It just confused me. How does anyone do well out of a failing startup? By jumping ship early? By going to a competitor? Surely there are no golden parachutes, right?

There definitely are golden parachutes. In my experience, when a startup is going to die, it will thrash around a bit before it quits living.

Most recently, the company was running out of money (~3 months runway), no significant progress was made on the large new funding round we were looking for, and a small round wouldn't do any good since there were big (expensive) milestones to meet. With about 2.5 months left, we laid off about 80-90% of the staff to extend the runway.

Most employees were let go with two week severance packages and about 1 week's worth of insurance. "Critical" employees stayed and were given sizable 'retention' packages with big bonuses if new funding was arranged.

For me, the applied skills I learned at a startup made me a much more attractive prospect. I can speak from experience about the relative strengths and weaknesses of machine learning models on real data, instead of the somewhat sterilized and over-used datasets often passed around in academia. It means a higher salary when I did jump ship into corporate-land.

Plus, it was just plain fun. Startups are the closest thing to the intellectual excitement and curiosity of grad school I've found in the private sector.

Still you get paid the whole time. You meet risk-takers. I've gotten almost every job for 20 years from connections I made at my first startup.

That's like asking "How does any do well out of failing?"

And the answer is surely the same: by learning from mistakes made.