Hacker News

points by dutchbart 17 hours ago | 0 comments

The even more insane thing to me is that the controller knew that there was another train in that block, knowing it was no signal failure. The driving train should’ve never been commanded to continue on full speed by the controller.

[–] cedws 17 hours ago

Trains shouldn’t even be manually driven in 2026. We have self driving cars now which is a way harder problem to solve. Relying on a system of signals and driver attentiveness is archaic.

[–] delta_p_delta_x 17 hours ago

> Trains shouldn’t even be manually driven in 2026

Potentially, but the point is that signallers should physically be unable to allow trains to proceed, and drivers should physically not be able to override them. British signalling from the 1800s has physical interlocks; this isn't a matter of manual versus automatic train operation. It is clearly systemic.

[–] M95D 15 hours ago

How do you pass a defective signal then? How do you couple a second locomotive? How do you add cars?

[–] kube-system 10 hours ago

Yeah, if you completely ignore availability as a problem. The state of the art self-driving technologies we have today only work in some conditions, some of the time.

The reason vehicles are still driven manually is because humans rely on transportation every day, all of the time.

[–] dns_snek 16 hours ago

Driving a train and driving a car are two completely different challenges.

If signaling worked properly then it wouldn't matter whether the train was driven autonomously or not, the "problem" becomes trivial. Establishing and continuously verifying that signaling is working properly is the hard part.

Without signaling you're limited to crawling speed just like humans are when they don't have any signalling .

[–] 1718627440 17 hours ago

We do not have self-driving cars in all environments that clear the necessary security regulation! Also the socially accepted number of deaths due to trains is way below that for cars. The current accident is just a Tuesday for car traffic. (Spain recorded 1785 road deaths in 2024[1], that's 4.9 per day) Detecting objects (and future objects!) on a train track is still a hard problem.

The current system is basically that the train is driven both automatically and by a professional attentive driver at the same time. Changing it to just the automatic system is going to reduce safety. Also the real world is way to messy to reach acceptable availability rates with only automatic systems.

[1] https://www.itf-oecd.org/sites/default/files/spain-road-safe...

[–] fn-mote 16 hours ago

> Detecting objects (and future objects!) on a train track is still a hard problem.

In this case the “object” was another train. That’s not the “hard” part of the problem. This sounds like an apologist comment.

[–] dns_snek 16 hours ago

What difference does that make, genuinely?

Are there systems that can warn drivers and prevent collisions without external signaling, like ACAS?

[–] 1718627440 15 hours ago

Yeah, that's a solved problem. Did you even read my whole comment? I answered why "just use the system from self-driving cars" wouldn't work and in fact would be less safe than the current system.

[–] AnimalMuppet 10 hours ago

What's the stopping distance of a high-speed train at full speed? I don't know, but I bet it's kilometers, not meters.

Detecting an object on the tracks 2 km away (say), as opposed to beside the track at the same distance, is in fact a hard problem. The tech used in self-driving cars is not going to solve it.