A very ill-advised attempt to get rid of attributes

[Teridax]

So here's the name of the game:

Trying to take attributes out of Pathfinder 2e is, in my opinion, a very bad idea. Trying to do this in earnest will likely not leave you or your players happy, and I wouldn't recommend it.

The name of the game here is explicitly to brainstorm how one would go about implementing this terrible concept while breaking as little as possible... but also poke holes and see what these ideas inevitably break. This can be as little as "there's a +/-1 difference in the math at this level" and as large as "X class just stops working entirely". So long as the criticism is accurate, it's valid!

Participation is encouraged. If you want to poke holes at what exists, that's good. If you want to come up with a solution to fix what you or someone else broke, that's also good. If you want to come up with your own implementation and post it here, that's also good!

Now with this in mind, I'll start:

The Assumptions:

The basis of the following take hinges on a few assumptions:

Players are generally assumed to boost the "big 3" attributes on their characters (Dex, Con, and Wis) every time they get 4 attribute boosts per level, with exceptions for Dex on classes that can wear heavy armor. This implies defense and Perception modifiers converge towards similar amounts at higher levels.

Players are generally assumed to boost their key attribute whenever they can, such that their attack and/or spellcasting modifier and class DC follow a pretty consistent progression.

Players are generally assumed to boost their skills relatively freely, such that there's a fair amount of variance between skill modifiers.

Based on these, and when factoring in item bonuses (but obviously not status or circumstance bonuses), there's the assumption that character modifiers and DCs tend to end up at one of five ranks:
Untrained: "I didn't increase this skill," with a final modifier of 0.
Trained: "I'm not very good at this," usually up-to-expert proficiency and the attribute isn't your key attribute, with a final modifier of 12 + your level.
Expert: "I'm okay at this," usually up-to-master proficiency and the attribute isn't your key attribute, or up-to-expert proficiency and the attribute is your key attribute, with a final modifier of 14 + your level.
Master: "I'm good at this," usually up-to-master proficiency and the attribute is your key attribute (and sometimes up-to-legendary proficiency and the attribute isn't your key attribute), with a final modifier of 16 + your level. You can sort of kludge spellcaster proficiency into this category even though spell attack modifiers get stuck at a -1 relative to martials.
Legendary: "I'm exceptionally good at this," with up-to-legendary proficiency and the attribute is your key attribute, with a final modifier of 18 + your level.

You'll notice that if you shave 10 off most of these ranks, the numbers end up the same as regular TEML proficiency bonuses! This is the fun part where we take that observation and run way too far with it.

The Concept
With the above assumptions in mind, here goes:

Get rid of attributes and attribute boosts and flaws, but also the item bonuses and apex attribute boosts you'd see replicated in Automatic Bonus Progression (plus the stuff that got missed like the Kineticist's gate attenuator bonus). Appropriately reduce the item bonus of certain mutagens and other effects that aim to provide you an item bonus over the baseline.

Your proficiency bonus past untrained equals your level and a half, rounded up, plus the usual increase from your rank.

Your level 1 proficiency rank for saves, Perception, and attacks is equal to the highest proficiency rank you can get (and the Fighter gets their weapon specialization moved to level 1 so they're legendary in just that until 19th level). If the proficiency relies on an attribute that isn't any of your class's possible key attributes, reduce that proficiency by one rank. Remove any current standard means of increasing your proficiency in these abilities beyond trained rank.

Your level 1 proficiency rank for spell attack modifier, spell DC, and class DC is one rank less than the highest proficiency rank you can get, to a minimum of trained.

Your level 1 proficiency rank for armor and unarmored defense is one rank less than the maximum you can get (so most casters would be trained, most martials would be experts, and Monks and Champions would be masters).

Remove any current standard means of increasing your proficiency rank in anything except a skill beyond trained rank. You still get improvements to your saves' degrees of success, and Canny Acumen could still boost your Perception or one of your save proficiency ranks from trained to expert (and nothing more).

Leave skill increases unchanged, so you increase them to the same ranks at the same levels.

Increase the Hit Points you get from your class at each level by 5. Increase the prerequisites for resiliency feats accordingly.

Remove attribute prerequisites from feats. When appropriate, include skill proficiency prerequisites instead, and don't grant training in those skills in the dedication.

Give classes that currently have Intelligence as a key attribute 7 extra trained skills at level 1 (or less down to a minimum of 4, if you're feeling stingy).

Change the weapon specialization track in the following way: if you have expert or better weapon proficiency, you gain a starting feature at level 1 that adds 4 to your melee weapon and unarmed damage rolls, increasing to 7 at 7th level, 10 at 15th level, and 13 at 20th level. You deal additional damage with your ranged weapon and unarmed attacks at those levels equal to 0 / 2 / 6 / 6. If you have legendary proficiency, increase the bonus by 1 at 7th level onwards, or by 2 at 15th level onwards, and if you only have trained proficiency (i.e. you're a caster), the additional damage is just 2 at 13th level.

Remove fundamental armor and weapon runes, save for weapon potency runes, which also add 1d6 to your weapon's damage rolls instead of giving you an item bonus to attack rolls (and unlike striking runes, these oughtn't count as weapon damage dice to avoid buffing things like the fatal trait). Weapon property runes that give you bonus damage instead convert this bonus damage from one of your weapon potency runes into their damage type.

Remove the base item bonus to AC, Dex cap, check penalty, Speed penalty, and Strength requirement from armor and equivalent items. Unarmored items that would normally accommodate armor runes instead automatically let you etch one armor property rune from the start. Light armor lets you etch two armor property runes instead, and medium and heavy armor lets you etch three. All heavy armor gets the bulwark trait, which gives you a +1 item bonus to AC, but a -2 item penalty to Reflex saves and a -5-foot penalty to your Speeds. Change the Kineticist's Armor in Earth to give you bulwark armor immediately, and change Hardwood Armor and Metal Carapace to just give you a (better) shield.

Whenever a weapon or armor specialization effect, or a similar effect such as the Kineticist's critical blast effect, would refer to an item bonus, just use +3 instead.

This is a bit of a laundry list, which is already a point against it, but just to pick a few examples:

Let's say you're a Ranger:

You're a Strength/Dex class with up-to-legendary Perception, so you start off a master.

You're a Strength/Dex class with up-to-master Fort saves, up-to-legendary Ref saves, and up-to-expert Will saves, so your Fort and Will saves remain unchanged while your Ref saves go to legendary.

Your starting skills remain unchanged. If you increase your skills, you might end up getting higher modifiers on skills that would otherwise not use your key attribute past certain levels.

You're a Strength/Dex class with up-to-master attack proficiencies, so you start off a master in those (and thus get a comparable starting modifier relative to now).

Your AC proficiencies go up to master, so you start off an expert in those for the same starting AC.

Your class DC goes up to master, so you start off an expert in class DC.

Effectively, your essential proficiencies would remain consistent, with your Perception modifier starting off more pronounced, your class wouldn't get the same proficiency and item bonus bumps they do now, and their progression in those stats would instead just be pretty consistent. You wouldn't be able to boost Intelligence as a fourth stat for more trained skills, but your Outwit Ranger would be able to get between a +2 and a +10 to all of their edge's boosted checks against their prey.

Now, to pick a caster, let's say you're a Wizard:

You're an Int class with up-to-expert Perception, so you remain trained.

You're an Int class, so your saves remain unchanged.

Your additional trained skills go to 9, 3 more than you'd normally start with.

You're a caster, so your attack proficiencies remain unchanged.

You're a cloth caster, so your AC proficiencies remain unchanged.

Your class DC is perma-trained, and stays that way.

Your spellcasting proficiency goes up to legendary, so you start off a master (and thus get a comparable starting modifier relative to now).

Even simpler here, you'd stick to your basic proficiencies and would just get lots more trained skills to play with by virtue of being an Int class. As a bonus, your spell attack would be just as accurate as any typical martial class's weapon attacks at all levels (like the Ranger's!).

Now with this entire wall of text out of the way, let's hear what you think!

[Helmic]

I'm gonna have to sit down and read and think through your specific changes in detail - I definitely get the qualifier you open with, because a genuine benefit to not having attributes is that attributes themselves introduce a ton of complexity into hte system and a system without them would be simpler, but if you have to homebrew them out you're instead introducing a ton of complexity into the system in order to cut them out.

But in doing so you kind of demonstrate how little depth attributes actually introduce into the system, because the assumptions of how D&D worked kind of made attributes a non-choice. You max out your attack stat, you boost your defensive stats (saves and AC), and then the leftovers are about whether you can carry your own gear or roleplay as a slightly smart or slightly gregarious person, with a handful of classes and subclass options having to be designed from the ground up with the assumption that their attack stat isn't the same thing as their key attribute as a really wonky way to say they have slightly less accuracy at certain level ranges than normal in exchange for some added benefit.

From there, attributes then severely constrain what archetypes will work together, which IMO is a bad thing generally and there's better ways to prevent combinations that are genuinely problematic. We actually saw Paizo seem to recognize this during the remaster when removed tradition-specific spellcasting proficiencies, this is still using attributes to restrict certain combinations but like the trend is towards permitting more combinations rather than restricting them.

And, again, a huge chunk of the book is dedicated *just* to generating your attribute array, when there's maybe 3 or 4 arrays players are likely to ever use. A lot of how the ABC system was designed was mostly a way to force players to have a range of numbers on their sheet so that they have a range of things that they are great, good, OK, and bad at.

The main thing attributes do that can't be done by just baking their numbers into the classes themselves is their interaction with skills, it's a way to improve a suite of skills. And I think that could be done so much better by just making that into a new feat category - take the Imposing feat to boost Athletics and Intimidation for someone that tries to scare people by being ripped rather than having a way with words, or Studious to boost Arcana, Religion, Nature, and Occult without that necessarily making you better at Society. And so on and so forth, grouping skills in ways that make sense as a personality for a particular character without the current restrictions make many members of hte same class take all the exact same suite of skills because those are the ones that share a skill with what the class uses, without Recall Knowledge having to contend with the fact that virtually everyone that doesn't key off INT dumps INT.

None of this I think would work out as homebrew for 2e, but it does make me wish that if/when we get a 3e Paizo seriously considers dropping attributes in favor of baking the numbers classes are supposed to have directly into the classes and giving us a more meaningful customization option in its place.

[Teridax]

What you describe are exactly the reasons why I want to see attributes removed in a future edition of Pathfinder -- and exactly the reasons why removing them in PF2e would be more trouble than it's worth. If you're interested, I've taken the above and refined it into this brew, which covers a few more bases and also shows how class stat blocks would look like under this model. This is actually part of what spurred my Druid thread some days back -- when you look at the class's base stats under the above implementation, they really have nothing special, because their proficiencies are brought to their late-game stage rather than the early-game stage where they have a major statistical benefit. The Druid's unique statistical benefits mainly come online in the early to mid game, and vanish thereafter.

But yeah, I have to say that attributes feel like a constant wrench thrown in the works of an otherwise really smooth character customization system. Many classes can't feel like they fulfil their fantasy effectively due to being pulled in too many directions, like the Inventor or Magus, and when character options don't work on a build, it's usually due to some clash in required attributes as well. I homebrew a fair bit, and one of the most consistent issues I've run into homebrewing for PF2e is trying to massage attributes into whatever I'm implementing. For instance, I'm currently working on a Skald class archetype for the Bard, and I feel like I'm going to have to implement some mandatory math-y feature to compensate for the fact that the archetype will likely want to be able to build Strength and Charisma on the same character, but not necessarily use heavy armor to make up for their resultingly awful Reflex saves. Were it not for attributes, and if character abilities came purely down to proficiencies, features, and feats, all of this would be much simpler and smoother to work with, and builds would be able to focus on customization options that actually do something, instead of patching up gaps in the system's math.

[Unicore]

My idea for a game with out set attributes is to replace the attribute system with a new pool of things that work like feats, but are something more like "Characteristics." So let's say a starting character got 4 (or x or whatever) characteristics, they could choose any 4 they want, and then they don't have to stay so static and boring. Like there could even be 6 of them called "Strong, Dexterous, Tough, Wise, Smart, Charismatic" but it would potentially be possible to get some of the same combat bonuses from different current attributes for characteristics (or traits) like "scrappy," or "stone-infused" or other more fantastical ways of being.

[Teridax]

I quite like the idea of letting adventurers start with some kind of unique characteristic, though I feel we could also do way better than just trying to copy the 6 attributes. "Strong" isn't really a standout characteristic when a whole bunch of characters are really strong, and that characteristic can be mostly represented even if PF2e with just Athletics proficiency and the Hefty Hauler feat. By contrast, a characteristic like "stone-infused," as you mention, is much more distinct, and I think comes closer to 13th Age and the One Unique Thing each character gets. It could be fun to explore what those kinds of unique characteristics could be, how they modify your character in unique ways, and how they'd work as an entire range of options to let party members begin with.

[Unicore]

I figured some very basic generic ones that could cover very common tropes of the genre and not make it feel like anything is being lost would help settle the sticker shock of no attributes. Strong would be like the equivalent of investing all relevant attribute boosts into STR, giving an appropriate melee accuracy bonus, damage bonus and ability to carry more weight (which will have to cover armor str requirements), but scrappy might give the accuracy bonus without the damage or carry benefit, maybe boosting hp instead.

[Teridax]

Trying to replicate attributes in a game without attributes is something I'd want to challenge, though, especially if it means just layering mandatory stat boosts back on. If starting out with the "strong" characteristic meant your melee attacks were more accurate and dealt more damage, every melee character would pick that characteristic. That to me doesn't really register as an interesting choice, or even all that much of a choice if not picking that characteristic made your character worse at their core functionality.

When it comes to depicting basic qualities like strength, we already have feats and proficiencies as examples of how that can be done in a more nuanced way: Hefty Hauler, for instance, does generally indicate a strong character, but specifically in the sense that they can carry more, whereas proficiency in Athletics suggests a strong character who can make powerful leaps and push people around. When a character hits particularly accurately or hard, that can already be represented by their Strike proficiencies, weapon specialization, and class features like Rage, and because these kinds of things are considered mandatory for certain classes, they're part of the base package. Moving beyond attributes in my opinion ought to mean focusing on what makes characters special: it's not enough for an adventurer to be strong, because lots of people are strong, but if they can jump freakishly long distances or punch walls so hard that they break, that is when they've got something less out of the ordinary going for them.

[Helmic]

Yeah the entire purpose of this is to remove the trap option of not taking the thing that increases your accuracy, anything that is giving the player the option to screw themselves on accuracy without any genuinely worthwhile tradeoff (ie, the tradeoff between a fighter's +2 and a rogue's sneak attack) is antithetical to what we're talking about. I don't know how many times I've had to sit and argue with players who do not understand that PF2e is not D&D fifth edition and that hte game expects them to be at full accuracy, and some number of people take that as having control of their character wrested from them.

I would rather bypass that entire argument and just not have that be an option, because usually what that kind of player actually wants when they roll a rogue with 16 INT and 12 DEX is to be able to say their character is smart. They don't care about the numbers behind it, they care about the label the numbers are applying to their character, and if I could have a system where they could be a smart rogue without it fundamentally underminining the class's ability to function I'd be over the moon. It's really the one way you can make a genuinely bad character concept in 2e, misallocating attributes, and the number of times I've had to deal with someone stubbornly insisting that it's their character and they get to do with the attributes as they please and then they get frustrated at the system when they can't ever do anything because they're missing all the time and dealing less damage and failing their skill checks is just so, so much more than the number of times I've seen a player do anything particularly clever or interesting with the attribute system. Which isn't hard, because the latter has happened zero times.