Hacker News Reader

Comment by lotu | Hacker News Reader

> How many pilots, trained or not, really have any experience with a 100,000 ton ship in a crash situation with responses where seconds matter?

I would expect anyone piloting such a ship in a harbor/under bridges. We requite airline pilots to train for many unlikely plane failures because the alternative is letting planes crash that we could have saved with better training.

View on HN

You can train for it, then how many years into your career you actually experience such a scenario are you likely to act instinctually and recover. The best solution would be to improve autopilot assist as it will never forget how to correct (if possible).

I’ve done some sailing but have no real authority when it comes to vessels like these.

Friends of mine are pilots on the Thames (London) and I seem to recall one of them telling me it was over 10 years training before you could bring a big boat in. Pretty fascinating really - they figure out all the tides and weather and plan the route. On the day they board along with a sensor system that sits in the bridge and gives the position to a high level of accuracy.

Lol you're right, controlling a ship with human inputs is so hard that is was the inspiration for a Russian shipmaster to create PID control

But I believe no big ships use any kind of autopilot while near shore. they only use it in the middle of the ocean.

If it’s anything like airline pilot training, there’s periodic retraining and evaluation to make sure pilots have the right reactions in case of an emergency.

I wonder if they’re required to keep up training to the degree pilots are. It wouldn’t surprise me if they’re not.

Part of the emergency reponse they are drilled with in simulators are "memory items" that are literally memorized responses to emergencies, i.e. "reverse engines, drop anchor" this is to prevent freezing or indecision. Of course I guess that can still happen but they are trained to the point where it should be automatic.

Or you just use the tried a true method of recurrent training and simulator work like in aviation

Nope. The military, aerospace, and medical industries have all refused full automation in life-or-death situations with cause.

Isn’t that interesting when compared with ‘machines don’t make mistakes’ wisdom

You'd have to be either a complete headass fresh out of a coding bootcamp or carting around The Agenda That Ate Calcutta to fall for that line of bullshit. Anyone who's spent the wrong parts of their week debugging ought to intuitively grasp that machines are inherently error-prone.

You are not wrong, but nobody gave the memo to the boss of my boss.

> How many pilots, trained or not, really have any experience with a 100,000 ton ship in a crash situation with responses where seconds matter?

The crew would have know for several minutes that the collision was imminent. These things happen slowly, it wasn't like the bridge suddenly jumped out of the water unexpectedly in front of them. The tragedy is that the crew had no options to avoid the disaster.

Reply on news.ycombinator.com