[nullc]

You don't necessarily need to make that comparison: If you make the trains too slow many people will drive instead which is radically more likely to kill them than the train.

So you can still find that there is a speed risk tradeoff the minimizes death which is not the same as minimizing the train risk.

Worse for NY's problems is that in that decision people are probably using something like the 95-percentile trip time as the criteria: It doesn't help them if the train is fast on average since they need to schedule for the worst-likely case. The fact that the crews are playing roulette with an kafkaesque speed limiter system and risking disciplinary action should they anger it essentially guarantees that the times will be both slow and highly inconsistent.

[jraph]

Indeed fully receptive to this argument.