Well, we shouldn't necessarily conflate overdose risks with risks of usual usage -- pure water will kill you in a nasty way if you drink twenty gallons of it. The ratio between effective dose and LD50 is often cited as a metric of how likely a drug is to cause accidental overdoses, but even that is sort of arbitrary.
I share your pessimism about tricyclics and think that people in general should be more cautious with how they approach these things, but stuff like Trobalt is in a different category for me for three reasons:
* has only existed on the market for 7 years; we don't know what the 30 year outcomes are
* relatively uncommon, so that serious side effect data is even more limited
* actual FDA black-box warning which isn't given out lightly.
Tricyclics have been around long enough that we have some grasp on how they work and a better one on what the likely outcomes are; they are certainly some number of sever adverse reactions, but they are extremely rare on the order of 1:10's of thousands of users, probably 1:millions of doses. Just from what we've seen with the bad, limited data we have here, Trobalt appears to cause serious or long-term consequences with more frequency.
I did experiment with Trobalt briefly, but given my prior history of enthusiastic exploration of little-known compounds, that is more a negative comment about the way I approach chemistry than it is an endorsement of Trobalt. Stuff was bonkers.