Percentage reduction also makes healing more effective and encourages threat/tanking mechanics.
Which is imo a lot better than "just do more damage lmao" that most games with a holy trinity fall into.
"reduce damage by 5% but move 5% slower" - or worse, "reduce damage by 5%, with no downsides" - then I don't really see much point.
Videogames are videogames. I was basically just going to say what
AW8 did.
I've thought a bit about RPG systems where every stat has both and positive and negative effect as it increases. In an ARPG, I think one option would outclass every other (speed) just because of how those games tend to play, so I think you'd need to make these qualities more abstract to keep some reasonable level of viable builds (this is less readable though and would probably cost widespread appeal.)
TB RPG's don't run into the exact same problem but it is kind of similar. In TB RPG's, I think the best stats would depend very heavily on specific encounters or bosses or whatever and it doesn't seem very plausible for a game to be so meticulously crafted that every option has roughly even value. Moreover, how would it be engaging on either end? If you're statted for a fight, you can basically cheese them and if not, you run away if you can? Or you go through an extremely grueling fight.
To tie this back into the general problem of armor types and calculations, it runs into both of the aforementioned problems. When you have the option to have either of the stats, one is usually far more advantageous (with percentage, it's usually because of numbers bloat while flat is usually based in it better fitting the game, I feel) and it creates iffy balance with encounter design. The latter is something most people can easily brush away like "bringing a fire wizard to fight a magma golem" but unlike being a fire wizard, there's no real inherent playstyle to being statted for percentage or flat reduction (ideally, a fire wizard should have a different playstyle to an ice wizard and not just being a different type but this isn't always the case.)
Something I haven't seen anyone else bring up is that different damage reduction and penetration mechanics fit different class fantasies even if the ludonarrative implications are a bit silly. Flat penetration for assassins, flat reduction for squishies, percentage reduction and penetration for tanks. So these kinds of stats are class-restricted or role-retricted (to the holy trinity.)
A couple of things should be noted though- no ability should ever just do something on a percentage alone because that's a nightmare to deal with balance-wise, at most it can be a flat damage/heal/whatever with a low percentage modifier.
Percentages should be kept at a "reasonable range" (let's say 5-25%, for example.) This makes percentage reduction/penetration weak in the early-game but that's an important part of the progression curve in typical RPG's, not a design deficit. Tanks *should* be weak in the early game, otherwise most of the challenge goes straight out the window. Early levels are feast-or-famine most of the time, tanking isn't and shouldn't be a factor. If you have early game abilities with percentages that have a real influence, it will quickly balloon out of control (unless you do the incredibly unintuitive solution of it being a percentage that decreases with level but then it might as well be a flat increase.)
Both of these apply to the next proposed solution: have one form of damage reduction be supplementary to the other. The original Fallouts are the closest example I can think of (even though they're not derived from the same base stat), regardless, it stands that if you have flat reduction and percentage reduction on armor, flat reduction should probably be applied first as to not make percentage reduction disproportionately useful (unlike in New Vegas.) While I'm one of the people who'd want both to be in the game because I like playing around with numbers, it seems very hard to make work right and almost seems like it's more trouble then it's worth.
If you want both in some form, I'd pick flat damage reduction (I think percent based reduction, almost by necessity, needs to be more of a supplementary mechanic) and have some percent based reduction on attacks past the damage threshold at different tiers, have some stance-dancing between percentage or damage threshold, or have "blocking" an attack make it reduce attacks by a percentage based on your damage threshold at the loss of the damage threshold functionality (you take less damage from big attacks but more damage from attacks under your damage threshold because you didn't outright block them.) I think systems like these avoid the problems of "problem 2" by avoiding the problems of itemization but still presenting the same kinds of choices- however, it's more involving in battle-by-battle decision making as opposed to win-or-lose scenarios based entirely in character building.
