Okay 99% worthless ;)

Those mechanics aren't always implemented.

I don't remember Dungeon Hack implementing carrying capacity, for example (even though the game's ancestor Dungeon Master did), though it did have an underwater level in each dungeon with that setting enabled (but I don't know how it handled the need for air, except that the Water Breathing spell is in the game and functional).

Underwater areas are actually quite rare in D&D-based games. I've heard that Dark Queen of Krynn apparently has underwater combat, but aside from that game and Dungeon Hack, that's about it to my knowledge. (Did any of the Eye of the Beholder games have any water levels?) I actually like having some sort of soft caps or diminishing returns (or reachable hard caps) when it comes to ability scores, particularly when they're improved through points earned at level up; it keeps overspecialization from being too dominant a category, and it turns late-game growth from getting way too good at one thing (which could break things) and into allowing characters to learn new things, even at that stage of the game.

There's, of course, multiple ways of doing this:
* Soft caps: Stat stops giving improvement after a certain point. (I think hard caps might be better here, where the option to raise the stat past that point is not present in the first place.)
* Diminishing returns: Need more of the stat to see an increased benefit. Alternatively, the cost of increasing the stat could increase (see Avernum 1-6, but not the later remakes), or for non-XP based systems, it might take more effort to gain one stat/skill point (most SaGa games, Wizardry 8 with skills, Elder Scrolls (from Daggerfall onward) with skills). If there's something like a skill point system, it would help if the game would communicate that detail.
* Hard caps: If you put all your points into one stat/skill, you can't put any more in, and must spend them elsewhere. Dragon Quest 8, as flawed as that game is (it's one of my less liked DQ games), at least does this; you can't raise a skill past 25 + 2 * level, and can't ever raise the skill past 100 (when you learn the final ability associated with the skill). (The thing I *don't* like is the skill point system where you need to level up to gain skill points, and the lack of any way to respec.)

One annoying thing is that the HP gains are not retroactive, so a character who waits to build Endurance (what the stat is actually called) ends up with lower health in the long run. I would much prefer it if your health didn't depend on when you decided to train Endurance, just that you did that at some point. (With this chance, raising Endurance early would still be a good strategy, as having low health early in the game would put you at a disadvantage; this change would just mean that Endurance gained later on would be just as valuable as Endurance gained early. Also, no other stat works like this.)

By the way, Arena uses the same formula as Daggerfall, though I note that, in Arena, temporary Endurance boosts affect HP gains. (Don't know whether Daggerfall works that way or not.)

In Battlespire, Endurance duesn't affect HP gains; rather, HP grows exponentially after you clear each level. (Note that the HP stat is actually called "Wounds", which makes no sense, just like the "Fatigue" stat name that wasn't fixed until Skyrim.)

Skill points gained per level up in Wasteland 2 depends on the character's intelligence attribute. The minimum points are 1 (2/level), 4 (3/level), 8 (4/level), and 10 (5/level). Anything between those gaps is an attribute point that could be used elsewhere.

Much like D&D 3.x where you don't get additional retroactive skill points for getting higher intelligence, meaning you want a really high intelligence as a rouge first and then other stats later in order to get more points for actual abilities.

Curiously the same problem appears in the D20 KOTOR games. As well as NeverWinter Nights

Fortunately, Pathfinder fixed this. It also fixed the issue where the first level class contributes more skill points than other levels; you don't get extra skill points at 1st level, but instead you get +3 to class skills.

You need to put 1 rank in to get the +3; but that only removes partial confusion verses max ranks of 3+level and 1/2 level bit, which made cross skills useless/pointless to get.

Curiously pathfinder also took the +5 to a skill feat, and made it +3, then when you put 10 ranks in it, it upgrades to +6.

The int skills being retroactive seems to be A potential unofficial ruling mostly because it probably wasn't clear enough the first time. Still it's meant to be so yes that does improve it. A little less juggling to do.

When was that feat (Skill Focus) +5? I distinctly remember that feat being +2 in D&D 3.0 and +3 in 3.5. (Did anyone take Skill Focus in 3.0 other than as a pre-requisite?)

Hmm i'm not sure, could just be remembering it wrong as i'd only touched 3.5, or it's specific skills that get such a bonus. Or maybe the +5 i see in various places just stuck with me (+5 BAB, -5/+5 Power attack, etc)

Looking it up, i see references to +3 and +2/+2 (agile, alert, acrobatics, etc), and +10 for epic level. Even a +6 but it looks like it would be a feat on top of another feat replacing it, which doesn't count.

That bonus may be too big.

Remember that you're rolling a d20, so a bonus of +10 is more than double the chance of success, assuming you're in the range where the die roll actually matters.

One problem with d20 and similar systems (basically any system where you roll a die, add a bonus based on the character's abilities, and have to reach a difficulty class), is that it doesn't scale too well. As the numbers get higher, you end up in the situation where, excluding automatic success/failure rules, every action is either guaranteed to succeed or guaranteed to fail; this means that, for example, AC becomes pointless, as attacks are either going to hit or going to miss regardless of the target's AC (and a similar situation applies to attack rolls).

The problem gets even worse as numbers get bigger. With epic level rules, around level 4000 a 1% level difference can mean the difference between missing except on a 20 and hitting except on a 1.

(Contrast this to something like Disgaea; while high level gameplay isn't exactly balanced, at least small differences in level or stats don't have that drastic an effect.)

Maybe. Epic level skills/feats/spells do epic level things. I heard multiple times levels that are 14+ tend to be heavily unbalanced.

Then again people would want to be able to actually do the impossible, walk on clouds with a mere 110 DC check on dexterity. Which i don't see being workable regardless how you slice it.

Which takes another of the silliness out of it, a lot of DM's won't do epic level characters and games, thus getting anything like that is far less likely. Most like the 5-10 level range, which is workable but it means half the material will never be of use from the books. More than that, getting to 20th level was a vague conceived idea in the original D&D and level 10 i believe was usually the cap of powers, while gods were probably level 30 and had an AC of -10 to -20 (using Thac0).

Then it's also said mixing the D20 which is a tactical system with skills was never a good mix. I don't know, i can only think WOTC was so desperate to make money on a purchased property they didn't think that far in advance and left huge gaping issues when they did 3e, and are paying for it since.

[ttRPG] There's a broad spectrum there. Some people prefer the random roll dominating the result. Others prefer the character's statics dominating the result. There are whole diceless systems that don't even randomize. Even within dice rolling systems, some use a single die (like the famed d20), or others use a bell curve (like Fate's 4dF).

I'm not surprised you called out a preference for the random element. Most games that favor the skill over the random are the less "gamist" and more "narrativist" style games, like Fate, or Amber. (Or in cRPG terms, Disco Elysium... again very narrative-strong, game-weak. And strongly values the character's stats over the randomized element.)`

Actually, it's not the randomness versus deterministic aspect.

It's more the fact that, with the random portion of the roll to determine a binary result not scaling, you get to the point where stats like attack bonus and AC become useless unless their relative values are "close enough", where "close enough" is a measure that does not scale. As a result, you get to the point where it's not worth raising these parameters unless you can get the value high enough; an AC of 2647 is no better than an AC of 10 when the enemy has +2683 to hit. (This is assuming higher AC is better, and that to-hit is handled like in the d20 system.)

There are approaches that would scale better at such high levels. For example:
* Have the hit chance be accuracy *divided by* evasion, perhaps with a constant factor. This way, small differences (relatively speaking) don't have such a disparate effect, and an AC of 2647 would be a lot better than an AC of 10, even in a situation like what I described above (provided that the scaling factor is less than 1).
* Have to-hit bonuses and evasion not scale. Instead, have damage numbers and damage reduction, along with HP, be what scales. (Also don't forget to have healing scale adequately as well.) Something like how Final Fantasy 5 does this would work well, I think.

FF5's approach is this:
* Each weapon or other attack has a hit chance. For some weapon types, like swords and daggers, the hit chance is 100%; for others, like axes and bows, the hit chance is lower. To hit, the attacker must first succeed on this roll.
* Some targets have the ability to evade. (In FF5, evasion is granted by shields if it's a party member; some enemies can evade as well.) If the attack is rolled to hit, the target can still roll to evade (though, with some weapon types, like daggers and bows, the chance to evade is halved). Note that the only way to prevent evasion entirely is to use an attack that's guaranteed to hit, like the Aim ability of the Archer job.

Dungeon Master, which I've been playing lately (along with Chaos Strikes Back), has some interesting mechanics here as well. When an attack is made, there's actually multiple chances for the attacker to hit. (Only one of these needs to succeed.)
* An attack roll that takes into account the character's dexterity and load and the target's evasion (what you might typically expect).
* A 1 in 4 chance of the attack hitting anyway. Hence, there is a lower bound to accuracy.
* A random roll based on the attack mode (like Swing, Stab, or Chop) being used.
* A random roll based on the attack mode and the hidden Luck stat. (Note that Luck increases when you fail and decreases when you succeed. Interestingly, some SaGa games use a similar Luck mechanic when it comes to stat growth and technique sparking checks.)

(Note that DM's system does have the "narrow window" issue, but with the other checks, combat at least doesn't degenerate into the situation where neither participant can hit the other.)

The issue here is that, if the character's bonus isn't reasonably close to 110, the check will either fail (and boosting it won't help if it doesn't get close enough to 110) or succeed (in which case increasing it further brings no benefit).

I'm sure I've mentioned it before, but instead I'll shift over and mention the Sims. Where it seems 0-3 are complete buffoonery, where even you or I could likely figure out how these things work instead of somehow burning waffles. 4-6 are average competency, while 8-10 are superhuman abilities. Meaning all numbers between those are worthless filler.

What does 7 indicate?