I think this is a good idea, but to make it a little less sucky for frail races, we could say that even morgia and such can raise potential maximums to, say, 18.

I agree with this suggestion. Attribute maximums are currently so meaningless that they almost qualify as an unimplemented game feature.

Characters who start with high current values, cannot reach their maximums.
Characters who start with low current values, can easily exceed their maximums.

I agree with the general idea that stat training is broken, especially herbs. Herbs are easily available, unlike gems which require a skill to appear in higher quantities. The main reason I dislike the current situation is that most races end up looking the same. Races with high starting To, Wi, Dex are sort of punished because they only get an early game advantage. Herbs are essentially unlimited, unlike gems, and I don't even engage in explicit farming, I just gather some herbs when I'm passing by and remove a patch if I see an easy to create 2x2 square.

In bigger picture, for me the problem is that stat limits are absolute, not relative. Not just with herbs, but also with corpses. The benefit of stat herbs and corpses is variable, it's (25 - starting stat) with a minimum of 0. One possible change is to cap corpses and herbs in a different way - cap the change, not max value. For example, a character could only gain +5 To, +5 Wi, +5 Dex from herbs and corpses. Eating more would have no benefit. Farmers would have these caps at +7 To, +7 Wi, +7 Dex.

I agree with the general idea that stat training is broken, especially herbs. Herbs are easily available, unlike gems which require a skill to appear in higher quantities. The main reason I dislike the current situation is that most races end up looking the same. Races with high starting To, Wi, Dex are sort of punished because they only get an early game advantage. Herbs are essentially unlimited, unlike gems, and I don't even engage in explicit farming, I just gather some herbs when I'm passing by and remove a patch if I see an easy to create 2x2 square.

In bigger picture, for me the problem is that stat limits are absolute, not relative. Not just with herbs, but also with corpses. The benefit of stat herbs and corpses is variable, it's (25 - starting stat) with a minimum of 0. One possible change is to cap corpses and herbs in a different way - cap the change, not max value. For example, a character could only gain +5 To, +5 Wi, +5 Dex from herbs and corpses over the course of a game. Eating more would have no benefit. Farmers would have these caps at +7 To, +7 Wi, +7 Dex. This would preserve the initial variety in stats among races, it would be only slightly diminished.

I've given this a little more thought, and come to the conclusion that the attribute potential maximum system is working as intended. Its goal is to slow down natural training, and that is precisely what it does, by a factor of 2 or 3, when an attribute reaches its "maximum."

The problem is the relative ease / difficulty of acquiring natural training, compared to "magical" training; and also relatively among various attributes.

I think it's fair to call herbs a magical means of raising attributes, even though the technical means by which they do so is the training system. If herbs and potions had the same relative availability, then Herbalism and Gardening wouldn't be nearly as interesting. I like their existence, but I still agree with the original poster that it would make for better gameplay (and easily rationalizable) if herbs were not limited to some arbitrary absolute value, but rather respected a character's attribute potential maximums.

The real problem IMHO is this:

Consider the difference between training Mana, Strength, and Dexterity.

A spellcasting class, whose prime attribute is Mana, which helps determine their ability to attack monsters, needs only do one thing to continuously, naturally, train Mana up into the 40's and beyond: cast spells. This comes naturally during the course of attacking those monsters, and while doing all the other things a caster naturally does by playing the game.

A melee class, whose prime attribute is Strength, which helps determine their ability to attack monsters, cannot train Strength naturally beyond 18.

A ranged class, whose prime attribute is Dexterity, ... ..., cannot train Dexterity naturally at all.

In summary:
Caster: Mana; trains into 40's during natural course of play.
Melee: Strength: cannot train past 18.
Ranged: Dexterity: cannot train.

IMO it is nonsensical (broken, actually) for weight training to be limited to an absolute value of 18, when the statuses Burdened, Strained, Strained!, and Overburdened! are themselves relative to the strength of the character. Is it impossible in the real world for a strong man to exercise his muscles by lifting weights?

I would also suggest that firing missile weapons of any type ought to provide natural training for Dexterity.

The current implementation of attribute potential maximums works great for slowing down natural training that is provided by spellcasting, weight training, and all other means of natural training. But the availability of that natural training is grossly imbalanced, and in some cases simply missing.

P.S. I realize it's also possible that we, the players, are simply unaware of the intended means of training Strength and Dexterity, and that no one in all these years has stumbled upon a reliable means of training Toughness, etc.

Pim: You can train Toughness from very low values by remaining well fed. Doesn't really change your point, as you can only get at about 12 or so like this. And I believe Dex is trained by avoiding undetected traps and by certain types of missile weapons.

Toughness can be trained by taking damage and leveling up if it's below 10. (I rolled a character with To 1 where this showed a few days ago.)

Dexterity can be trained by catching returning weapons and not falling into pits.

All stats receive a training boost at level up, and very low stats are more likely to increase, IIRC. Nothing to do with taking damage or being well-fed. Athletics improves chance of this.

I have never found any training effect related to catching thrown weapons. I've seen a lot of people say this, but I've never seen any actual evidence to that effect. Or it only works at very low values of Dx. Maybe I need to pull out the 1 point in all stats character and run some tests...

Toughness to 10 - another absolute value, which is fine as far as it goes, i.e. that's by design and although it leads to characters all looking the same, that's ... by design.

Dexterity training by those methods is not sufficient to overcome the 1000 pt hurdle, and is therefore only supplementary. My evidence for making this statement is purely anecdotal: I had an elf with dexterity 18, and spent several hundred turns walking back and forth over a pit, falling about 20% of the time, and did not get a dexterity increase. This was 1.1.1 iirc, and not even close to being a statistically valid test. (I wanted to raise dex in order to have better attack values with a whip.)

At any rate, spending the entire game walking over a pit or using only boomerangs is unfeasible nonsense, compared to a caster or melee class's option to "just cast spells" or "just be Strained!".

As it stands, only caster classes have a means of naturally training their primary attribute. Melee can only go up to 18 and ranged classes get squat.

P.S. It goes without saying that I love the game even in its current incarnation, and that my comments are attempts to be constructive.

Interesting idea, however the problem is that herbs are also useful for retraining stats depleted by monster attacks. This would needlessly punish that use. The goal is to maintain racial differences, not to punish attack recovery.

How about going with the "change cap" idea, but cap the change in potential? E.g. you could raise potential via morgia as you do now, but you could only do it by, say, 5 points. Then if your stat gets drained, you can retrain it. And you'd have the desired difference between the races.

I think that's a good idea if one wants a compromise solution. Although, to be honest, I'll probably set the limit to 0 points rather than 5 and go with the suggestion in the original post, which is simple to implement, straightforward for the player to understand, and good.

Anyway, most of the solutions that are being proposed here look fine, then it'd all be a matter of fine-tuning for balance.

I'll try to brief.

It's wrong that my orc beastie who started with Le at 6 is going to be a decent spellcaster towards the endgame, even with the spell learning bug.

Garth's steroid gym must go, no training of stat potentials from him, $*0d10, or whatever it is. Each spin on a casino machine should use up time/energy, although it doesn't make sense that a faster PC can make a machine spin faster.

Herb bushes must die (have a lifespan). New herb bushes type must be randomized upon spawning. Gardening may yield seeds when applied to herb bush.

Stat potentials can only be raised by potions. No training, corpses, herbs can raise it. Example: PC's Strength is 16/16, eating a corpse will not raise it. Drinking a potion of Strenght will.

If these make the game more difficult, good. Let orcs be orcs and hurthlings be hobbits, not everything a super intelligent, archmage, nimble Greater Titan thing.

The rub is that it doesn't necessarily congrue with the stated intention behind the potential system in the manual. The manual suggests that exceeding the maximum potentials through "natural" means is rare and difficult, not flat impossible. So maybe make it take 10-20x the gold to raise the potential by 1 point as to raise the stat by 1 point. Combined with nerfing the casino (so that it's no longer possible to bring millions of GP to Garth), that should be good, I think, in making it agree with the game plan. The herbs are supposed to be magical, I believe, so allowing them to boost potential by a few points, as suggested in my post, would seem to agree with the intended system, which was that "magical means" could exceed the limits with ease. And being able to buff your toughness, etc. by just a few more points over the potential (vs. fixed-limit caps that can be defeated anyway with scum training) isn't going to obliterate the race differences, which is the point here.

Also, changing the herb growth system would seem to punish the ability to use them for food and healing. I thought the idea was just to keep the character from getting too far "out of his race", so to speak. In that case, then simply limiting or removing the potential boosts would suffice. No need to punish food/healing. If the stat potential increase was limited or removed, I fail to see any other point to dinking with the herb growth system other than punishing food, healing, and ability to recover from wight attacks.

Just to add, if herbs are set to not raise stat potentials, then scum training won't raise them. With regards to my suggestion, one would still be able to gather and use food/healing herbs, just not enough in a short period using script-macros (or your fingers, which can drive you insane) for easy piety.
With current herb availability it is simply too easy to raise Wi,Dx,To. This is why I'm suggesting those herb bush changes.

However, with it capped by the potentials, then why do the bushes need changing? Sounds like you're not just after racial variation but outright making the game harder, like it isn't already hard enough as it is.

After further investigation, I concur. It will in most cases make very little, if any, difference in the gametime a pc needs to have it's relevant stats reach 25 or potential value. Which is not the point of this discussion anyway.

Those suggestions will be removed from my original post.

In my view "natural" would exclude: potions, corpses, herbs, gems, pools.
And include stuff like: digging, casting, catching, training with Garth etc.

Changed the rate at which potentials can rise. Also adjusted the way in which training affects attributes. It will be interesting to see the results.