Hacker News Reader

Comment by eterevsky | Hacker News Reader

When a manufacturer is selling millions of cars, I think it's better to judge their safety using statistics rather than some qualitative reasoning.

Are Teslas measurably less safe than other cars? I.e. is the accident probability per km driven on a Tesla higher than on another car? I haven't seen any data suggesting that, but I'm happy to be corrected.

View on HN

The number of accident might be lower (no idea) but still, If I'm responsible of killing/injuring someone with my car, I'll pay the consequences. If Tesla is responsible of killing/injuring someone with their car (when they drive through the autopilot), they must also pay the price (probably damages or jail depending of the situation).

How is that any different from any other product from any other manufacturer, though? Tort law has a thousand year history in our culture. Is there something specific about Tesla that needs something new?

I mean, what you say is sorta specious. Of course that's true, it's always true. The interesting question is "are the cars dangerous?". And the answer seems to be a pretty emphatic no, at this point. So instead everyone wants to argue about abstractions ("they're still liable") or absolutisms ("no failure is acceptable").

And that seems increasingly counterproductive, and frankly to have more to do with the somewhat questionable mental stability of the CEO than to the behavior of the actual products.

Tort law requires evidence. The only entity with actual evidence of what the car did is the entity who is the defendant in the litigation.

All of the elements of the Tesla business model like no repairs, no dealers, etc are all about maintaining that control.

Well, it is a bit different. FSD theoretically can get driver into bad situation, then beep at them "I can't handle that", and as long as beeping was early enough that's no fault of Tesla even if the start of the event chain was caused by it.

Also Full Self Driving is extremely deceptive name for feature that does not do that

> Is there something specific about Tesla that needs something new?

Is this even a real question? The product acts on it's volition! Ofcourse it's different.

Imagine you buy a crate of crate of C4 - well, its super dangerous, but it does not have a mind of it's own. Untill you touch it, it stays put. So long as you store it properly, it doesnt go off. If you kill someone with it, you go to jail.

But with a car, it can suddenly decide to kill you because of bad code. It could allow someone else to take control and kill you. It could kill someome else and make it look like you committed murder.

The fact that it has a computer and follows someone else's orders is a foundamental difference.

What does any of that have to do with liability concerns? Sure, if an outrageously hyperbolized murder robot decides to kill your parents you have the ability to sue its manufacturer for damages. No one contests that.

The question is whether or not the murder robots are a real thing or just a meme. And I think we all know the answer. Look, the cars are safe. Another story on the front page today noted that the Model Y is currently the best selling car on the market. If they were dangerous in any measurable way, we'd know. They aren't. They just aren't. But sure: if they do decide to kill their overlords, it's 100% on Tesla.

That is not how the real world works.

If a car kills someone (e.g. it explodes, or burns down, or brakes fail) but the manufacturer took all legally required precautions & followed all the rules & regulations, then they're not legally on the hook. It was "an accident".

Queue in the Ford Pinto controversy: https://youtu.be/jltnBOrCB7I

> That is not how the real world works

Yes, we noticed that corporation are never held accountable for their actions.

After all, personal responsebility is only for little people.

The fact that it's an accident doesn't remove the liability.

If a human driver fail to brake somehow (like they have a bad reflex and push the wrong pedal), they might have followed all the rules and regulations, it might be "an accident", but they will still be liable for dammages they cause.

I'm not sure why it should be any different for a software bug in the autopilot for example.

Since Tesla works really hard to hide who is responsible in case of an Autopilot crash, witholds and deletes information as well as deny any responsibility anyways, being just as good as a human driver isn't nearly enough.

Additionally, there's no data because Tesla releases nothing except heavily doctored numbers meant to make them look good. (To Tesla, a crash means airbag deployed. Sorry, random person that got ran over and killed, but you're out of Tesla's statistics)

You can count on Tesla for one thing, and that is to lie.

I'm not talking about Autopilot specifically because it may be difficult to judge. But surely there is some statistics based solely on the car model?

Nope. This whole whitsleblower thing is because it is company policy to hide from everyone every single malfunction event.

>When a manufacturer is selling millions of cars, I think it's better to judge their safety using statistics rather than some qualitative reasoning.

Why do you think that? If we used statistics to guide criminal justice policy we’d be in big trouble

> If we used statistics to guide criminal justice policy we’d be in big trouble

Uh... we do use statistics to guide criminal justice policy? Not sure I understand what you're saying. How do you think they decide where to police and what to prosecute? How they decide on legislative penalties? It's true that this process isn't necessarily 100% data based, and in a bunch of ways ends up very unfair, but it's absolutely driven by a mostly-sincere attempt to get the most social good out of our limited enforcement and regulatory budgets.

Dog whistles aside, the entire justice system is driven by stats.

The statistics are the best part of criminal justice policy. DNA tests, Blood typing, fingerprint matching all run on statistical analysis. Much better than the memory of a witness.

There is nothing fundamentally wrong with statistics, you just need to maintain an appropriate threshold for where you draw the statistical line.

Tela insurance is way more expensive than other cars.

This hasn’t been my experience.

To my understanding, that is because Tesla car replacement parts are hideously expensive. Not because they are unsafe.

If they don't cause many collisions, the cost of replacement parts isn't significant.

Tell that to all the Hyundai/Kia owners whose cars are just spontaneously getting damaged while sitting parked. ;)

But they are when they get involved- the odds of getting into a crash with a Tesla goes up significantly when you happen to be inside of it all the time.

Insurance covers risk. More cost exposure, more cost gets passed on to the customer

That doesn't really matter.

Lets assume the ridiculous best case scenario: Tesla cars are 2x as safe as other cars.

If replacement parts for a Tesla car cost 2x the amount of regular cars, there is practically no difference in cost. It will still cost more to repair that Tesla comparatively to other vehicles per accident.

Remember, insurance also calculates the driver's individual risk level.

Combine that with reckless Tesla owners letting the car just drive for them while asleep at the wheel. Tesla cars still have a high risk overall.

I paid $129/mo for insurance ($500 deductible) on a Toyota Highlander. It went up to $179/mo for a Tesla Model Y with the same insurer, and when I switched to Tesla for insurance it went down to $124-144/mo (dynamic pricing, $1000 deductible).

maybe that depends on your driving habits (which telsa monitors)

Are you referring to insurance on a Tesla car sold by third parties, or the insurance that Tesla sells for their cars in some but not all states in the US?

Reply on news.ycombinator.com