So go roll a starting GE Wizard. With potentials usually close to 40 in Le, Dx, Ma and Ap, at 90% that you mentioned, you're looking at a wizard with 30-35+ learning, dex and mana.
You will read the initial spellbooks and get enough castings to last you for the rest of the game. By level 10, you will have 250 PP.
Way to go, OP wizards.
Similarly, a trollish barbarian will start with typical 40-45 St and To. You will slash through early and mid and late game one-hitting everything with a simple dagger 1d4.
Dark elven archers will start with ~25 Le for 6 points per level to increase all those nice skills like alertness, find weakness and archery. You will have a 1-shotter archer by level 10, with that starting 35 dex.

You see where I'm going with this?
This isn't the way to go. Many r/c combinations that have synergies over the same attributes and this makes it so important to select the right race for the right profession.
Without extensive rework of attribute system, their potentials and starting values, you can't possibly consider introducing changes like these.

I also understand and agree with what JellySlayer says - more differentiation between races is attractive and makes sense from the role playing point of view.
This forces players to play their r/c combinations differently, with respect to their parameters.
The game benefits from hard caps but only if they are REALLY hard caps - i.e. you can't raise potentials with herbs; only respective potions and training for gold with Garth can do that.
It takes time and some micromanagement, as well as a dose of luck. Makes sense.

I generally still agree with SirTheta's approach of class having a bigger impact on starting attributes, but perhaps it's a good idea for race to be the only determining factor on potentials (not actual attribute values, which should only be determined by class).
Some stats like To and St for elves could get a slight increase of possible starting potentials at the cost of lower overall pool of points.
This way you cover situations like mist elven beastfighters or barbarians not having enough of those to justify their choice of profession.

You did realise that I'm not proposing my approach applying on top of the current system, right?

In fact, I'd suggest that base potential for Grey Elves in Learning would be around 25 or so. Even with maximum genetic randomness and the Wizard boost, the Potential only gets to 33 from a base of 25. The highest the actual starting Learning would be is 31 (before the opening question system), and the highest potential would be 35. And that's based on my initial guess, and would naturally be subject to tweaking (for instance, it could be a base of 22, or the max genetic randomness could be 1.1, or the Wizard could just get a multiplier of 1 instead of 1.1). I intentionally chose big numbers to emphasise the effect. If you dropped it to 22, randomness of 1.1, and wizard multiplier of 1, the highest you'd get would be a potential of 26 and stat of 23.

Aielyn: I still disagree. The problem with your system is that it requires a complete and total reworking of how stats are determined - which is a lot of damn work! It's the wrong solution to the problem for that reason. It's also the wrong solution for other reasons - you're either going to require starting stats to be higher (bad idea) or you're going to lower potentials across the board for melee classes and make the game even harder for them (bad idea). The problem is not that "classes need to have max closer to their potential" - all this rhetoric about being "trained" in a stat means you MUST be closer to your potential is just baffling to me. TRAINING can and does increase your potential in-game. Swamp hydra corpses, Garth, etc. People are creating this artificial dichotomy between "race" training and "class" training that simply doesn't exist. I have the potential to bench 300 pounds if I put enough training in, even though I've never benched before, and while that IS capped somewhat by my physical structure, it's also capped by how much training I put in.

Training does increase your potential, but mostly only when your stat is at your potential. Hence the requirement that the difference be smaller.

You have the potential to bench 300 pounds. But if you'd spent your entire life training to be a warrior, you'd probably already be benching 300 pounds. This wouldn't mean you magically have better genetics and thus a dramatically higher potential.

And you'll notice that I never said that class shouldn't influence potential. As I pointed out, I'd expect a Dwarven Beastfighter's potential Learning to be significantly lower than that of a Dwarven Wizard, and vice versa for potential Toughness. But I'd also expect the Beastfighter's Learning stat to be lower relative to potential than the Wizard's, and vice versa for the Toughness stat.

If you follow the reasoning I put forward in my examples, you'll see that race and class both influence potential, but from there, the class determines how far away from the potential the PC is - which addresses what you claim is the main issue, doesn't it? Didn't you claim that the gap between stat and potential seemed to be determined by race, rather than class?

You note that it would involve a "complete reworking" of the system. While this is technically true, the "system" in question is probably about 8 lines of code, maybe 16, and the substitutions would be pretty straightforward. The main reworking involved would be selecting the values (like the base potentials for races) in order to produce reasonable numbers through the process. The remainder can be written in pseudocode so easily it's not funny, and would translate to real code even more easily. And a reworking is probably necessary, now that some of the rules of the game have been altered so noticeably. Now that potentials are so much "harder" than they used to be, the system needs at least some tweaking to make sure that it's not overly disadvantaging certain races/classes, anyway. Might as well consider larger-scale adjustment while we're at it.

Yes, but with more training I could bench 500 pounds, 600 pounds, etc. [which is exactly how it works in-game if you have enough gold or low enough St]. Leveling will randomly raise your potential sometimes, instead of increasing a stat, even when it's nowhere close. There's no reason that a certain class enters the chain near their potential except that some people think so - there's very little in-game or physical justification for it. If they're at their peak physical/mental/etc. fitness, why even gain levels? Why does gaining levels then train your stats?

My problem with your suggestion is that it severely disadvantages all melee classes - instead of being able to up their To, they're stuck with a potential close to their starting value. In most non-troll cases, this is below 25 to some extent or another. It's fundamentally unfair to them and actually reduces the differentiation between them and other classes (because differences in potential means everyone can climb to almost the same place).

I'm also approximately 500% sure you're wrong about how complicated stats currently are. Anytime someone thinks changing something in ADOM is easy, they're proven wrong very quickly. There's a lot of complicated interplay in sections of the ADOM code, and this would be no exception. Just off the top of my head, here's some things that complicate stats: classes have a multiplicative effect on stats, you have to take burden status into account when determining strength, various backgrounds affect stats [sickness as a child], etc. Changing the whole system from determining stats to determining potential creates a real headache that is not at all simple, and a complete reworking is asking a lot for very little gain when you can do the exact same thing by just using the current system and factoring in class (because, yes, I agree the system needs tweaking now that potentials are "hard").

Well, no. There are limitations on what the human body is actually physically capable of doing. Nobody will ever be able to throw a baseball 200 km/h. It's impossible. The force required would literally break your arm first. Without special equipment, it seems doubtful anyone will ever be able to bench press more than about 320 kg. These are hard limits. You can't beat them with more training. Moreover, you aren't talking about potential here. If you can lift 300 pounds now, and with more training, could maybe lift 600, then 600 is your potential. The extra training is increasing how much you can lift, not the maximum that you could ever hope to lift.

However, that's not really how things work in-game because potentials aren't completely and totally absolute with no possibility for change - ever (if they were, then yes, I would agree). It is actually analogous to reaching your "potential" of benching, say, 200 pounds (with a modicum of training) then 300 pounds (with as much training as it took you to get to 200) then 400 pounds (with twice as much as it took you to get to 300), etc., etc. (i.e. the old exponentially decreasing progression of work put in:results). You could come into the chain benching 200 pounds with a potential of 300 pounds, but that doesn't mean your potential can't go to 400 pounds (okay, this metaphor is becoming really convoluted).

As I see it, the potential should be the limit that can't be increased by "normal" means. So if you start at St:12 and potential is 15, you could train by carrying heavy stuff up to St:15, no more. Same with herbs (overtraining past 25 can stay imho, but overtraining past potential has to go) and with Garth training. Only way to move past potential are "magical" which means potions. And perhaps corpses, or maybe just some of them.

Well, you're wrong about strength. Getting to St:18 is a necessity for all so they can actually carry things without being burdened all the time (hurthlings, for example, can easily start with single digit strength and it'd be absolutely cruel to restrict them). Any opposition on that front is really just silly. Herbs have already been fixed...

Potential isn't really the limit that can be obtained by "normal" means, no matter what the manual says. Garth isn't magical at all, for instance, you just put a *lot* of hard work (i.e. gold) in, which is pretty "normal". There are legitimate ways to raise your potential (considering p21, where herb bug is fixed) and legitimate reasons to give people the option to do that.

Well I'd say they need more ogres in their diet.

I think that Garth shouldn't be allowed to train past potential, but corpses should allow that. Or maybe if an increase would be due, but potential is in the way, the potential could rise first.

You are confusing what is with what ought to be. Yes, at the moment Garth can train potentials. That doesn't mean he should be able to. I would say that Garth is bugged in the same way that herbs were, and that should be fixed.

I don't think Garth training potentials has ever been the subject of a bug report and this is the first time I've seen people suggest he shouldn't. I never use him, so I could care less either way, but I see no reason he shouldn't be able to train your potentials just like swamp hydra corpses can (he imparts a lot of training, after all).

Again, you are confusing is/ought. The fact that nobody has reported training with Garth and swamp hydras, etc. as a bug, doesn't mean it isn't a bug and shouldn't be reported as such. Exactly what the Creator did to potentials in the previous RFE is somewhat unclear. But here you go.

I don't think I'm confusing is/ought. Should potentials be increased by leveling, Garth, and corpses [I in fact support a more liberal corpse policy than currently exists]? I do think so, and that happens to be the case in-game, currently. They are related but neither mutually exclusive nor causal.

No... but the person with better genetics is probably more likely to have picked to BE a warrior. And the person who naturally has perfect pitch (as far as I know that's genetic) is more likely to take a career as a bard than someone born deaf.

The average troll may be fairly dumb (average potential learning 9 or whatever), and the average gray elf is really smart, but occasionally you're going to have a really smart (for a troll) troll. And that hyper-intelligent troll, with his potential learning of an astronomical 15 or 16 is the kind of troll who is going to set out to prove that trolls CAN be wizards.

I would pick a min and max potential stat for each race, and put classes naturally good at that stat at the top end, and classes naturally bad at that stat at the average or low end.

Which is why I suggested that potentials should be influenced by Class. A Trollish Wizard will have a higher Learning potential than a Trollish Farmer, not just because those with higher Learning potential would be more likely to try to become a wizard, but also because all of that training makes them more capable, thus increasing their potential.

But to increase your potential, you first need to max out your stat. Thus why a class that makes a lot of use of a stat should start with that stat closer to potential... but the potential should be higher. Or would you suggest that the PC spends their life training in their chosen profession, and then spends years just doing nothing before heading to the Drakalor Chain, thus allowing their actual stats to drop?

A person who has gotten their bench to 200 kg is going to have a current "potential" of maybe 220 kg, not of 300 kg. Your assertion basically amounts to say that, because your "potential" right now for bench might be 150 kg (with your max being 40 kg), your "potential" when you're benching 240 kg after a lot of training should be 350 kg. Things just don't work that way.

I actually said back in the original potentials RFE that Garth should not raise potentials.

Anyway, class should not have any influence on starting potentials. Potentials are determined at birth prior to apprenticeship, it is not something that governs profession choice.

I would still like to play individuals that made bad profession choices, so voting no.

I don't understand any of that - potentials being determined at birth is just not true - even in-game your childhood significantly affects your stats and thus potentials, for instance. Going out to the real world - nurture and nature are pretty much inseparable and have significant effects on your "potential" (e.g. childhood malnutrition, not ever being formally educated). The trajectories available to you at birth diverge very widely as you grow up, choose a profession, etc.

I have no idea why this would prevent you from playing individuals that made "bad" profession choices (what does this even mean??).

he means that he still wants to be able to push elves to 18 strength by just carrying a lot of junk.

He might have been responding to me. An anemic human with genetically brittle bones and no family history of physical strength could still decide to be a warrior, and thus choice of class should have no bearing on potentials.

Whereas I say, if I want to pick an individual who makes a bad profession choice, I'll pick a new race, but count on him still being one of the better exemplars of that race for that profession. So an anemic human with brittle bones might not try to be a warrior, but a really buff mist elf might - for the same potential strength (but higher than normal mist elf potential strength)