Retiering the Classes: Home Base

[eggynack]

JaronK's tier system for classes is pretty neat. Least I think so. A hierarchical arrangement of game objects is tricky to design in the best of times, but it's perhaps at its trickiest in a system as complex as 3.5, and cooperative for that matter. The original system has prevailed throughout the years, sticking around while more complex variations pop up and often fade away. It's a great thing, but also a somewhat problematic one, as various classes in the game were mistiered, or, if they're sufficiently obscure, they weren't tiered at all. Moreover, some of the underlying rules for tiers are a bit on the wonky side. Our goal here is to retier the classes over a long period of time, knocking them out approximately three at a time in a variety of subthreads, and create as perfect a tier list as is plausible. Our other goal is to discuss classes, cause doing that is neat.

This thread, then, has the basic purpose of discussing stuff like procedure, and what classes we should tier, along with any other essential functioning stuff. It will also be a nexus for all the child threads that are devoted to class sets, linking to them and having useful information and such. So, with that out of the way, let's get to the important stuff.

What are the tiers?

The simple answer here is that tier one is the best, the home of things on the approximate problem solving scale of wizards, and tier six is the worst, land of commoners. And problem solving capacity is what's being measured here. Considering the massive range of challenges a character is liable to be presented with across the levels, how much and how often does that character's class contribute to the defeat of those challenges? This value should be considered as a rough averaging across all levels, the center of the level range somewhat more than really low and really high level characters, and across all optimization levels (considering DM restrictiveness as a plausible downward acting factor on how optimized a character is), prioritizing moderate optimization somewhat more than low or high.

A big issue with the original tier system is that, if anything, it was too specific, generating inflexible definitions for allowance into a tier which did not cover the broad spectrum of ways a class can operate. When an increase in versatility would seem to represent a decrease in tier, because tier two is supposed to be low versatility, it's obvious that we've become mired in something that'd be pointless to anyone trying to glean information from the tier system. Thus, I will be uncharacteristically word light here. The original tier system's tier descriptions are still good guidelines here, but they shouldn't be assumed to be the end all and be all for how classes get ranked.

Consistent throughout these tiers is the notion of problems and the solving thereof. For the purposes of this tier system, the problem space can be said to be inclusive of combat, social interaction, and exploration, with the heaviest emphasis placed on combat. A problem could theoretically fall outside of that space, but things inside that space are definitely problems. Another way to view the idea of problem solving is through the lens of the niche ranking system. A niche filled tends to imply the capacity to solve a type of problem, whether it's a status condition in the case of healing, or an enemy that just has too many hit points in the case of melee combat. It's not a perfect measure, both because some niches have a lot of overlap in the kinds of problems they can solve and because, again, the niches aren't necessarily all inclusive, but they can act as a good tool for class evaluation.

Tier one: Incredibly good at solving nearly all problems. This is the realm of clerics, druids, and wizards, classes that open up with strong combat spells backed up by utility, and then get massively stronger from there. If you're not keeping up with that core trio of tier one casters, then you probably don't belong here.

Tier two: We're just a step below tier one here, in the land of classes around the sorcerer level of power. Generally speaking, this means relaxing one of the two tier one assumptions, either getting us to very good at solving nearly all problems, or incredibly good at solving most problems. But, as will continue to be the case as these tiers go on, there aren't necessarily these two simple categories for this tier. You gotta lose something compared to the tier one casters, but what you lose doesn't have to be in some really specific proportions.

Tier three: Again, we gotta sacrifice something compared to tier two, here taking us to around the level of a swordsage. The usual outcome is that you are very good at solving a couple of problems and competent at solving a few more. Of course, there are other possibilities, for example that you might instead be competent at solving nearly all problems.

Tier four: Here we're in ranger/barbarian territory (though the ranger should be considered largely absent of ACF's and stuff to hit this tier, as will be talked about later). Starting from that standard tier three position, the usual sweet spots here are very good at solving a few problems, or alright at solving many problems.

Tier five: We're heading close to the dregs here. Tier five is the tier of monks, classes that are as bad as you can be without being an aristocrat or a commoner. Classes here are sometimes very good at solving nearly no problems, or alright at solving a few, or some other function thereof. It's weak, is the point.

Tier six: And here we have commoner tier. Or, the bottom is commoner. The top is approximately aristocrat. You don't necessarily have nothing in this tier, but you have close enough to it.


If you have any suggestions for modifications to this setup, go ahead, but I think the general format of, "Each tier is worse than the one above, and can be that in a large number of ways," is solid. It's a structure that can plausibly handle the addition of random new classes and systems without breaking down.

Important notes, procedural and otherwise

Tier Voting Procedure:The basic procedure here will be to vote on classes in the suitable thread. Threads will stay open for an indefinite quantity of time, though I could get bored in like a year and someone else could take up the thread altering game. I'll be checking and altering the numbers reasonably often to match changes in vote, and you can alter your vote whenever you want. I expect each thread to handle roughly three classes, though I could see some going up to five or six. We probably don't need to spend two entire threads covering classes that are obviously tier one, for example. Votes don't necessarily have to take on integer values, though I'm gonna say you should stick to rationals, cause supporting the alternative seems too hard and not worth it. I don't really have much I can do with, "Tier three sometimes, tier four other times," and, "High tier two," is just going to be a two, so make sure that whatever you do can be reasonably put into a fancy spreadsheet.

One really important thing here is that you can't just toss a vote out into the void with no information and then just leave. You need some solid justification for your vote, and preferably some interaction with the discussion. If you don't, it's not that big a deal, cause you can always add justification, but your vote may not be counted until then, and you'll be notified if you've been left out. Our goal here is accuracy in tiering, and if you think a class is tiered in a certain way premised on incorrect knowledge, then that should be plainly visible from what you're saying about the class. A few sentences is reasonable, a paragraph or two is quite good, and a few posts on the topic is great. Generally speaking, the more controversial a class is, the more explanation you're going to want to give, and the same is the case for individually weird votes. Just giving a wizard a 1 and saying, "Jeez, you called this a one explicitly in your initial post. How much detail do you really want me to give?" is likely sufficient. Doing the same about a class that's had ten pages of individual attention is probably not.

On ACF's, feats, items, dips, prestige classes, and so on:The default here will be to consider everything that isn't a class or prestige class, and nothing that is a class or prestige class. A general assumption is that the more obscure something is, the less likely it is to be on any particular character, and thus the less it should factor in. What matters most are things that a class has access to or makes good use of by dint of their class features. If a commoner can do it just as well, it's not a major class consideration. One major exception to this is individual game objects that merit a tier adjustment in and of themselves, and that largely lack for substitutes. The same may sometimes be the case for two object interactions, but that's more of an edge case and should be looked at on a case by case basis. When these things happen, we'll split off the ACF or feat (or item, but that's rarer) altered class and call it its own entry.

In all cases, use your best judgement and discretion. I think we'll get some good results here.

The Threads

The Fixed List Casters: Beguiler, Dread Necromancer, and Warmage


The Obvious Tier One Classes: Archivist, Artificer, Cleric, Druid, Sha'ir, and Wizard


The Mundane Beat Sticks (part one): Barbarian, Fighter, Samurai (CW), and Samurai (OA)


The Roguelikes: Ninja, Rogue, and Scout


The Pseudo-Druids: Spirit Shaman, Spontaneous Druid, Urban Druid, and Wild Shape Ranger


The Jacks of All Trades: Bard, Factotum, Jester, and Savant


The Tome of Battlers: Crusader, Swordsage, and Warblade


The NPCs: Adept, Aristocrat, Commoner, Expert, Magewright, and Warrior


The Vaguely Supernatural Melee Folk: Battle Dancer, Monk, Mountebank, and Soulknife


The Miscellaneous Full Casters: Death Master, Shaman, Shugenja, Sorcerer, and Wu Jen


The Wacky Magicists: Binder, Dragonfire Adept, Shadowcaster, Truenamer, and Warlock


The Slow Casting Melee Folk: Duskblade, Hexblade, Paladin, Ranger, Sohei, and Spellthief


The Pseudo-Clerics: Evangelist, Favored Soul, Healer, Mystic, and Spontaneous Cleric


The Auraists: Divine Mind, Dragon Shaman, and Marshal

The Rankings

Beguiler: Tier two.

Dread necromancer: Tier two.

Warmage: Tier three.

Archivist: Tier one.

Artificer: Tier one.

Cleric: Tier one.

Druid: Tier one.

Sha'ir: Tier one.

Wizard: Tier one.

Barbarian: Tier four

Fighter: Tier four

CW Samurai: Tier five

OA Samurai: Tier five

Ninja: Tier four

Rogue: Tier four

Scout: Tier four

Spirit Shaman: Tier two

Spontaneous Druid: Tier one

Urban Druid: Tier two

Wild Shape Ranger: Tier three

Bard: Tier three

Factotum: Tier three

Jester: Tier three

Savant: Tier four

Crusader: Tier three

Swordsage: Tier three

Warblade: Tier three

Adept: Tier four

Aristocrat: Tier six

Commoner: Tier six

Expert: Tier five

Magewright: Tier five

Warrior: Tier six

Battle Dancer: Tier five

Monk: Tier five

Mountebank: Tier five

Soulknife: Tier five

Death Master: Tier two

Shaman: Tier one

Shugenja: Tier three

Sorcerer: Tier two

Wu Jen: Tier one

Binder: Tier three

Dragonfire Adept: Tier three

Shadowcaster: Tier four

Truenamer: Tier five

Warlock: Tier three

Duskblade: Tier three

Hexblade: Tier five

Paladin: Tier four

Ranger: Tier four

Sohei: Tier five

Spellthief: Tier four

Evangelist: Tier two

Favored Soul: Tier two

Healer: Tier three

Mystic: Tier two

Spontaneous Cleric: Tier two

Divine Mind: Tier five

Dragon Shaman: Tier five

Marshal: Tier five

And here's a link to the spreadsheet.

[eggynack]

Here's a reserved post. Dunno if I'll even need it, but better safe than sorry. Posting away should be good now. I'll put up the fixed list casters thread in like an hour.

[Hurnn]

I like the tier descriptions mostly, except the language of tier 1 and 2. Tier one isn't Incredibly good at solving all problems to me. I think of tier 1 as: Solves every problem perfectly, and then does the extra credit questions at the end of the test. Tier 2 is solves some problems perfectly, and the rest really well.

[eggynack]

I like the tier descriptions mostly, except the language of tier 1 and 2. Tier one isn't Incredibly good at solving all problems to me. I think of tier 1 as: Solves every problem perfectly, and then does the extra credit questions at the end of the test. Tier 2 is solves some problems perfectly, and the rest really well.

Yeah, might be a bit hyperbolic. I might tone it down to, like, incredibly good at solving a ton of problems. I'm mostly trying to capture the essence of, "Top of the curve." I dunno if you're right on two. As I noted in my description, I view each step down the tier system as a sort of toning down of one factor or another. You can keep power or versatility, but you can't keep both at the exact levels you once had them. You can lose some of both too, but your "stats" will be lower than another tier two's "high stat". With that in mind, tier two doesn't necessarily solve some problems perfectly and the rest really well. Maybe one tier two solves a ton of problems very well, instead of a ton of problems incredibly well. Maybe another does as you said, and has a core competency that's excellent, and a surrounding region that's just great. Point being, I think that basically a single classification for tier one makes sense, but anything lower is going to be... complicated.

[AnachroNinja]

I'm on board, I like that problem solving is the prime criteria. I feel like it's worth noting that while we obviously do need to consider access to Items and feats, we should generally not put to much weight on things that are not relevant to the class itself. The classic example being candles of invocation. Whereas a ring of blink or weapon crystals is definitely a big factor for rogues and needs to be considered.

I'm not sure obscurity is really a good criteria though. For better or worse, everyone's got the internet now and as such, access to way more then just the books they physically own. I think a better process might be judging how obvious it is to use a feat or item. Any druid who sees a summoners totem is going to almost instantly see the advantages, but not every scout is going to immediately recognize the impact of chronocharms of the horizon Walker. That's honestly not the best example, because that one is pretty straightforward even if not as obvious as the totem, but it's what I could think of.

Anyway, just some thoughts. I'm on board for this.

[eggynack]

I'm on board, I like that problem solving is the prime criteria. I feel like it's worth noting that while we obviously do need to consider access to Items and feats, we should generally not put to much weight on things that are not relevant to the class itself. The classic example being candles of invocation. Whereas a ring of blink or weapon crystals is definitely a big factor for rogues and needs to be considered.

I already have that noted under the third procedural paragraph. Maybe I should clearly mark the stuff in that section better.

I'm not sure obscurity is really a good criteria though. For better or worse, everyone's got the internet now and as such, access to way more then just the books they physically own. I think a better process might be judging how obvious it is to use a feat or item. Any druid who sees a summoners totem is going to almost instantly see the advantages, but not every scout is going to immediately recognize the impact of chronocharms of the horizon Walker. That's honestly not the best example, because that one is pretty straightforward even if not as obvious as the totem, but it's what I could think of.

Maybe. I'm hesitant to fully drop obscurity as a factor. A lot of people talk about their core only games, or their game that has core as well as the PHB II, some completes, the magic item compendium, and some wacky setting specific book. Not all games use all books. I think the criteria you're talking about could fall roughly under "optimization".

[eggynack]

Just opened up the first tier voting thread. You can find it at the bottom of the first post, or you can find it here.

[GilesTheCleric]

The descriptions for each tier confuse me (as a native english speaker). I can probably guess what you mean when you say "solves a lot" vs "solves many". But, the vague comparative quantities could perhaps be a bit more explicit in just how much they encompass, or could be better contrasted with each other. Thank you!

[eggynack]

The descriptions for each tier confuse me (as a native english speaker). I can probably guess what you mean when you say "solves a lot" vs "solves many". But, the vague comparative quantities could perhaps be a bit more explicit in just how much they encompass, or could be better contrasted with each other. Thank you!

I'm mostly just trying to indicate less. Fewer problems solved, less power applied to the problems in question. Conveying that is tricky without getting weirdly recursive (fewer problems than the tier above this solved, with roughly equal power, or vice versa). Any suggestions for how that should be constructed? I've been really iffy on the whole thing. The main guiding principle, I think, should just be that a lower tier is worse than a higher one. Obvious, perhaps, but the other stuff strikes me as weirdly distracting.

[GilesTheCleric]

I'm mostly just trying to indicate less. Fewer problems solved, less power applied to the problems in question. Conveying that is tricky without getting weirdly recursive (fewer problems than the tier above this solved, with roughly equal power, or vice versa). Any suggestions for how that should be constructed? I've been really iffy on the whole thing. The main guiding principle, I think, should just be that a lower tier is worse than a higher one. Obvious, perhaps, but the other stuff strikes me as weirdly distracting.

Comparatives are always tricky, but what about a sort of sliding scale of "none - few - many - most - all"? That's 5 different amounts, which probably roughly correspond to the tiers. For me, things like "lots" or "tonnes" connotes a limit of infinity with no real guidance on how practically large those amounts are (is "lots" equal to 20? 100? 10.000?). With the above terms, there's a limit of 100%, which I think makes mentally organizing them easier. How do you feel about this?

Edit: Sorry, but I want to clarify, since I'm getting a suspicion that we're looking at different things. I mean the comparatives used within each tier individually, not from one tier to another.

[eggynack]

Comparatives are always tricky, but what about a sort of sliding scale of "none - few - many - most - all"? That's 5 different amounts, which probably roughly correspond to the tiers.

Sounds solid, though I think I'ma replace all with nearly all.

Edit: Replaced none with nearly none, cause I think having none as a category winds up not telling you much. Might swap many with some too. Haven't swapped out the "very good"s. You think that should be changed too?

[Grod_The_Giant]

I feel like we might want to fiddle more with tier definitions and "what to consider" before we start really ranking them. Making definitions vaguer doesn't seem terribly likely to result in more productive discussion. In particular, it seems like we should settle some of the past debates about baseline optimization levels and transformative feats/ACFs...

(It also seems to me that the "Tier 3.5 problem" is still exists-- it doesn't exactly feel right to put both the Warblade and Barbarian in the same the same tier, but both are pretty mediocre at all noncombat things.)

[Gullintanni]

One of the nuances inherent to JaronK's tier list that I find causes a lot of problems for those who haven't spent a lot of time analyzing the Tier list is his take on optimization thresholds. JaronK's tier list assumes equivalent levels of optimization, and some would be quick to point out that some classes get a heck of a lot more mileage out of heavy optimization than others.

Compare Monks and Rangers, for example. Even with heavy Op, it's difficult to move a single classed monk out of Tier 5. Really the best you can hope for is low Tier 4. Rangers, on the other hand, through ACFs, can go from mid Tier 4 to high Tier 3, especially with Dragon Magazine content. So if the premise here is performing a full re-evaluation of all classes, it may be useful to assign optimization approximations. For example, one might rate Warblades as follows:

Base Tier: 3
Optimization Floor: Low Tier 3 (Warblades come with all the tools they need to perform as Tier 3 classes within their chassis, and even sub-optimal feat choice can't screw this up).
Optimization Ceiling: High Tier 3 (No combination of tricks that a Warblade brings to the table can really be considered encounter breaking.)

A Ranger might be assessed as follows:

Base Tier: 4
Optimization Floor: Tier 5 (Sub-optimal feat and skill choices can make it difficult for Rangers to contribute even within their assumed primary roles)
Optimization Ceiling: High Tier 3 (With options like Mystic Ranger and SotAO, Rangers can combine potent spellcasting abilities with their already strong martial chassis)

The parenthetical notes needn't be included in the final tier list, I've included them here in order to better explain the concept I'm putting forward. So...would a Base Tier//Optimization Threshold format be useful? I don't think it'd be a great deal more work, since the playground's basically done all the analysis, it's just a matter of gathering a consensus.

[rrwoods]

I like this!

The second half-ish of your opening post here is just a tad hard to parse. I find myself kind of mentally losing what I'm reading about. Something to help anchor/re-center/whatever might be helpful:

Spoiler: Maybe headers, like this

Show

Basic Procedure: Vote on classes in the suitable thread. Threads will stay open for an indefinite quantity of time, though I could get bored in like a year and someone else could take up the thread altering game. I'll be checking and altering the numbers reasonably often to match changes in vote, and you can alter your vote whenever you want. I expect each thread to handle roughly three classes, though I could see some going up to five or six. We probably don't need to spend two entire threads covering classes that are obviously tier one, for example.

Explain Your Vote. One really important thing here is that you can't just toss a vote out into the void with no information and then just leave. You need some solid justification for your vote, and preferably some interaction with the discussion. If you don't, it's not that big a deal, cause you can always add justification, but your vote may not be counted until then, and you'll be notified if you've been left out. Our goal here is accuracy in tiering, and if you think a class is tiered in a certain way premised on incorrect knowledge, then that should be plainly visible from what you're saying about the class. A few sentences is reasonable, a paragraph or two is quite good, and a few posts on the topic is great. Generally speaking, the more controversial a class is, the more explanation you're going to want to give, and the same is the case for individually weird votes. Just giving a wizard a 1 and saying, "Jeez, you called this a one explicitly in your initial post. How much detail do you really want me to give?" is likely sufficient. Doing the same about a class that's had ten pages of individual attention is probably not.

On ACFs, feats, WBL, and Similar: The default will be to consider these somewhat. A general assumption is that the more obscure something is, the less likely it is to be on any particular character, and thus the less it should factor in. What matters most are things that a class has access to or makes good use of by dint of their class features. If a commoner can do it just as well, it's not a major class consideration. One major exception to this is individual game objects that merit a tier adjustment in and of themselves, and that largely lack for substitutes. The same may sometimes be the case for two object interactions, but that's more of an edge case and should be looked at on a case by case basis. When these things happen, we'll split off the ACF or feat (or item, but that's rarer) altered class and call it its own entry.

On Multiclassing/Prestige Classes: We're mostly not considering these. This is the tier system for classes, so considering classes separate from the one you're supposed to be considering is... bad. I think of this as kind of the inverse of the above situation. The default is to never consider these things, but if the class has a lot of substitutes, and you're not using it that extensively, then there may be situations where it makes sense to think about it. Dipping into one of like a dozen different classes to pick up a domain on the cheap is probably fine. If you're spending ten levels in sublime chord, then we're just not talking about a bard any more.

In all cases, use your best judgement and discretion. I think we'll get some good results here.

[Aimeryan]

EDIT: I have changed this post fairly drastically since it was first posted in order to update it.

Good work on the opening post - it hits the nail of every point that needed to be covered.

I want to look at how the tiers work, both for JaronK's tiering and for what I hope this Community tiering might go for. Rather than use words, since I think that has been done enough and still ends up being misunderstood, I went for something more visual. The bottom axis, which I forgot to label, is the problem space; the range of problems for which there are distinctly different solutions for. This is how JaronK's tiering works as I understand it:

Tier 1	Tier 2	Tier 3 (upper)	Tier 3 (lower)/Tier 3.5	Tier 4	Tier 5	Tier 6
488 CP	326 CP	315 CP	210 CP	115 CP	70 CP	30 CP

There are two particular issues here though:

1. Tier 2 and upper Tier 3 are very similar; the only difference of note is that Tier 2 can break the game - except if that occurs then, well, your game is broken. It is something worth noting about a class but not worth tiering differently, especially because classes in Tier 3 are often as good as those in Tier 2. In fact, the only reason this exists is because of the barred entry being forced by requiring Tier 2 classes to break the game.
2. There is a "hidden" Tier 3.5, mostly consisting of gishes. It exists because the gap between upper Tier 3 and lower Tier 3 is so huge - although JaronK himself didn't use it. Without this tier those classes tend to get dumped in with those that are upper Tier 3 (and described by what described them), which is unfortunate because those classes should be in different tier to the gishes.

I propose something a little different:

Tier 1	Tier 2	Tier 3	Tier 4	Tier 5	Tier 6
425 CP	315 CP	210 CP	115 CP	70 CP	30 CP

Tier 1 is the same, just anything above strong has lost meaning to record or differentiate upon, so it isn't there. Tier 2 has similarly gone through the same process - now it is fine for those in the upper Tier 3 that would have been excluded by this. Tier 3 is now just what lower Tier 3 was, otherwise called Tier 3.5. Tier 4 and onwards are unchanged.

Mostly this would likely put Beguiler, Dread Necromancer, etc., in with Sorcerer and the other Tier 2s. Gamebreaking potential would be noted in the respective class notes, like any other particularity. Gishes can happily call Tier 3 home.


Here is the spreadsheet I made to create these graphs (download/copy for own use). Umm, I couldn't figure out how to copy a graph, so I've just been changing the range of the one graph I have when I want to look another Tier (I didn't just create new ones because it was finicky to customise each axis and other things to how I desired).

The graphs use a 30 point range for the problem space with 15 values each can take (20 for JaronK's); this gives a total of 450 points possible (600 for JaronK's). I've included the total for each graph; CP stands for comparison points.

An important point to note is that the Tiers are not fixed to the state of those graphs; the states above are simply to demonstrate the tiers matching the regular classes put into them. Below I have posting some alternative states that a Tier could take, but these are again just a demonstration, not something that is fixed. The important point to take away is that from Tier to Tier you are losing or gaining that power/versatility interaction somewhere overall, as Eggynack mentioned. The comparison points should indicate this if the visual graphs do not.

Here are some alternative states (none for Tier 1 and Tier 6 for what I hope are obvious reasons):

Tier 2	Tier 3	Tier 4	Tier 5
315 CP	210 CP	115 CP	70 CP

Note that these particular alternate states for Tier 4 and Tier 5 would probably not exist in the JaronK tiering; Tier 4's lack of a single strong area would disqualify it, and Tier 5's lack of anything competent would just land it in Tier 6 regardless of how much noticeable areas such a class reached.

[Fizban]

I've always thought it funny how people would say that "tier 3 is the best," when they couldn't actually define what tier 3 is. While I think he's taken it a bit too far, eggynack has essentially hit it right on the nose: tier 3 is less than tier 2, that's pretty much it. Tier 1 is the best at everything, tier 2 is the best at one thing or good at a lot of things, and tier 3 is worse than tier 2. The only actually concrete definitions are 1 and 2, so anything more removed is just mincing words: tiers 4 and 5 are the same thing (dependent upon so many other build elements that the base class matters little, the best you can say is that tier 5 is classes that actively work against themselves), and tier 6 is just fancy way of saying NPC classes.

The main difference I seemed to be getting is that in some people's minds, tier 2 meant broken spells and cheese. Sorcerer was tier 2 because it could use Contingency and Arcane Fusion and Arcane Spellsurge and Time Stop, etc, while Beguiler is supposedly tier 3 because it doesn't have the cheese, even though it's essentially the best mind mage if cheese is barred or people just aren't doing ridiculous Contingency shenanigans and whatnot. For people that don't use any of that, Beguiler and Sorcerer are both obviously tier 2.

So my suggestion for definitions:
Tier 1: can be the best at anything (essentially limited to full casters with variable spell lists)
Tier 2: can be the best at one thing, or good at multiple things (essentially limited to full casters)
Tier 3: clearly worse than tier 2 but still clearly more capable than-
Tier 4: clearly lacking in power or options, usually due to lack of magic or supernatural abilities
Tier 5: classes that actively work against themselves or unduly restrict the player
"Tier 6": NPC classes that are already labeled as not for PC use

Tier 1 has your big 3, tier 2 has your sorcerers and beguilers, tier 3 has all those popular classes that aren't full casters, tier 4 is your barbarian/fighter, and tier 5 is your monk/samurai.

Then you have the high-op and cheese notes, which don't affect the tiering but are simply there so you know what major spikes the class has printed. The tier 1's could probably have this omitted for brevity, since they by definition have access to almost all of them, but this is where you put the notes for "beguiler lacks access to the same cheese a sorcerer can get, having only. . . "

Edit: to be clear, I don't much like the criteria of "problems solved," because the problems depend on the campaign and some specializations are just obviously better than others. Unless you nerf it into oblivion or run campaign focused on mindless foes, mind magic is always more capable of "solving problems" than blasting. That doesn't make beguiler a higher tier though, because an undead or dungeons+traps+constructs campaign is perfectly valid. It should be assumed that anything a PC is trying to do is of equal value to that of the other PCs, because the DM should be making sure they get equal spotlight.

[bekeleven]

I've always found the simplest explanation of tiers to be a grid:

Versatiliy	High	Low
High Power	1	2
Medium Power	3	4
Low Power	5	6

Even though it disagrees with JaronK's definitions, particularly at tiers 2, 4, and 5.

That said, I still think that one of the primary issues with community tiering efforts is that people play at different optimization levels, and the tiers change as you optimize.

[OldTrees1]

Good work on the opening post - it hits the nail of every point that needed to be covered.

I want to look at how the tiers work, both for JaronK's tiering and for what I hope this Community tiering might go for. Rather than use words, since I think that has been done enough and still ends up being misunderstood, I went for something more visual. The bottom axis, which I forgot to label, is the problem space; the range of problems for which there are distinctly different solutions for. This is how JaronK's tiering works as I understand it:

Excellent analysis. I especially liked that you used graphs to accurately express the curves rather than just default to 1 or 2 variables.

Now unto your proposal:
At first I was mildly skeptical about removing game breaking abilities from the system. However, by narrowing the strengths you were considering, you were able to create a fairly smooth transition from your Tier 1 to Tier 6

Tier 1	Tier 6

The only place I can see you further improving is the Tier 3 state(old Tier 3.5).

Tier 2	Tier 3	Tier 4

Personally I would have expected a state that is noticeable in all cases but only strong in a few cases and competent in a few more. I say this because Tier 2 is showing always noticeable & almost always competent in addition to being strong twice as often. If Tier 4 -> 3 was filling out the noticeable and increasing the competent, then Tier 3 -> 2 could be filling out the competent and increasing the strong.

[Aimeryan]

Argh, I spelt noticeable wrong! Eh, the word was due to be modernised, viva la revolution!

I did think about whether the new Tier 3 (old Tier 3.5) would be noticeable in all problem spaces, but I think just about all gishes do suffer from areas where they are just plain useless. I wanted the tiers to represent average fit for the classes assigned to them.

The same reason is why Tier 4 has one strong peak (that has some width to it, if minor, because of how just hitting things actually ends up being a pretty broad solution), and then not even reach noticeable anywhere else. That said, I am wondering if Rogue would fit the new Tier 3 or Tier 4, hmm.

Edit: I think Rogue fits the new Tier 3 fine, but the one state I posted may not look quite right for it. An alternative, but roughly equal in terms of area-under-the-line covered, Tier 3 state would look like this:



That may be more what you were thinking of? It trades reaching strong in two places (with some width to each) for general competence and, with a little further optimisation, easy to reach being all-round noticeable. It is weaker than Tier 2, but stronger than Tier 4 (unless your campaign is just fight after fight).

You could, of course, have something in between the two Tier 3 states visualised; the states for any one Tier just need to be roughly similar in area - and even then you can have high/low in the Tier.

[eggynack]

I feel like we might want to fiddle more with tier definitions and "what to consider" before we start really ranking them. Making definitions vaguer doesn't seem terribly likely to result in more productive discussion.

I dunno if we're in the exact right vagueness spot, but I think this is a whole lot closer. All this specificity and definition focus takes us further from some essential truths of the tiers. Is tier two really this incredibly specific structure that has some game breaking power, but with only a percentage pulled off of a longer more versatile list? I'd say no. Tier two, in the simplest terms, means, "About as good as a sorcerer." No more, no less. Information beyond that should maybe guide us, but if the system does not fulfill the logical statement, "You are tier two iff you are about as good as a sorcerer," then the system has failed. It has, in fact, brought us further from truth. There's a lot of r

In particular, it seems like we should settle some of the past debates about baseline optimization levels and transformative feats/ACFs...

Do you think it should be different from how it is? I've generally thought that, say, wild shape ranger should be separately considered. My understanding is that that's a pretty unanimously fine with folk. And, if we're not cutting something off for separate consideration, then it makes a lot of sense to me to consider it as a more or less fringe element of the main class. As for baseline optimization, there's definitely room for debate, but I don't think my position, that even optimization levels a bunch of standard deviations off the center should be considered a little, is that incompatible with the position that we should only consider things maybe one or two standard deviations away. The whole idea is weirdly like an integral of that bell curve.

(It also seems to me that the "Tier 3.5 problem" is still exists-- it doesn't exactly feel right to put both the Warblade and Barbarian in the same the same tier, but both are pretty mediocre at all noncombat things.)

It might still be an issue, but it's not as much of one. My system operates on what is essentially a cross sectional power/versatility thing. Even a large versatility loss can generally be compensated for by a sufficiently large gain in power (though the gain might be massive), and vice versa. Maybe the warblade's cross section does merit a three, or maybe it doesn't, but if it doesn't then that might just be the nature of the class. The 3.5 thing is interesting, meanwhile. Should we really add a 7th tier for it? It seems kinda extreme.

One of the nuances inherent to JaronK's tier list that I find causes a lot of problems for those who haven't spent a lot of time analyzing the Tier list is his take on optimization thresholds. JaronK's tier list assumes equivalent levels of optimization, and some would be quick to point out that some classes get a heck of a lot more mileage out of heavy optimization than others.

Yeah, it's complicated. My thinking on the topic is weirdly calculus oriented.

The parenthetical notes needn't be included in the final tier list, I've included them here in order to better explain the concept I'm putting forward. So...would a Base Tier//Optimization Threshold format be useful? I don't think it'd be a great deal more work, since the playground's basically done all the analysis, it's just a matter of gathering a consensus.

I'd prefer not to, personally. It sounds like it'd be... complicated. Even a low/medium/high thing would roughly triple the effort, and it'd add a lot of talk about what qualifies in each section.

The second half-ish of your opening post here is just a tad hard to parse. I find myself kind of mentally losing what I'm reading about. Something to help anchor/re-center/whatever might be helpful.

Sounds solid. Prolly gonna do it up like that. I did notice that whole section kinda getting away from me. Wasn't sure precisely where I was going or when I was stopping when I started.

Good work on the opening post - it hits the nail of every point that needed to be covered.

I want to look at how the tiers work, both for JaronK's tiering and for what I hope this Community tiering might go for. Rather than use words, since I think that has been done enough and still ends up being misunderstood, I went for something more visual. The bottom axis, which I forgot to label, is the problem space; the range of problems for which there are distinctly different solutions for. This is how JaronK's tiering works as I understand it:

It's an interesting model. Pretty helpful way to think about the overall structure of tiers. The idea of ditching things above a certain power level is nifty. Still, I'm not entirely sure how big a fan I am of applying some universal graph to all classes of a particular tier. It's like, what if you have a kinda flat always competent graph? Not sure if that's what you were aiming for.

So my suggestion for definitions:
Tier 1: can be the best at anything (essentially limited to full casters with variable spell lists)
Tier 2: can be the best at one thing, or good at multiple things (essentially limited to full casters)
Tier 3: clearly worse than tier 2 but still clearly more capable than-
Tier 4: clearly lacking in power or options, usually due to lack of magic or supernatural abilities
Tier 5: classes that actively work against themselves or unduly restrict the player
"Tier 6": NPC classes that are already labeled as not for PC use

Interesting, though it might actually be vaguer than what I already have. It's clearly workable as a system. Maybe we should have a weird vote on how each tier should look, like that odd meta-thread was essentially suggesting.

Edit: to be clear, I don't much like the criteria of "problems solved," because the problems depend on the campaign and some specializations are just obviously better than others. Unless you nerf it into oblivion or run campaign focused on mindless foes, mind magic is always more capable of "solving problems" than blasting. That doesn't make beguiler a higher tier though, because an undead or dungeons+traps+constructs campaign is perfectly valid. It should be assumed that anything a PC is trying to do is of equal value to that of the other PCs, because the DM should be making sure they get equal spotlight.

I think it's a bit more expansive as a definition than you're giving it credit for. It's not about whether you can solve a particular problem completely, or how many problems you can solve, but, when considering the total scope of problems, what percent of "problem space" your class takes up. I might be obfuscating this more. But, for a simplified numerical example, if you can solve a particular problem with reasonable ease with the help of a roughly equally leveled party member, then we can call that 50% problem solving capability in that case, and if you can do it alone with reasonable ease, we can call that 100%. If you get all problems in the gaming universe at every level at 50%, then you should be tiered similarly to a class that's 100% on half and 0% on the other half. It's not exactly that, because I think we should expect each percent added to have somewhat lower marginal utility, but that's the overall idea. Problems do depend on campaign, so some classes will always be stronger, but that's gonna be a problem for literally any tier system. Someone's always gonna pick up a tier four rogue and fall behind a tier five fighter in the all undead campaign, y'know?

That said, I still think that one of the primary issues with community tiering efforts is that people play at different optimization levels, and the tiers change as you optimize.

Yeah, that's always gonna be a problem. I think there's value in getting the generic variation right though. Then, when you try to work out the variations, you aren't quite so stymied by the weirdness of the base system.

Edit: Went with two headers instead of four. Looks pretty reasonable, I think.

[OldTrees1]

It's an interesting model. Pretty helpful way to think about the overall structure of tiers. The idea of ditching things above a certain power level is nifty. Still, I'm not entirely sure how big a fan I am of applying some universal graph to all classes of a particular tier. It's like, what if you have a kinda flat always competent graph? Not sure if that's what you were aiming for.

Their idea gets detailed further in our discussion. The graphs for Tier 3 & Tier 3 alt as well as their last paragraph in the quote below should answer your question.

Using such graphs is a powerful tool.

Argh, I spelt noticeable wrong! Eh, the word was due to be modernised, viva la revolution!

I did think about whether the new Tier 3 (old Tier 3.5) would be noticeable in all problem spaces, but I think just about all gishes do suffer from areas where they are just plain useless. I wanted the tiers to represent average fit for the classes assigned to them.

The same reason is why Tier 4 has one strong peak (that has some width to it, if minor, because of how just hitting things actually ends up being a pretty broad solution), and then not even reach noticeable anywhere else. That said, I am wondering if Rogue would fit the new Tier 3 or Tier 4, hmm.

Edit: I think Rogue fits the new Tier 3 fine, but the one state I posted may not look quite right for it. An alternative, but roughly equal in terms of area-under-the-line covered, Tier 3 state would look like this:



That may be more what you were thinking of? It trades reaching strong in two places (with some width to each) for general competence and, with a little further optimisation, easy to reach being all-round noticeable. It is weaker than Tier 2, but stronger than Tier 4 (unless your campaign is just fight after fight).

You could, of course, have something in between the two Tier 3 states visualised; the states for any one Tier just need to be roughly similar in area - and even then you can have high/low in the Tier.

Tier 3		Tier 3 alt

I was thinking something between Tier 3 and Tier 3 alt. However your description of area under the curve made me realize that for a linear scale, the area under the curve is a good metric (although still best referenced via graphs). There still are qualitative differences between Tier 3 and Tier 3 alt but not large enough of differences for additional tiers.

You have my support!

[Fizban]

Interesting, though it might actually be vaguer than what I already have. It's clearly workable as a system. Maybe we should have a weird vote on how each tier should look, like that odd meta-thread was essentially suggesting.

Really? I felt like I was basically saying the same thing, just with the added caster/self-sabotaging notes.

I think it's a bit more expansive as a definition than you're giving it credit for. It's not about whether you can solve a particular problem completely, or how many problems you can solve, but, when considering the total scope of problems, what percent of "problem space" your class takes up. I might be obfuscating this more. But, for a simplified numerical example, if you can solve a particular problem with reasonable ease with the help of a roughly equally leveled party member, then we can call that 50% problem solving capability in that case, and if you can do it alone with reasonable ease, we can call that 100%. If you get all problems in the gaming universe at every level at 50%, then you should be tiered similarly to a class that's 100% on half and 0% on the other half. It's not exactly that, because I think we should expect each percent added to have somewhat lower marginal utility, but that's the overall idea. Problems do depend on campaign, so some classes will always be stronger, but that's gonna be a problem for literally any tier system. Someone's always gonna pick up a tier four rogue and fall behind a tier five fighter in the all undead campaign, y'know?

It basically works out to the same thing as build vs potential tricks, but from the other side, campaign vs potential threats. I just don't like phrasing it in terms of problem solving because then people focus on how their favorite option solves more problems because of some convoluted reason and then making up things like the "same game test" to "prove" the difference, when the game is actually up to the DM and players. Classes with more problem solving are what I'd call "clearly better" than those which lack it, no need try counting how many ways because down that path lies madness.

[danielxcutter]

So... if we go with the graphs, then is the total amount of power a class has proportionate to the area the graph covers?

[OldTrees1]

So... if we go with the graphs, then is the total amount of power a class has proportionate to the area the graph covers?

Yes. The larger the region the more power they have and different shapes of the region describe different ways the power is distributed.

The x axis is the problem possibility space (basically every possible situation they can be in).
The y value for a given x value is their strength at contributing in or otherwise handling that situation.
So the area under the curve (the shaded region) represents the sum of all these situations.

[eggynack]

Their idea gets detailed further in our discussion. The graphs for Tier 3 & Tier 3 alt as well as their last paragraph in the quote below should answer your question.

Using such graphs is a powerful tool.

Yeah, it's a useful thing, I think. A lot of my thinking on the tier system takes place in graphs. Theoretically, it actually winds up fitting onto a weird multi-dimensional graph that also covers stuff like optimization and level, with that problem space success area being the main thing calculated at the end. My thinking, the way I constructed things, is that you wouldn't simply have one, or two, or even a bunch of graphs associated with each tier. You'd have an infinity of them, all falling within certain area bounds. As long as there's the underlying premise that you can always modify the shape of the graph as long as you retain the same area, I think the whole thing works out really well. It's possible I'm going too far with that modifiable nature though.

Really? I felt like I was basically saying the same thing, just with the added caster/self-sabotaging notes.

Maybe. I've been a bit out of it on this thread, owing to distraction from the other one, mostly. I like tier one, and two is pretty good as well. Three is a bit bare bones, not really covering much of the actual possibility space we're working with, though that could be fine. Really not sure on four. Defining things by what they're not seems like it could cause problems. Especially when somewhat magical classes like the paladin, ranger, and adept are likely to wind up there, and when the monk, a notably supernatural class, is likely to get below there. Five seems pretty good. The notion of self sabotage is interesting. Definitely reflects the monk. A bit less sure whether it captures, say, a fighter. Six, I dunno if I like characterizing it as the NPC section. Yes, it has NPCs, and, if the CW samurai lands in five, only NPCs, but NPCs are going to show up outside of six. As we went over it last time, it almost seemed like six should be exclusively for commoners. It was that, aristocrats, and warriors, I think. And the latter two could maybe rise up if you squint.

It basically works out to the same thing as build vs potential tricks, but from the other side, campaign vs potential threats. I just don't like phrasing it in terms of problem solving because then people focus on how their favorite option solves more problems because of some convoluted reason and then making up things like the "same game test" to "prove" the difference, when the game is actually up to the DM and players. Classes with more problem solving are what I'd call "clearly better" than those which lack it, no need try counting how many ways because down that path lies madness.

So you're talking modes of interaction as a metric? Like, we take the niche system, normalize for the fact that the different niches aren't necessarily equivalently useful (though not that much, because we should expect the DM to sorta call upon various niches at various moments), fix the thing where sorcerers are ranked as though they can do everything at once, and grind that together into a tier system? Not necessarily exactly that, but is that the general criteria you're working with? It kinda makes sense. Little out there, but it could work.

[D.M.Hentchel]

Yes. The larger the region the more power they have and different shapes of the region describe different ways the power is distributed.

The x axis is the problem possibility space (basically every possible situation they can be in).
The y value for a given x value is their strength at contributing in or otherwise handling that situation.
So the area under the curve (the shaded region) represents the sum of all these situations.

Love this method; concise, visual, and easy to understand.

However I think it is ridiculous that we should factor in High and Low optimization.

Low Op games are a crazy place, the rules are regularly ignored, unspoken house-rules abound, encounters are weirdly designed, and characters often denied their abilities "because story". It's so hard to quantify these variables and if we are being honest how many Low Op groups are even going to see this.

High Op games suffer a similair problem in that the environment of the game is so wildly different that it's difficult to even distinguish classes and anyone in a high Op game is going to need a more complex understanding of the game than a tier list could ever offer.

On the other side though there are handbooks for every class, barring the extremely obscure ones, that a quick google leads you to.

I think that the tier list as a whole benefits from working under the idea that the character is making good decisions, not necessarily the best decision and certianly not just randomly.

Then we can still make note of how certian details effect high Op and low Op in our discussion without detracting from the tier list's primary use and debate.

[OldTrees1]

Love this method; concise, visual, and easy to understand.

Nice, another supporter.

However I think it is ridiculous that we should factor in High and Low optimization.

Low Op games are a crazy place, the rules are regularly ignored, unspoken house-rules abound, encounters are weirdly designed, and characters often denied their abilities "because story". It's so hard to quantify these variables and if we are being honest how many Low Op groups are even going to see this.

High Op games suffer a similar problem in that the environment of the game is so wildly different that it's difficult to even distinguish classes and anyone in a high Op game is going to need a more complex understanding of the game than a tier list could ever offer.

On the other side though there are handbooks for every class, barring the extremely obscure ones, that a quick google leads you to.

I think that the tier list as a whole benefits from working under the idea that the character is making good decisions, not necessarily the best decision and certainly not just randomly.

Then we can still make note of how certain details effect high Op and low Op in our discussion without detracting from the tier list's primary use and debate.

I see no problem with presuming they are picking good options rather than the best options or random options. Just make sure it is an equivalent optimization level.

[Fizban]

Maybe. I've been a bit out of it on this thread, owing to distraction from the other one, mostly. I like tier one, and two is pretty good as well. Three is a bit bare bones, not really covering much of the actual possibility space we're working with, though that could be fine. Really not sure on four. Defining things by what they're not seems like it could cause problems.

What's the difference between defining things by what they don't have, and what they do have? The only concrete definition anyone agrees on is the top, the first definition is "if you're not a wizard/cleric/druid you're probably not tier 1." It's much easier to work from the top down looking at what things can't do than it is to try to define how many things or how strong you need to be from the bottom up. Otherwise you'll just end up arguing endlessly about how many skills are worth what amount of spells and how much these class features are and on and on.

Especially when somewhat magical classes like the paladin, ranger, and adept are likely to wind up there, and when the monk, a notably supernatural class, is likely to get below there. Five seems pretty good. The notion of self sabotage is interesting. Definitely reflects the monk. A bit less sure whether it captures, say, a fighter.

When I say supernatural I mean heavily so, Incarnum, Binding, Invoking, etc. Monks are self-sabotaging (and barely supernatural at all), they go in 5 by default. Pal/Ranger go in 4, as does Fighter, as they are all clearly lacking in magic compared to 2/3 casters and at-will supernatural effects without heavy optimization.

Six, I dunno if I like characterizing it as the NPC section. Yes, it has NPCs, and, if the CW samurai lands in five, only NPCs, but NPCs are going to show up outside of six. As we went over it last time, it almost seemed like six should be exclusively for commoners. It was that, aristocrats, and warriors, I think. And the latter two could maybe rise up if you squint.

NPC classes aren't meant for PC use, so I don't see why they should even be rated myself. They are all deliberately inferior to PC classes and rating as if they're meant to be only muddies the waters. That's why "tier 6" is in quotes, because it's not actually a tier- it's a statement that NPC classes don't need to be tiered

My tier list makes little distinction between 3 and 4, because it's the bulk of the classes and the only distinction is up or down. 1 is clearly definable, 2 is clearly a step below that, but once you're more than one step removed from the starting position everyone's different step sizes start piling up. A firm bottom end is needed, so we define the bottom as self-sabotaging for a clear opposite end. Once you've sorted out the 1s, 2s, and self-sabotaging 5s, that leaves you with a pile to be split between 3s and 4s, which is a simple question of better/worse. Stuff that's clearly low-end (like Fighters and their lack of both magic and skills) goes to 4, stuff that's clearly high-end (stuff with lots of magic and skills like Bard) goes to 3, and work your way to the middle. The more stuff outside the class needed to match whatever is above it, the more likely it's a 4 than than a 3. This is where the meat of the arguing goes.

So basically I'm saying that there should only be two actual tiers of contention and trying to cram more in there is a fool's errand. Everything below tier 2 is so dependent on build and game state that trying to divide it into more than two groups isn't going to work, and doing so is misleading.

So you're talking modes of interaction as a metric?

I think you're reading a lot more into it than I am. All I'm saying is that trying to quantify the sum of all potential encounters/problems is the same as trying to do it for all potential builds, the same pointless logistical problem from a different angle. One man's edge case is another man's central proof of awesomeness. No math can be agreed upon or sufficient, down that path lies madness.

[Bucky]

I've always found the simplest explanation of tiers to be a grid:

Versatiliy	High	Low
High Power	1	2
Medium Power	3	4
Low Power	5	6

Even though it disagrees with JaronK's definitions, particularly at tiers 2, 4, and 5.

I explain it as three super-tiers - incompetent, competent and overpowered - each of which has generalist and specialist divisions. Generalists are ranked higher than specialists except in the incompetent range, where specialists have things they suck less at.

In practice, there's room for more divisions that don't correspond to numbered tiers; mid power hyper-generalists who can't competently solo most level-appropriate challenges but who can meaningfully contribute to any of them (e.g. dedicated group buff characters), multi-specialists in between tier 3 and 4 that do several things competently, and narrow tier 2 power spreads that lack competence in secondary areas vs. broad ones that could be in tier 3 even if they ignore their best tricks. And tier 0, for TO cheese that renders normal play irrelevant.

[eggynack]

Love this method; concise, visual, and easy to understand.

However I think it is ridiculous that we should factor in High and Low optimization.

Low Op games are a crazy place, the rules are regularly ignored, unspoken house-rules abound, encounters are weirdly designed, and characters often denied their abilities "because story". It's so hard to quantify these variables and if we are being honest how many Low Op groups are even going to see this.

High Op games suffer a similair problem in that the environment of the game is so wildly different that it's difficult to even distinguish classes and anyone in a high Op game is going to need a more complex understanding of the game than a tier list could ever offer.

On the other side though there are handbooks for every class, barring the extremely obscure ones, that a quick google leads you to.

I think that the tier list as a whole benefits from working under the idea that the character is making good decisions, not necessarily the best decision and certianly not just randomly.

Then we can still make note of how certian details effect high Op and low Op in our discussion without detracting from the tier list's primary use and debate.

I can see this as a position. My thinking is that there are two ways to consider optimization, with a small range and with a large range. Small range is what you're proposing. The bottom optimization sorcerer prepares some blasting and the occasional utility, and has decent charisma. The top optimization sorcerer

What's the difference between defining things by what they don't have, and what they do have? The only concrete definition anyone agrees on is the top, the first definition is "if you're not a wizard/cleric/druid you're probably not tier 1." It's much easier to work from the top down looking at what things can't do than it is to try to define how many things or how strong you need to be from the bottom up. Otherwise you'll just end up arguing endlessly about how many skills are worth what amount of spells and how much these class features are and on and on.

I think my problem might just be that the second part was a bit too specific. We could absolutely imagine a class with no magic whatsoever that lands above tier four. I was thinking of that tier in terms of a construct that'd disprove it, something that lacks that object but is still great, without considering the fact that we could construct an equally exacting definition in the other direction with equal inflexibility. Something like, at tier three, "Everything at and above this point has tons of magic."

When I say supernatural I mean heavily so, Incarnum, Binding, Invoking, etc. Monks are self-sabotaging (and barely supernatural at all), they go in 5 by default. Pal/Ranger go in 4, as does Fighter, as they are all clearly lacking in magic compared to 2/3 casters and at-will supernatural effects without heavy optimization.

I'm honestly not entirely sure what self-sabotaging means in this context. It might make sense when explained, but even if that's the case, the fact that it strikes me as this confusing in the moment means it probably shouldn't fit into a finalized tier system.

NPC classes aren't meant for PC use, so I don't see why they should even be rated myself. They are all deliberately inferior to PC classes and rating as if they're meant to be only muddies the waters. That's why "tier 6" is in quotes, because it's not actually a tier- it's a statement that NPC classes don't need to be tiered

They should be tiered cause they're surprisingly decent. An adept can plausibly fit into either a tier four or tier five game without too much issue. Also, I think people play them sometimes. If someone comes into an NPC class game with an adept, they should know to expect something maybe problematic. If anything, this information might be more important to log because it's assumed to be false, in a sense. People see NPC classes and assume crap. Breaking assumptions is a core goal of the tier system.

My tier list makes little distinction between 3 and 4, because it's the bulk of the classes and the only distinction is up or down. 1 is clearly definable, 2 is clearly a step below that, but once you're more than one step removed from the starting position everyone's different step sizes start piling up. A firm bottom end is needed, so we define the bottom as self-sabotaging for a clear opposite end. Once you've sorted out the 1s, 2s, and self-sabotaging 5s, that leaves you with a pile to be split between 3s and 4s, which is a simple question of better/worse. Stuff that's clearly low-end (like Fighters and their lack of both magic and skills) goes to 4, stuff that's clearly high-end (stuff with lots of magic and skills like Bard) goes to 3, and work your way to the middle. The more stuff outside the class needed to match whatever is above it, the more likely it's a 4 than than a 3. This is where the meat of the arguing goes.

So basically I'm saying that there should only be two actual tiers of contention and trying to cram more in there is a fool's errand. Everything below tier 2 is so dependent on build and game state that trying to divide it into more than two groups isn't going to work, and doing so is misleading.

You're correct that the gap between two and three is much larger than the ones that come later. I think it's possible to create some meaningful distinctions below that. Maybe. We've obviously had some trouble with it. Maybe there should just be three tiers or something. 1-2, 3-4, and 5-6. Wouldn't be crazy. I feel like we'd be losing something there though.

I think you're reading a lot more into it than I am. All I'm saying is that trying to quantify the sum of all potential encounters/problems is the same as trying to do it for all potential builds, the same pointless logistical problem from a different angle. One man's edge case is another man's central proof of awesomeness. No math can be agreed upon or sufficient, down that path lies madness.

I'm not exactly sure what your metric is then. Ya gotta have a metric of some kind.