Cool, then let's talk good solutions. What are yours? What priority are they to you?
Well, for one, general skill gain rates can probably be tweaked, but that's an issue of abstraction of time. Realistically, dwarves should take years to get really really good at something, but in-game that would be a pain. However, I think it would be less of a pain if some other things were changed, such as the game getting less FPS-trampled over time (and other changes that allow longer-term fortresses to be desirable or successful), being able to start with higher skills, immigrant skill levels being more sensible (or at least following some kind of logic instead of being insane-super-high at random)... so really, that's an aspect that ties into many other aspects of the game that also would have to be looked at.
Also, realistically speaking, a person can only learn so much in a day/week/month. If I'm trying to learn, say, chemistry, I
can read the entire textbook in one week, but that would cause me to learn much
less (and less well) than reading it over the course of a month. Same with combat: Practicing for eight hours per day won't give four times the benefit of practicing two hours per day. It'll give some amount more benefit, but not that much. I think that it would be possible, feasible, and both realistic and good for gameplay if skill gain were somehow attenuated by time in such a manner; training a skill should be worth a bit less if you've already done a whole bunch of training recently. This might seem pedantic or overanalytical, but in my opinion it's actually an extremely important part of balancing skill gain, because if you have skill gain very linearly with how many times you "train" without regard for how often you're doing it, then you're always going to have weird cases like this, on some level or another. It would prevent things like an invading goblin mysteriously gaining three levels of skill over the course of a few steps just because he happened to get hit with enough weak blows from a weapon trap.
You did mention lowering the training from something like a trap, and lowering the amount you'd get from one kind of repetitive training. I think those are actually both good ideas; a mechanical trap shouldn't be as valuable training as fighting another, live person, and the more repetitive your training is, the less of a benefit it should have (tying into my above idea). This would encourage players to mix up their training more, between things like drills, sparring, demonstration, active combat training, wacky things the players themselves come up with (like danger rooms), and so forth.
Bugs? Some things are clearly bugs. BSD is a bug. Training via traps? I'd call that more unintended behavior. This game is all about using unintended behaviors to our advantage. That's why the DF model is so much cooler than games with less complexity. Toady doesn't forsee me using caged FBs to milk extract to defend my fort. Toady doesn't forsee me using temperature to make my FBs milkable. To answer Krenshala's argument (if you can't tell I'm editing post-preview), I would rebut rather that any game that doesn't involve manipulation unforseen by the game designer is an insufficiently complex game.
There's a difference between emergent gameplay and unintended behavior. Yes, emergent gameplay is a good thing, but sometimes an element of emergent gameplay comes about that not only isn't something the developer intended, but is
contrary to what the developer intends, or is only possible because of systems that are flawed in the first place. If it makes sense for forgotten beasts to be able to be trapped and for their extracts to be handled safely, then yeah, using them in the way you mentioned is a good example of emergent gameplay. However, taking advantage of, say, being able to trap a dragon in a wooden cage is a bit different, because the only reason that's possible is because of highly flawed or incomplete systems. Interesting game mechanics leading to interesting behavior isn't the same as bizarre or nonsensical game mechanics leading to bizarre or nonsensical behavior.
Also: "BSD"?
Okay. So calling it an exploit is a super vague suggestion that Toady change the behavior? Any time, or soon? Because if it's anytime, I'm down with it (although I want him to have the framework to do it right before he bothers to do it). Won't call it an exploit though.
No, calling it an exploit is a way of noting that not only should the behavior be changed, but that the current behavior is unfairly, well,
exploitable. It's not just nonideal, but it's nonideal in a way that the player can use to very great advantage. See what I said above about emergent gameplay.
Seriously, if all that is intended is a suggestion for a change, then fine. It's just that the tone (forum used, language used) of the discussions around danger rooms in particular doesn't lead me to believe that that's all it means to call danger rooms exploits. It sounds like unnecessary judgment of people that use them.
I don't know about anyone else, but I'm not trying to make any judgment calls on people like that.