grade Cygnus are both possible, & function as expected.

Noblesse

Treated identically to the beginner class.

Dawn warrior

Warrior archetype. Only two canonical weapon types: one-handed sword & two-handed sword. Tripartite stat régime identical to other warriors.

Blaze wizard

Treated identically to other mages.

Wind archer

Treated identically to bowman class → hunter throughclass.

Night walker

Treated identically to rogue class → assassin throughclass. Mildly problematic in that 1^st grade favours claws by omitting Double Stab; nonetheless, no mastery/booster/&c. until 2^nd grade.

Thunderbreaker

Treated identically to pirate class → brawler throughclass. Very mildly problematic in a similar way to night walker, because Double Shot is omitted. However, SSK is included & Improve maxHP is only present in 2^nd grade.

Dual blade

Thief archetype; indeed, its 1^st grade is the same as that of an assassin or bandit (viz. rogue). However, it enters its throughclass ten levels earlier: at level 20, it advances to 2^nd grade (viz. the blade recruit class). However, other than this ten-level discrepancy, & the fact that it has more grades than pre-Cygnus throughclasses do (6 > 4), its 2^nd grade advancement is perfectly ordinary, & results in specialisation into the dual blade throughclass.

The only canonical weapon type of the dual blade throughclass is katara. Its tripartite stat régime is thus determined by the katara; I’m not exactly sure what it is, but it’s probably the same as for assassin (= claw) & bandit (= dagger wielded by a rogue). Kataras don’t use ammo.

Aran

0^th grade (Aran beginner or legend) is treated identically to the beginner class.

Perma-1^st-grade Aran is extremely problematic, mostly because it already has access to Polearm Booster. However, it doesn’t get access to Polearm Mastery until 2^nd grade, so we can technically argue it either way. In honesty, I would conservatively consider the Aran to have entered its throughclass already at 1^st grade.

Warrior archetype. Only canonical weapon type is polearm. Tripartite stat régime identical to other warriors.

Evan

0^th grade (beginner master, Evan beginner, or legend) is treated identically to the beginner class.

Perma-1^st-grade Evan is possible. Magic Missile is identical to Magic Claw in every way, except visually.

Technically, Evan has a whopping ten(!) grades. Nonetheless, it specialises as expected: 2^nd grade is when it gets its first access to an elemental attack, & its first access to Teleport.

Treated identically to other mages.

Induction

We typically think of certain jobs in ways similar to the $J$ representation above — which exemplifies automatic representation — but without actually involving that automatic structure in the definitions of said jobs:

Location-induced jobs

Campers & islanders are — conventionally — defined entirely by their location. These locations induce permabeginnerhood, & the latter is what actually makes them odd. Yet this permabeginnerhood isn’t included in the definition per sē — it’s part of the extension, & not of the intension.

Weapon-induced jobs

Similarly, a weapon like e.g. the Scythe has requirements that induce oddness in its wielder. In this case, it’s a one-handed axe equippable only by thieves; the thief archetype has no classes with one-handed axe as a canonical weapon type, so the dedicated Scythe-user is what we fancifully call the grim reaper odd job. But we define the grim reaper solely by its weapon; the archetype restriction isn’t part of the definition (& technically, neither is the weapon type restriction).

The $J$ schema cannot capture what makes induced jobs special, even though consulting it with a concrete (read: extensional) example of a PC with an induced odd job yields the answer that we expect: it’s odd.

Oddness vs. jobness

Indeed, an “automatic job” (in the sense of $J$ or similar) is not really a job — hence the use of the word representation instead of job. We could make $J$ & its formula increasingly complex in an attempt to more & more closely capture what we think a job can be (or what we think a job definition can be),^[6] but this is essentially an abuse of the automatic representation: it captures oddness, & not jobness.

This turns out to be an important distinction here, & is a distinction that I don’t consistently make elsewhere. There are two sides to the automatic representation’s inability to capture jobness:

• Given a job, it can be translated — perhaps lossily — into $J$ (coërced, as it were), which allows us to use the automatic formula to determine whether it’s automatically odd.^[7] But the reverse operation — translating some instance of $J$ into a flesh-&-bone build/job — simply isn’t meaningful. We know perfectly well that we had to mangle the job to fit it into $J$ in the first place: the nuänces of induction, the aspects of the job not strictly relevant to its oddness, &c. were all lost.
• An automatic representation is entirely bereft of insight into what makes a job useful or worth defining. This is most straightforwardly evinced in two ways:
  • The representation cannot distinguish between a quintessential odd job (e.g. the dagger warrior) & other odd “jobs” that aren’t quite comparable:
    • A pure-INT brawler^[14] is clearly automatically odd. But… why?
    • Is there really that much of a difference between a maxMPless STRginner & other STRginners? Because $J$ no doubt encodes them distinctly.
    • A gun warrior is automatically odd, seeing as $J$ is contentedly oblivious to the fact that such a job isn’t even playable.^[8]
    • ⋮
  • An automatic representation has no need to encode things in ways that can represent useful, general definitions. To give just one simple example: $J$ cannot encode the blood warrior. Yes, all blood warriors are automatically odd according to $J$’s formula — but only because one of the following must obtain, & any one of them will result in automatic oddness:
    • The blood warrior is a permafirst.
    • The blood warrior is of the fighter throughclass, which has STR (which is importantly not the same as maxHP) as its primary stat.
    • The blood warrior is of the page throughclass, which has STR as its primary stat.
    • The blood warrior is of the spearman throughclass, which has STR as its primary stat.

On the other hand, an automatic representation is good at what it does: if you want to know whether… something is odd, & you specifically want this very particular automatic version of oddness, then the automatic representation won’t hesitate to spit out a binary “odd” or “not odd” answer.

Jobness

With that, we ought to clarify the notion of “job” to the extent necessary to disentangle what we know so far (from this appendix, as well as On odd jobs proper). In ascending order of abstraction:

Build

The particular way in which a given PC is — bodily, & not on hypothesis — built in-game, potentially including its destiny (insofar as the latter may or may not be realised).

Within certain limits, what constitutes the PC’s “build” is conventional. Nonetheless, we typically observe regular habits of equipment (including, but not necessarily limited to, weapon types, particular weapons, shield use, clothing habits, scrolling habits, &c.), class (but not so much abstractions of class like throughclass, archetype, &c.), AP & SP build (again, largely unabstracted from raw numbers), precinction (i.e. fixed habits of location), &c..

Job

An abstraction of build that’s general enough to be nameable, & also durable enough to allow for reasonable judgement of whether a given build conforms to the job in question.

Paradigmatic job

A job that’s distinguished for being paradigmatic (“exemplary”) & atomic (“cardinal”). As always, paradigmaticity is relative to the paradigm in question.

Nonetheless, in general,^[9] the paradigmatic job must be defined in a way which distils its exemplary position within the paradigm down to a single defining feature, or a single cluster of defining features that’re mutually interdependent in the sense that all must be present in order to produce the exemplary effect. This fulfils the atomicity requirement.

Automatic representation

A mathematical abstraction of a job (or of a fragment of a job, or of multiple jobs), encoded in purely logical terms, with semantics corresponding to game-mechanical entities (i.e. ludemes sēnsū strictō) that’re salient to the paradigm.

Revisiting the old definition of classicality

We’re now equipped to unpack what’s really going on in the old definition of “classically-odd job”, as quoted in On odd jobs:

• Defining a particular odd job (e.g. permawarrior) is simple.

Already, this definition is pointing specifically at paradigmatic jobness. If defining a job is simple to do, & results in a simple definition, then we can be reasonably confident that it at least satisfies the “atomicity” requirement given above.

• Each particular odd job satisfies some intuitive notion of natural.

This fulfils the “paradigmatic; exemplary” requirement of a paradigmatic job. Many jobs can be defined simply, but not every such job is worth canonising.

This is where it becomes particularly obvious that we need to appeal not just to our intuition (“intuitive notions”), but also to the nuänces & crannies of the game’s ludemic structure. It’s difficult to make generalisations about what “natural”, “paradigmatic”, &c. might mean when the game itself is deeply — if lovably — uneven.

A simplistic, automatic example is that $J$ sees no qualitative difference between a pure-DEX brawler & a pure-INT brawler,^[14] & neither does our atomicity principle!

A more subtle example is the question of how — or whether — the LUKless assassin should be refined. The STR assassin historically emerged as the “natural” result of meleeifying the assassin throughclass, & finds kin in other “STRangers” like the woodsman & the swashbuckler, as well as in the similarly related brigand — the latter of which is overwhelmingly played STRfully (or may even be defined as such). Moreover, STR increases damage with claws at least as much as any other non-LUK stat does. For these reasons (& potentially others), the STR assassin has usually been conceived as primary, with the LUKless assassin left unnamed, or made synonymous with STR assassin, or considered as a purely theoretic generalisation. Yet the shadowy sibling (the shadow partner, if you will…) of the STR assassin — the DEXassin — enjoys the same damage output with claws,^[10] whilst enjoying superior AVOID & WAcc. Moreover, other odd jobs often get analogous “lesser odd-statted” versions as well: our famous STRginner sits alongside the LUKginner^[11] & DEXginner, the DEX warrior alongside the LUK warrior, & so on.

• Insofar as an odd job is “odd”, it is pure in its “odd” aspect.

This condition is designed to rule out odd jobs that’re hybridised with nonodd jobs (or however you want to think of it), so that the oddness of the job is substantially sacrificed. It’s clear that satisfying this condition is necessary for the job to be paradigmatic. However, it also appears to hint at automatic representation territory.

That’s because this is the first condition considered so far that actually has even the slightest thing to do with oddness per sē. The simplicity & naturalness conditions essentially just enforce the notion that a genuine paradigm is at stake. But a paradigm is just a paradigm, & to show that we mean real odd business, we insist that there’s a qualitative binarity: if a job is to be “classically odd”, then it needs to fully separate itself from other (“classically nonodd”) jobs via its purity. We can imagine paradigms under which purity is irrelevant or inapplicable; as a simplistic example, the naïvest (& commonest) conception of job conflates job with throughclass, & thus has no notion of “purity”. Therefore, this is finally something distinctive — if perhaps not unique — about classical oddness in particular.

But we’ve still yet to really articulate oddness. The tacit assumption here is that the reader already has a nebulous idea of what oddness is, & this list of conditions is only here to refine that idea into a paradigm capable & worthy of entering into discussion.

• Odd jobs are, to the extent possible, atemporal.

This is the most pointed item on the list, as it’s so characteristic of (or perhaps even unique to) classical oddness.

Nitpick about the poor wording

The formulation quoted above is, as written, incorrect (as pointed out in On odd jobs & much earlier), because it states that the job itself is what needs to be atemporal. In reality, it’s not the job, but the odd aspects of that job (i.e. the ludemic relations that actually make it odd) that’re atemporal. For instance, although the pirate archetype was released relatively late in the game’s history, the swashbuckler is just as odd as the woodsman is, because their oddnesses (not their jobs) are identical — & therefore equally atemporal.

Presumably, the intended effect is to exclude perma-2^nd-grade, perma-3^rd-grade, perma-4^th-grade, & perma-5^th-grade jobs from being classically odd. This condition actually turns out not to be strictly necessary — at least, not for this purpose — because there are even stronger additional reasons for excluding these jobs from being odd; see: “Late grade-advancement restriction”. Nonetheless, the atemporality condition is indeed characteristic of classical oddness, & it helps us narrow our conceptual scope.

Overcoming & reïncorporating history

In its general survey, On odd jobs indulges in some clarification of things that’re commonly misconstrued as odd. In the process of excluding odd–nonodd hybrids from classical oddness, we find ourselves in need of explicit clarification that this doesn’t apply to odd–odd hybrids (so long as at least one job that enters into the hybrid remains intact — perhaps you get the idea):

Note that this generally does not apply to hybrids of two or more jobs, all of which are classically odd […]. Although the resulting hybrid job no longer satisfies the “simplicity” requirement (& arguably not the “naturalness” one either), we nevertheless consider it odd on the simple logic that mixing odd with odd could only produce more odd.

This simple logic is sound only because the paradigmatically classical (that is, both classical & not hybrid) odd jobs operate like cardinal directions or atoms. […]

But the quoted approach is specifically an historic one. In the primordial soup, there’s not yet any possibility of an odd–nonodd distinction; how such a distinction can arise, what its implications are, &c., are part of the focus of On odd jobs. What’s relevant here is that this formation of “oddness” from specific exemplars — treated here metaphorically as atoms or cardinal directions — isn’t entirely necessary from a mature standpoint.

All language is formed through usage & example. As a language matures, however, it becomes powerful enough to serve its own purposes; a dictionary’s definition of a word, for example, is only possible to the extent (admittedly, a deeply limited one) that a language can describe itself. To this extent — again, with very palpable limits — the paradigm can be more than the sum of its exemplars.

If we trust our own judgement — well-informed as it is by history, gameplay, & game-mechanical knowledge — then we can identify paradigmatic jobs. If we can put together our gradually-congealed understanding of what even unites these exemplars under a unified paradigm, then we can rationalise the paradigm to the point of making automatic representation possible. Oddness (in the desired sense) can be defined rigorously. A reversal occurs: the exemplars no longer collectively germinate the paradigm — the paradigm holds aloft the exemplars.

Oddness

We’re now equipped to give a set of definitions analogous to those in the “Jobness” section above — but this time for oddness, rather than for jobness:

Odd sēnsū lātō

Not well-defined (yet…‽);^[12] includes builds & jobs that’re “odd” in their closeness to more strictly odd jobs, but that fail to pass muster by any stronger definition. For example: a hybrid STR–LUK warrior. (This category also includes, as a subset, all jobs that’re more strictly odd.)

Traditionally odd

Oddness by virtue of conforming to a job traditionally (not to be confused with “historically”) understood to be odd. We scarcely question the oddness of these jobs, because they’re handed down to us directly as part of a canon.

Unsurprisingly, the traditionally odd jobs tend to overlap considerably with the paradigmatically odd jobs. However, this is purely an extensional overlap, & not an intensional one — hence why paradigmatic job is defined in the above list, rather than in this one. We’ll later see how it’s possible for the two to diverge; but for now, it’s important to simply note that a job can be traditionally odd without a paradigm for it to exist within — whereas this is clearly untrue for a paradigmatic job of any sort.

Classically odd

A rationalisation of traditional oddness that embeds the traditional canon into a paradigm that can stand on its own two feet.

Because — as its name implies — classical (perhaps neoclassical would’ve been more suitable) oddness is based in traditional oddness, it cannot, in general,^[9] be fully rationalised.

Automatically odd

A build or job is automatically odd according to some automatic representation schema ${\rm A}$ (the example given in this essay is called $J$) iff coërcing it into ${\rm A}$ results in ${\mathrm{\alpha }}_1,{\mathrm{\alpha }}_2,\cdots ,{\mathrm{\alpha }}_n:{\rm A}$, s.t. every ${\mathrm{\alpha }}_i$ (i.e. all $n$ of them, where $n$ is a positive integer)^[13] is odd according to the schema’s formula.

The key phrase here is “according to”. We must first invent an automatic representation schema. Once we do, its results are automatic (duh) — but not unique, inasmuch as a different representation schema might produce a different answer in some cases.

One good reason for including this as an appendix (as opposed to part of the main text) is that the above list of definitions looks a bit like some kind of miserable tier list. Needless to say, that would be horseshit, & I think that I make a pretty strong case in On odd jobs that an odd–nonodd distinction — really, any one whatsoever — should not affect players’ character-build choices at all — much less should it confer any sort of prestige.

More to the point, thinking of the above list as a tier list — or really, any kind of meaningfully-ordered list — is simply categorically confused. There’s no pair — out of all six possible pairs — of items in the above list where both items in the pair exist at the same conceptual level.

In any case, I’m now reasonably confident — excepting the occasional little snag here or there — in the current version of the list of odd jobs on the Oddjobs website. Specifically, in its ability to present a broad array of classically odd jobs in a way that’s coherent, & just about as rationalised as we can reasonably hope to get. Is it the ✨⁠perfected form⁠✨ of classical oddness? Presumably not. But I don’t expect to see anything considerably superior any time soon…

Case studies

And, speaking of rationalisation, it’s instructive to give examples where traditional oddness diverges from classical oddness and/or from automatic oddness; that is, concrete instances of mismatch between paradigm & tradition.

The gish

Even the relatively meagre available historical material is enough to establish the gish as traditionally odd. In particular, it’s so intimately tied up in the entire notion of odd-statted mages that I had to fish around the WWW for a goofy name like gish just to differentiate them from STR mages.

But there’s something very strange going on with mages: they can use magic! Essentially everything about magical attacks is either game-mechanically unique, or mirrors physical-attacking mechanics in ways that’re incompatible with that mirror image. This puts the mage archetype in a uniquely difficult position: whereas archers, thieves, & pirates can, broadly speaking, pick up new weapon types in ways that allow for hybrid combat styles (woodsman, STR assassin, DEX brawler, swashbuckler, &c.), the mage analogue apparently has no plausible way in which to allocate its AP.

There’s no way around the need for INT+LUK (considered as a single monolithic stat): there aren’t any other sources of MAcc, & without MAcc, you don’t really have magic. Moreover, INT is the only efficiently-scalable source of TMA, & indeed, the only one of the six stats into which AP is allocable that increases magic damage output (considered separately from accuracy).^[15]

But if the gish is defined by no more than a combination of AP allocation & archetype, then doesn’t adding INT kinda just make it an ordinary mage?

The old version of the odd job list placed a “STR + DEX > INT + LUK” constraint on the “gish” entry, in an attempt to make clear that a gish’s AP build is really distinct from that of a nonodd mage. But this inequality is entirely ad hoc: it bears no special relation to the job, nor to the paradigm that it inhabits.

The new version of the list fully rationalises the definition of the gish’s AP constraints, so that it’s expressed in the very same language as every other job’s AP constraints.^[16] The new definition is certainly accurate, but it places very little actual constraint on a hypothetical gish’s build in general.

The fundamental problem here is that there’s no way to make the constraints more narrow (precise, if you will) without breaking the level of abstraction at which “gish” exists: the level of the job (in this case, job under the paradigm of classical oddness). Once we involve exact AP distribution numbers at particular character levels, details of the game-content in a particular MapleStory implementation, itemisation at the level of e.g. “Bathrobe with 20⁢ INT”, the social nature of the PC’s interactions with other PCs, &c., &c., in an attempt to delineate the “true gish”, we’ve already left far behind the land of abstraction above the level of character build — or perhaps even abstraction to the level of build.

Putting aside the gishlet, there is, in truth, no way to make the gish automatically odd. It separates itself from the nonodd mage via AP allocation — weapon type is no help here, thanks to magical attacks’ suī generis relationship to weaponry — & yet may, in principle, have an arbitrary amount of base INT.

The result can be stated reasonably simply, but only thanks to our conceptual framework developed so far: the gish is traditionally odd, & yet cannot be automatically odd. It can no doubt be classically odd — but only insofar as the classical paradigm incompletely rationalises its mages. It cannot be a paradigmatic job: it’s nonatomic, & not really exemplary either, because it presents a barrier to rationalisation.

“Salvaging” the gish?

In no way should this be read as condemnatory. I stand ready to vigorously defend the traditional oddness of the gish, in addition to its oddness sēnsū lātō, & even its classical oddness (and, of course, its 🆒ness 😎). There is thus simply no need to “salvage” the gish as an odd job.

Nonetheless, it’s interesting to consider a nonclassical oddness paradigm that explicitly includes (amongst other things, of course) every job that makes use of at least one of its throughclass’s^[17] tertiary stats as a primary (or secondary…?) stat for at least one of its weapons. The phrasing “makes use of […] its weapons” is meant to imply that the weapons in question are actually meaningfully used to perform physical attacks; at the same time, this is “weapons” plural, where some of those weapons might not have any particular relationship to stats.

In the case of the gish, both STR & DEX are tertiary for mages, yet also contribute to damage dealt by melee weaponry. So long as the gish uses one or both of those stats to power meaningful physical damage output, it’s odd in this sense.

In the case of warriors, archers, & pirates, claws offer a way to enter this niche: LUK is tertiary for the relevant throughclasses, but is the primary stat for claws. Like the gish, they’d be free to allocate — within certain limits — AP into their primary stats. Under the definition given above, they’d also be able to switch weapons to make use of those primary stats in the usual ways.

Thieves don’t get to join in on the fun, because all of their tertiary stats are physically worthless. 🤷🏽‍♀️

The claw bandit

With the gish, we have traditional oddness, but not automatic oddness. With the claw bandit (= sindit sēnsū strictō), however, we have the reverse!

We must first contrast the strict & eternal sindit with the sindit sēnsū lātō; the latter only opportunistically uses claws during the early (roughly ≈20) 2^nd-grade levels.

Because the sindit apparently encompasses little more than the bog-standard way of playing bandits at the most crucial levels of specialisation, it hasn’t traditionally been seen as odd. Moreover, most PCs don’t make it beyond (or don’t make it far beyond) these levels, thus making the sindit playstyle appear even more perfectly ordinary.

Even in cases where PCs do achieve arbitrarily high levels, the specialisation argument seemingly still applies: for game-mechanical reasons, both traditional & classical oddness hinge many of their judgements on the crucial transition from pre-2^nd-grade into a determinate throughclass (again, see: “Late grade-advancement restriction”). The sindit sēnsū lātō palpably exists right across this exact boundary — otherwise it wouldn’t be called “sindit”. If the sindit sēnsū lātō isn’t odd by traditional standards, then, the traditional Mapler has no way to tell the difference between this nonodd build & a hypothetical purified version (i.e. our sindit sēnsū strictō).

However, this cutoff point for specialisation exists at the game-mechanical level, thus making it a category error to conflate this level of abstraction with the level of empirically-observed gameplay at which the sindit sēnsū lātō exists. The result is that an automatic representation schema (e.g. $J$) has no problem whatever determining that the sindit sēnsū strictō is odd. Moreover, classical oddness is more than free to rationalise the traditional conception, so that the claw bandit & the dagger assassin are equally classically odd, just as much as any other classically odd job.

There’s no contradiction in the fact that the sindit style is opportunistically played by those with otherwise no interest in odd jobs. This very phenomenon occurs with many odd jobs, because such a great number of them are so powerful in their own right (indeed, this is part of the thesis of On odd jobs).

This isn’t an attack on the traditional conception. We respect the traditionally odd canon as it is handed down to us, & preserve it in its place — for its historic value, if nothing else.

The islander

The islander (& camper, which isn’t meaningfully different for this purpose) is a job whose oddness is induced by location; as a result, an automatic representation schema like $J$ cannot capture its nuänces. See: “Induction” above. Nonetheless, the islander is — at the very least, extensionally — odd by all accounts, including any reasonable construal of the four items defined in “Oddness” above.

But one thing that makes the islander relevant in this context is that it’s traditionally odd without being a paradigmatic job under the classical paradigm. Inasmuch as a classical paradigm is incompletely rationalised, it still treats the islander separately from other permabeginners, as a kind of holdover from the traditional conception.^[18] Indeed, the islander is viewed somewhat differently by each conception of oddness:

• The traditional conception sees the islander as self-evidently odd. In addition to being a permabeginner, it’s largely causally isolated from the rest of the game, and forms a strong & enduring social niche.
• The classical conception, as always, inherits tradition; but to rationalise this inheritance without sublimating it requires adding “location” as a dimension of jobness, which in turn adds complication to the paradigm. Thus, the islander quā job cannot be any more than an annex — far from being a paradigmatic job.
• The automatic conception has no choice but to completely rationalise what the classical conception did not. Lacking a notion of location implies that the islander quā job cannot be automatically represented. Nonetheless, the islander quā build can be coërced into an automatic representation like any other build, & it’s automatically odd in this sense.

Perhaps unsurprisingly, this state of affairs is the cause of some confusion. See: “Islanders & campers”.

Summary

It’s a bit unfortunate how many words that took to articulate coherently, but I suppose I shouldn’t be surprised. If, for some reason, you actually read this appendix, email me or something about how you learnt a lot & all of it was useless. Thank. 🧡

If you’ve been following along so far, then this diagram will, hopefully, help you review the basic concepts. It’s not a model for anything real, but more like the world’s worst flowchart. Y—you’re welcome.

	description	+odd­ness	+job­ness
I.	The primordial soup. There are PCs, & they have builds in the barest sense, but those builds aren’t yet concieved as such.
II.	Players begin to consciously shape their builds strategically; builds are now concieved as such. The individual choices made in each build are — at least sometimes — purposeful, but never quite categorical.		build
III.	The community has collectively crystallised a largely unsystematic array of identifiable, nameable build strategies: the notion of job can now be meaningful. This array gradually evolves over time, & in the process, begins to shake out what we now know as traditional odd jobs.	pro­to-tra­di­tion­al	job
IV.	As traditional odd jobs become reasonably well-established, a paradigm begins to form around them. Some odd jobs are virtually set in stone, which allows newer odd jobs to be innovated merely by analogy with other odd jobs.		pro­to-para­digm
V.	Analogic reasoning by players of odd-jobbed PCs, in combination with the now extreme ossification of conventional character-build wisdom, precipitates odd-jobdom as a distinct concept. A model of classical oddness can now be developed.	tra­di­tion­al, pro­to-clas­si­cal	para­digm
VI.	With the progress of classical rationalisation, odd jobs that’re traditionally odd & that fit particularly easily into the rationalised paradigm become exemplary — i.e. paradigmatic jobs.	clas­si­cal	para­digm­atic job
VII.	With the maturation of the classical model, the only way to further rationalise and/or ossify oddness is to do what the classical model cannot. An automatic representation schema makes possible the completed rationalisation of oddness, but in exchange, has an austere ontology that makes it inferior to the classical model for most practical purposes.	au­to­mat­ic	au­to­mat­ic re­pre­sent­a­tion

---

Footnotes for “Appendix A: Classicality in depth”

1. [↑] The archetypes are: beginner, warrior, mage, archer, thief, pirate. That is, an archetype is either a 1^st-grade class along with every class that it can possibly advance to, or else the set of all 0^th-grade classes (including noblesse &c.). The notion of archetype is encoded into — but not named by — the game: every equipment item defines which of the six archetypes may equip it (an equipment item that allows all six is conventionally said to be “common”).
2. [↑] For more info, see: Heal.
3. [↑] Typically, an SPless PC still spends its 0^th-grade SP. Not only is this SP clearly as unspecialised as it gets, but the game itself explicitly isolates 0^th-grade SP from all other SP in a way that it doesn’t for any other grades.
4. [↑] The alert state is that condition whereby certain actions — equipping items, dropping items, changing channels, &c. — are prohibited. Attempting these prohibited actions only results in an error message like: “You can’t do it right now. Catch your breath and then try again.” — hence the use of catching one’s breath to refer to alert.
5. [↑] To last-hit a monster is to cause its HP to go from being positive to being nonpositive, thus slaying the monster once & for all. Several sources of damage exist that cannot last-hit monsters; in such a case, the damage can, at best, reduce monsters to 1⁢ HP.
6. [↑] The astute reader may’ve noticed that the contrast between $⧺$ and $?$ in the definition of $J$ is actually entirely superfluous to the automatic formula given in “The formula” above. I leave the contrast there largely to keep $J$ from being too conspicuously artificial & purpose-built. Note that we can recover the concept of pure $a$ (for a stat $a$, e.g. DEX) by simply assigning $𝟘$ to all other stats (or to all other stats excepting maxHP & maxMP, if you prefer).
7. [↑] Strictly speaking, some jobs (depending on your definition) might require multiple representations as various instances of $J$. In such a case, for the job to be automatically odd, the automatic formula must obtain for all representations. This fact also reäppears later within this appendix.
8. [↑] It would be sooper cool, tho…
9. [↑] In the mathematical sense.
10. [↑] That is, debuffs like venom & Ninja Ambush notwithstanding. Plus, neither of the aforementioned are claw-specific, nor even weapon-restricted at all. We should also note that both are 4^th-grade skills, & thus do not generally exist.
11. [↑] There’s another “fun” little conundrum: is it a besinner, defined by its exclusive use of claws? Or a LUKginner, defined by its pure (or nearly pure?) LUK? Both? Or perhaps they’re two distinct jobs…?
12. [↑] I mean, I can’t be the only one out there writing dense, technical essays about MapleStory jobs, r—right…? Try it yourself! I wanna read it!! (😭)
13. [↑] Having multiple $\mathrm{\alpha }$s might seem unnecessary — & it should be, in most cases. Nonetheless, consider the example given earlier of coërcing the blood warrior into $J$. We need four representations: one for the permafirst, & three for the throughclasses. It’s a bit annoying, but noöne’s gonna actually use $J$ anyway, so I’m not worried about it.
14. [↑] As always, HP/MP washing is totally irrelevant here. Not only is this particular quirk functionally or really nonexistent in most versions of the game, but in cases where its presence is felt, the INTful physical-attacker in no way includes INT as part of their build; rather, the allocation of AP into INT is fraudulent, & intended to be reset out later.
15. [↑] Excepting Heal, of course.^[2]
16. [↑] Viz. various individual stat constraints — each one being one of {Sless, Sful, pure S}, for some stat S — in something like conjunctive normal form. But the astute reader might notice that “gish” & “gishlet” are the only two entries that actually make use of disjunction. ({Sless, Sful, pure S} is also unnecessarily expressive, but makes it easier to read, I hope.)
17. [↑] Or — where the generalisation is warranted — class’s.
18. [↑] This is the approach taken by the Oddjobs website — albeit armed with the notion of subjob, so that islander ⊂ permabeginner.

---

Appendix B: The exoteric & the esoteric

One of the misconceptions refuted by On odd jobs is that “certain odd jobs are just weaker versions of their nonodd counterparts”. Although a few other sections of the essay are also relevant, this one in particular cuts to the core of the tension between the exoteric & the esoteric.

In MapleStory, some things are publicly observable (or simply public) to relevant players. If a monster is in a map, then anyone in the map (putting aside visibility concerns in large maps) observes the monster in morally the same way. If a PC attacks the monster, then the result — damage lines, movements, status effects,^[1] &c. — are similarly publicly observable. By double-clicking a PC, its equipment becomes publicly observable to some extent; but many relevant details remain publicly unobservable: stats, scroll pass count, nametags, expiry dates, &c..

The exoteric is always public, but most public observables are typically not exoteric. The exoteric only includes things that’re customarily observed.

Things that aren’t public are private. For our purposes, this includes subjective things; the latter aren’t immanent to the game itself — that is, they aren’t game-physical — but they do supervene, at least in part, on game-physical observables (public & private).

The esoteric is the aggregate of all subjectivities & private observables that would be consciously available to the player of a PC if the player were, hypothetically, to have an idealised expertise in playing said PC.

The esoteric in practice

If you’ve read my diary before, then you’re familiar with the passages that go into some considerable detail about my experiences playing one or another of my characters, at the level of gameplay — perhaps in a particular situation. As just one example, consider: “DK rev. 3”.

Needless to say, not every Mapler has the time + tolerance for writing absolutely too much (& learning how to write along the way…) — the combination necessary to pen these sorts of things in detail. But the point is that it’s entirely possible, because otherwise, I wouldn’t’ve gone & done it on multiple occasions.

This simple possibility alone is enough to demonstrate that there’s a lot going on at the esoteric level (including things that’re difficult or impossible to even write about at all), to the point that exoterica are totally eclipsed — at the very least, in sheer magnitude.

Subjection?

Since the crux of the argument that I’m in the process of making here is clearly that the importance of the esoteric ought to be emphasised, the natural objection is that this places unwarranted emphasis on subjectivity at the expense of both sociality & game-physicality.

My response to this is largely in three prongs:

• What is (or will be) at stake here is character build/job. In On odd jobs, I use the term egocentric to describe the basic nature of the act of building one’s own MapleStory character.

  Furthermore, to the extent that a character’s build is not determined egocentrically, the build is significantly underdetermined by the relevant social factors, inasmuch as those factors are driven entirely by exoterica. This doesn’t necessarily imply that social factors can’t determine whether the build will turn out to be odd,^[2] but it does imply that esoterica still have a strong role to play, even in a particularly allocentric case.

• In no way does this approach actually prioritise subjectivity over game-physicality; indeed, quite the opposite in some ways! This is made particularly obvious by, for instance, the example given in “The esoteric in practice” above. Even in principle, it should be self-evident that exoterica provide meagre game-physical insight in comparison to esoterica.

• On a more philosophic level, the bridge to sociality is intersubjectivity, which in this context is — insofar as it’s mediated by game-physics, rather than by ordinary communication methods — something like approximate perception of esoterica via exoterica. Still, for the purpose of this appendix, this is more of a technical point than an important argument.

But why would you do that?

The myth of “certain odd jobs are just weaker versions of their nonodd counterparts” is one particular case of a more general sentiment: “but why would you do that?”.

The honest answer to this (usually) honest question obviously depends on whom the interrogatee is. Components of the answer typically include things like:

• Because so many odd jobs are so powerful in their own right (that is, in their own ways), we can point to game-mechanical advantages, which directly supervene on game-physics.

• A familiar trope is the idea that certain character-builds^[3] are done for the “challenge”, or as a “challenge run”.

  Insofar as the “challenge” is challenging for the player & her character — that is, to the extent that the player is the one who’s challenged, & that she elects this challenge for its own sake — this is an instance of prioritising esoterica. However, we should hesitate to attribute justification to the “challenge run” trope, for basically three reasons:

  • It appears attractively obvious what “challenge” is supposed to mean here: much like some videogames offer multiple difficulty levels to the player, so here does the Mapler elect her “difficulty level”. But there’s such a vast variety of things that can be done to plausibly “increase difficulty” in MapleStory that even the most ardent “challenge run” lover (i.e. a given individual) can scarcely be expected to enjoy even half of them. We thus find that the “challenge run” trope fails to actually explain anything, and we’re forced to drop to the level of the particular “challenge” & what makes its gameplay enjoyable.
  • As explored in On odd jobs, MapleStory has no win condition. Therefore, we need to do at least a little more work to explain what “challenging” someone might mean, if that someone isn’t necessarily trying to do anything in particular. For a “challenge run” that genuinely is a run in the sense that it provides its own win condition, this is, of course, an irrelevant worry; but here, we’re concerned with build/job.
  • Not everything is a “challenge run”, even putting aside what that might really mean. Like any other trope, the “challenge run” is widely recognised (if perhaps not always sympathetically), & so we naturally find ourselves trotting out the trope merely for lack of consideration or creativity.

• Because it’s 🆒. It’s got that swag factor. More broadly, the player gets to specially distinguish her character from other PCs. Her character is more unique — even if that uniqueness is, as it happens, not actually 🆒, & fails to attain the relevant social recognition. The “meme build”^[4] trope also falls under this category.

Taking into account the dubiousness of the second item, the above three items, in any combination, are totally valid reasons & ways to understand why one’s PC has been made how it has. Nonetheless, the list is missing something, & we can’t adequately (or perhaps even partially) explain certain phenomena with it.

But there are also unknown unknowns

Very simply: we play characters because they’re fun to play. This is so obvious that it hardly seems to warrant rigorous investigation!

But esoterica are nonetheless largely ignored, to the point that a statement like “[insert odd job here]^[2] is just a weaker version of [insert another job here]” appears to make sense, simply because the PC instantiating the job in question exoterically appears, under particular circumstances, largely indistinguishable from the latter job. For instance, a gish(let) can, by its very nature, easily be mistaken for a “weaker version” of an INT mage or of a STRginner. But anyone with any experience with gishes can tell you how deeply mistaken this is at the esoteric level: the gish unlocks, nearly from whole cloth, a new mode of gameplay.

Worse yet, like so many misconceptions, this can extend even into the realm of conjecture: mere consideration of a job can seem uninteresting or trivial, in spite of the considerer having none of the actual experience with the relevant esoterica upon which such a judgement could even be based. And I’ve fallen into this trap myself!

Believe it or not, it gets worse yet. In some cases, even a player with personal experience playing the build/job still lacks the expected esoteric insights, so that she fails to fully differentiate even the build of her own PC from that of other wildly different PCs. For example, one of the various palpably core aspects of playing a DEX warrior (speaking as someone with the relevant experience) is the avoidability advantage over STR warriors (& various other jobs). Yet this was news to another fairly experienced DEX warrior player with whom I spoke one day.

We’re forced to accept that we ought to be humble in these matters: if you’ve no knowledge of something, that in no way implies that there’s no knowledge to be had; & this is doubly true for esoterica, by their very nature. Every new build teems with esoteric possibilities that may or may not be actual — but the only way to know is to make them actual, by rigorous investigation & by amassed experience.

Physics–content duality

In point of fact, appreciation of esoterica is required by any approach to gameplay that respects the game itself.

At its core, the game partially presents itself game-physically; it has rules, & these rules are expected to be universal to the game, regardless of any particular game-content. The game-content, then, forms the dual half of the game’s core.

But in any game of sufficient complexity — such as MapleStory, for example — the abstract nature of the game-physics doesn’t make the game’s possibilities immediate to the player (nor to any other observer). This is because said possibilities emerge through gameplay.^[5] To give another example: I know the rules of chess. But I don’t have much experience playing it, & I’m rubbish at it. A grandmaster has vastly more knowledge & understanding of what chess is actually like than I do, even if our knowledges of the rules are technically equivalent: she’s played positions that I could scarcely dream of, & has far more understanding of chess strategy than I do. (All of this is prior to the additional fact that she would kick my arse at chess.)

MapleStory is, however, not chess. The rules of MapleStory are brought out — made interactive, made relevant — by the game-content. A game-mechanic that’s prominent in one implementation may be forgotten in another, simply due to differences in game-content, in spite of technically identical game-physics.

But this broad sense of game-content — as opposed to the narrow sense of “structured game-activities designed to give players something to do” — includes all particular instantiations of any game-objects that conform to the game-physics. The game-content is what there is, & the game-physics is merely its integument. One of the most important aspects of game-content, then, is the PCs themselves: what they are, & what they do.

What the PCs are — i.e. their builds — is so decisive that, if we ignore this element, we can no longer make sense of the game-physics. More to the point, in order to appreciate the game-physics, we need to know what that game-physics is actually like. Although esoterica aren’t our only window into the game-physics, they form by far the most insightful one; without access to wide-ranging esoterica, we simply cannot penetrate into the game’s inner logic, & can only content ourselves to play the game unconsciously — or not at all. And I know this from personal experience, above all else!

Why, indeed

With all this in mind, we can launch a soft “critique” of the three items given in “But why would you do that?” above:

• The fact that a particular odd job — or indeed, any job whatever — is so powerful in its own right is undoubtedly important, but only of secondary importance. Of primary importance are the game-physical ways in which the job empowers its PC, & the specific ways in which that empowerment affects gameplay at an esoteric level.

  This perspective allows us to see that, even in cases where an odd job doesn’t appear to have any particularly shining “powers”, the job can still be “powerful in its own right”, in a slightly different sense: it traces out a unique path through the game, in a way that both highlights the game’s structure in an interesting way, & is productive of new gameplay experiences (again, at the esoteric level).

• We can now account for the things that the “challenge run” trope couldn’t explain on its own. Whereas the notion of “challenge run” provides no insight into why it makes gameplay enjoyable, the esoteric provides the necessary conceptual link: a particular “challenge run” allows us to trace a novel path through the game’s structure, & thus experience new gameplays — new esoterica — that we would otherwise never experience.

• Anyone can do something a little 🆒. But if that “something” is within the realm of build/job, then it’s a commitment: in MapleStory, a PC is not really something that one “does” — it’s something that one plays as. If we can’t explain why the build is enjoyable at an esoteric level, then — be the build 🆒 or not — not only is it quite unlikely that anyone will have the motivation to commit to it, but we’re left to wonder what the whole point really is. A PC that really, truly is no more than a “meme character” in the usual sense^[4] is less than a full PC; not a character, but a mere bauble.

Conclusion

We’ve seen that, by taking esoterica seriously, the differentiae specificae between builds/jobs — or more generally, between gameplay styles — reveal themselves to us. Moreover, these esoteric revelations don’t exist in a vacuum, & as a result, they shed light onto what makes MapleStory unique, what gameplays are possible, & how the game can even be fun to play at all. This perspective, by the way, collapses the distinction between “odd” & “nonodd”: both are just various ways of tracing paths through game-space.

So, if anyone questions your character-build choices, or seems a bit indifferent to the differentiae specificae that separate one build from another, & you suspect that they’re unlikely to read a dense philosophic paper, then perhaps you can simply respond with: “anyone can do a lot of damage, but it’s another thing entirely to have a reason to log in in the first place”.

---

Footnotes for “Appendix B: The exoteric & the esoteric”

1. [↑] Excepting that of Homing Beacon.
2. [↑] This appendix consistently uses odd in the specific sense of “odd sēnsū lātō” (for which, see: “Oddness”), unless otherwise specified.
3. [↑] And certain other things which, for simplicity, we here leave out of consideration.
4. [↑] Don’t be too confused by the fact that this term has nothing to do with memes in the usual sense. This is in the specialised sense of meme “joke”, so that a more appropriate term would be joke build — but here I use the conventional nomenclature.
5. [↑] Emergence is also a common theme throughout An all too brief polemic against the post-Big-Bangification of nominally “pre-Big-Bang” MapleStory.