I avoided going there because any substantive change is going to make the system a PF3. I’d put almost everything you listed as too big of a change to call it a revision. Even though PF1 was quite different from 3e, it was still more or less built on the same chassis.Let's assume Paizo decided to do a PF 2.5 - what changes would you make?
If I were going to do a revision, and assuming I was just iterating PF2 into PF3, I’d make the following changes:
- Get rid of skill actions. If you look at the ones included in the CRB, they follow a couple of similar patterns (succeed → get a result, critically succeed → get a better version of success; set a benchmark for opponents; etc). Paizo recognized this during the playtest for spells, creating the idea of a basic saving throw. There be a couple of standard types of skill checks that the GM can use to adjudicate most situations.
- Make the VP subsystem core. It’s not well-written, but the VP subsystem is powerful. Look at how Blades in the Dark uses clocks. Most of the skill actions that don’t fit in the standard types of skill checks could be handled via the VP subsystem. In particular, this would let them drop the awful Make an Impression and Request actions. Those are holdovers of 3e’s disposition subsystem, which I’ve never seen used in play as written. It just doesn’t align with how people want to use skills in social situations.
- Rename Diplomacy to Persuasion. Intimidate and Deception describe approaches. Diplomacy is another holdover from 3e that could be given a better name.
- Give all martial classes a free martial archetype at 1st level. Casters differentiate themselves by tradition. Martials would differentiate themselves by their archetype. This would also address the complaint that all the martials get variants on the same feats or don’t get access to ones they want for certain concepts. You could then build class-specific feats to further differentiate the archetypes between classes.
- Let the fighter change archetypes. Right now, the fighter’s primary differentiation is being better at hitting things than everyone else. I like to think of fighters as someone who can make anything a weapon in their hands, like the fantasy equivalent of John Wick with a pencil. Find a cool axe? Fighter is axe guy. Find a cool polearm? Fighter is polearm guy. People complain about having golf bags full of weapons, but that to me is what a fighter is. They bring their tools to battle and use the right one for the job.
- Simplify monster creation. Do we really need the ranges of values for every degree of every level of a monster? Make it so GMs can quickly improvise a monster just using the tables. I used the class breakdown by level in the DMG all the time while DMing 3e. It’s way more handy than the useless NPC roster in the PF2 GMG.
- Be more prescriptive about exploration mode. You have a good foundation, so build it up into a full-fledged subsystem. People who don’t care about that stuff are ignoring it already anyway, so give those who want to use it a cool mode to play with. Include things like wandering monster and reaction. That is where your disposition stuff should come into play. This would also help walk back the idea that “encounter” is synonymous with fight. Use the VP subsystem here where appropriate.
- Enumerate expectations when discussing encounter-building and provide knobs for groups that fail to meet them or exceed them. If the game assumes a decent level of tactical play, then groups that don’t will have a miserable time. The same goes for groups that are really good at synergyzing. That helps keep the encounter-building guidelines as a useful tool for assessing difficulty.
- Allow the game to be played without balanced encounters. Rather, make it clear that this is a valid approach. You can do that today. If you’re doing an old-school style dungeon crawl, you can use a suitable exploration procedure that lets PCs control engagement. That doesn’t mean they will win every fight, but the expectation in this style of play is that fair fights is a failure state. This would be a supplement to the expanded exploration mode.
- Pay Justin Alexander to write your adventure-building chapter in the GMG. Paizo has pulled fairly liberally from various sources for its other advice, but I think they’re too steeped in their own adventure-writing culture to see other approaches. Justin’s advice on the Alexandrian is really good for running dynamic, exploration-based games. This would be good for the expanded exploration mode, but it would also give tools to GMs who wanted to run mysteries and other styles of games that weren’t just fighting through encounters to tell a story.
- Get rid of the Incapacitate trait. I’m with CapnZapp. This is the kind of thing that players will remember when it screws them out of stopping a Big Bad. It feels too much like a gacha. I get that Paizo doesn’t want people trivializing encounters, but I think we need to get back to that’s being okay. Let the PCs be clever and win, and include that as part of the expanded exploration mode.
- Look at spell balance. People are unhappy with how casters feel. Mathematical balance with martial characters feels crappy. Figure out a way to let non-casters get cool stuff as they progress that isn’t just hitting things harder, so casters don’t “have” to be balanced against it. You don’t need to resort to stronghold-building or getting an army, but let higher-level play be more than just lower-level play with bigger numbers.
- Make the arcanist core. It’s baffling that they created this class in PF1 then went back to traditional, Vancian-style casting in PF2. If you were going to do full Vancian (meaning you could reprepare empty slots), that would be one thing, but forcing players to choose how many of each spell isn’t very fun or exciting. I know OSE does it the old-school way, and I’m fine with it there because your relationship with casting is different, but PF2 wants you to be using magic regularly. It’s not really a strategic resource, so don’t make players choose.
- Rethink higher-level play. I’m going off my experience with PF1 and how CapnZapp has described higher-level AP play. In a sense, you’re just doing the same thing that you did at lower levels, but the stakes are higher, and you do bigger numbers. If higher-level play is about saving the realm, then it should involve mustering realm-level resources (and not just because you went into the Dungeon of Really Dangerous Stuff to get the magical super-doodad).
- Devise a proper hero point economy. I don’t like how you just earn them whenever the GM remembers to give them out (once per hour). Tie them into certain activities or for introducing certain difficulties into the game (e.g., like compels in Fate, intrusions in Numenera, or invoking a flaw in Open Legend).
- Add more uses for hero points. In particular, make my house rule core because it adds an element of off-turn engagement: you can spend a hero point to increase the degree of success of another player’s result.
- Rethink magic items. I agree with CapnZapp here too. After getting into OSE, the magic items in that game are just fun. If we really need to have boost items, then tie them into the encounter-building guidelines as knobs you can turn. If you want to give everyone +3 items, then this is how it affects the game. If you never give them any, then that is how it affects the game. Get rid of all the little boost items and replace them with fun stuff. Do more things like the bag of weasels (except not cursed).
- Make shields passive. I don’t like the “shields shall be splintered” variant for OSE. I don’t like Raise a Shield and Shield Block in PF2. It encourages PCs to bring along carts full of shields to use as ablative armor, or you never use it because you don’t want your precious magic shield to be destroyed. It also encourages people to be less mobile than you really ought to be in combat.