Though its not particularly directly relevant here (it wasn't a D&D game or even fantasy) I sometimes tell the story about my wife putting together a Mutants and Masterminds 2e character and talking to me about it (we were both players in the game) and her watching as a look of dawning horror came over my face as I realized she'd quite innocently stumbled into a really, really bad corner case in the rules that was going to blow up the first time she used an ability she was talking about taking on her character. I use it as an example of design mistakes having potential implication that reaches well beyond deliberate abuse.
Absolutely. I don't think deliberate abuse is more than like, MAYBE 25% of issues? And that's being quite generous--I'd expect it's closer to 10%, or less. Far, far more issues come from the rules making people
believe X Y Z when none of X, Y, or Z is true--or people thinking game A is built for X Y Z, when it's actually built for L M N O P, and can't even
do Y without houserules; or the GM aiming for a toned-down, down-to-earth, low-power adventures, because they know how to make it do that, and one or more players pick options because they want a power-fantasy wish-fulfillment thing because they know that's what the system does really well. Or maybe the GM is aiming for a high-power campaign and that's why they chose this system, and the player has no idea that that's why it was chosen, and goes for pure flavor regardless of strength, causing the same problem from the other direction.
The more the system presents one face, while mechanically supporting something different, the more likely players and GMs will misunderstand each other. 3rd edition has possibly the single greatest spread between what it
tells the reader it does, and what it
actually does, of any game I've read. They tried to address this later on (e.g. with the "<Class Name>s with Class" articles), but ultimately the damage was done--and Pathfinder inherited that same issue. Mr. Bulmahn communicated this quite clearly when he asked folks to give PF2e a chance: PF1e had hit a brick wall because the fundamental design of the 3rd edition system simply conflicts with what they wanted it to be.
I still maintain that there must have been either very limited blindtesting there, or those in charge of it must have ignored results that seemed off from anything they'd seen in direct playtesting. It seems impossible some of the problems that became endemic later would have remained unseen otherwise.
We've been told, point-blank, that active testing did not occur beyond level 6--
just far enough to test the most basic of scaling elements and no further (e.g. level 6 is when you first get iterative attacks, when you get your third level-based feat, etc.) So no need to maintain it, it's simply true. 3e was
very minimally-tested, and that's a big part of why it goes off the rails so quick--and why E6 and such exist. They're part of the levels actually playtested.