Hacker News Reader

I never understood the liability aspect of vehicle repair from a manufacturer's standpoint. The common refrain seems to be that these new vehicles are not serviceable without the dealership/manufacturer. It seems to be a narrative to drive out the owner from repairing vehicle or independent repair facilities.

The liability seems to be the same as historic vehicles, for... vehicle repair. So why is this any different?

The burden of proof is on the manufacturer to prove that a modification caused damage to the vehicle.

I'm also surprised nobody (well the press and regulators): Why are you allowed to design vehicles whose safety is brittle enough that it can inadvertently undermined by an independent mechanic?

There's no way to design vehicles that can't be made unsafe by a mechanic making a mistake. Even for purely mechanical things like brakes, a mechanic can leave out a bolt or crack a structural member so that it breaks on the road. EVs require careful procedures, for instance to make sure that any of the high-voltage wires can't rub against anything and wear through the insulation.

Dealer mechanics make mistakes too, but at least they're trained on the specific car, and they can collect data to notice a pattern of things breaking after certain people worked on them.

Most software is secured my limiting access. That’s not really an option with a car.

> So why is this any different?

It is only different in that by making cars internet-connected, the manufacturers see a golden opportunity to make them all cars-as-a-service that you must buy but no longer own. In other words, profit and more profit, greed.

Yea but the burden of action is on the victim to have a more expensive lawyer so it doesn't matter one way or another.

While I don't doubt your enthusiasm, I don't want my car experience to feel like Arch Linux. I want something I treat like an appliance or utility to just work.

I am with you of course.

What I am highlighting is that if enough people are affected by a bankrupt OEM, it is not the end of the world to fix them so all is not lost in that case.

I also don't think it is a particularly smart idea to have your car needing to chat to the mothership before it decides to do it's job of driving you somewhere. At least for the consumer I see limited benefits of this trend.

The "just work" people are the reason for the prevalence of phone apps and cloud services in so much today's gadgets, and that dependency on services is exactly why all these things fail as soon as the company who made them fails.

I'm in the same boat, which is why I've soured greatly on a lot of things internet-connected. Internet connectivity almost always means needing perpetual updates (whether for security or just for keeping up with remote APIs changing). The economics don't make sense for providing that for free forever, so anything under that model must be either a rental/subscription or isn't built to last.

It’s definitely too much work for everyone to repeat and duplicate.

The obvious solution (to those of us here) is open-source and right-to-repair requirements.

I don't think the average fisker or lucid purchaser is willing to do 500V hacking. In fact, I would wager the DIY fisker/lucid community might be less than 6 people.

I admire the spirit but I just don't think hackers will do much with $100k bricks beyond salvage components.

Absolutely. But they may be willing to pay someone to swap a component. EV hacking is a small community still, but with more of them on the road every year more and more people get interested. If $100k are turned into a $0 brick and you can turn it back into $100k there is a commercial as well as academic interest there.

CAN bus is in many ways too difficult to encrypt. (not that it can't be, it's just got way too small a packet size). Nothing prevents manufacturers from getting complicated though - eg "knock three times on this ID, then request info from this ID then .....". (example from an old Shadowrun netrunning campaign idea)

I'm working in EV charging domain. EV chargers that comply with ISO 15118-20 may prove difficult to get working once more (due to key signature requirements). Otherwise, all should continue working as long as the charging protocol they use continues to be supported. And that said, the EV charge managing component should be replaceable.

for e-bikes at least, it seems that Bosch e-bikes use AES encryption between their controller and the battery, to prevent third-party batteries, perhaps you have knowledge about this, or how it could be possible to circumvent?

You'd probably have to extract the encryption key (which has to be the same between all controllers and batteries, otherwise it wouldn't work).

Depending on the hardware this can range from needing to attach a debugger to JTAG, to dumping a flash chip, to decapping a microcontroller at the hardest.

I can't crack encryption I am afraid. Got a link? It's the first I hear about this.

Yikes, how is this even legal for a EU company.

The 2015 "Jeep Hack" (affected a huge number of Fiat Chrysler cars) was actually one of the coolest moments of hope & inspiration for me. Ok, so the embedded OS isn't linux, but it still runs good old FreeDesktop D-Bus? You can just RPC to the window & tell it to roll down? You can just RPC to the head unit & adjust the volume? Nice. https://www.wired.com/2015/07/hackers-remotely-kill-jeep-hig...

I'd assumed everything was some terrible impenetrably dense proprietary cruft, but oh, the whole system runs some standard well known protocols that would be super super super great & convenient to use, if we had network access to the car? Yeah, nice.

Wouldn't it be so great if someday things actually got better. The "Hack" was, oh, you can engage the parking break if you get network access. Oh no! So the response is that now we for sure will never be empowered useful-to-ourselves consumers again. It sucks that the low-trust society keeps dominating, that it keeps overshadowing the can-do is-possible universe, that making possible is forever dwarfed by Fear Uncertainty and Doubt.

The one really bright spot of the past, in my view, was Webinos, which was a super cool IoT system that had some automotive interest. Notably, unlike most systems, it kept users sovereign, let devices (like cars, or car windows/stereos) expose themselves up to user-specified gateways, securely without intermediaries. Didn't make it, but I have yet to see anyone with 1/5th the promise, with anywhere near as low-ickiness as modern devices.

How far can you go with these modifications before the vehicle is no longer street legal? I suspect that even minimal changes might cause an issue there?

Depends, but it's possible: All vehicles I built were signed off by the licensing bodies & fully insured. We told both the lawmaker and the insurer about all our modifications and with some paperwork it was possible.

The community I linked above is full of people doing this kind of stuff. Some if it illegal, some of it legal.

Jailbreaking an EV I haven't thought about yet, but if the original manufacturer is bust, and you can show some degree of care to the licensing body I am sure it's negotiable.

Obviously it would be better to not be in this position in the first place, but problems are fixable.

Yes, for sure. Thanks for your response, the info you share is very insightful. Hopefully I will never have to use it :)

>Fisker Ocean was in a similar place recently and I started looking at their owner groups to see if there would be a business case

Rich Rebuilds got one for this exact reason recently https://www.youtube.com/watch?v=b_OxgAYG0Io