*by significant stress problems I mean 3 or more (out of 100) dorfs I expect to lose to insanity; obviously higher in evil biomes buuut that would be expected no? I also always have one level of discipline on all 7 dorfs- I continue to maintain that if you think stress is broken in 44.12 you need to "git gud"
Um, no.
3 dwarfs out of 100 is not an acceptable number of losses when the fortress is otherwise running well.
The first insane dwarf should be a sign that your fort has started to fail.
No, I disagree with this. Problem dwarves should always be part of the game so long as there are ways of dealing with them. We had the "no Dwarf ever feels stress because his dining room is nice" version of the game. It was boring.
I generally agree with Shonai_Dweller.
A dorf lost to insanity in a full size fortress is not a big issue, assuming you're willing to accept a shift from the "every lost dorf is a failure" of 0.40.X+ to the current (0.44.10+) "you'll lose some along the way". What is not acceptable is if that lost dorfs grabs hold of the whole tapestry and pulls it down with him, or if that lost dorf is just a portent of where the rest of the fortress is heading if just given a bit more time.
If the balance is somewhat correct (assuming you accept the balance goal), the next siege should see fewer losses among those who were present during the first one, rather than each siege adding another stone to the burden (which seems to be a big issue currently). Again, some may snap under the stress of repeated sieges, but most should endure, carry on, and gradually return to some base stress level once the siege has been taken care of unless something special happens on the individual level.
The issue isn't the occasional problem dwarves, but the current inability to prevent those that should be preventable from getting there and overseer means of detecting that problems are brewing (at least without checking up on each and every dorf once per mongh or so), as well as the lack of means to rescue those that ought to be possible to rescue (and it ought to be a viable playing strategy to just take the losses without going out of the way to save individual cases, just replacing them with migrants [that might dry up, if the losses on the battlefield are too great, of course], as long as the general conditions are kept at a reasonably good level with most of the basics taken care of (food, booze, bed, socializing, praying, reading, and reasonable free time to handle needs, and the avoidable stressors dealt with [handle dead citizens with burial or slabbing, do away with corpses of sapients from the day-to-day view of citizens, etc.]).
There are two important parameters ThreeToe and ToadyOne ought to test: Long term (at least one generation born and matured), and somewhat poor playing, such as "forgetting" to set up e.g. temples for the first 5 years, occasionally "forget/fail" to produce enough clothing (e.g. through too small farm plots, or all thread producing plants "accidentally" used for booze production, which is very easy to do, especially for a newbie), and "detect" the mistake later when symptoms appear.
One restriction might be to just do underground farming (that's what dorfs do, isn't it: the surface is horrible!), while still logging (and thus hauling logs through the rain) and have an above ground pasture (how many newbies will realize without prompting that cave moss is useful for grazing except possibly for reindeer, and how many will rush for the dangerous caverns to release spores anyway?).
Another, preferably parallel, alternative is to recruit newbies and watch which mistakes they make (or have Kitfox do that and report the results, which is probably way more efficient).