Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> at the time the design decision was made, Tesla was still a startup.

If you're going to sell a product, the Model S, at the same price range as a BMW M5, 7-series, or the equivalent Mercedes S class, it had better be a premium quality product. Start-up or no.



Wow, I guess I touched a nerve :-) Have you ridden in an early model S? Tried to repair a Roadster? I am a big fan of Tesla and impressed at what they pulled off, but I don't forget that when they started they cut all the corners.

It is common, perhaps even expected, in startups when an engineer says, "Well this won't last more than 5 or 6 years" to be told by upper management, "Well if we're still around in 5 or 6 years, we'll do something about it."

Given that Tesla is not a startup these days, and is on the other side of that chasm, I was surprised that their response here was "oh we'll replace the board with another board with the same issue."

Perhaps there is not a good understanding about the differences in the legal ramifications of a vehicle failing and resulting in property damage or loss of life from the failure of something considered a 'wear' item versus something that is not classified as a wear item.

I am not a lawyer, one of my friends who does consumer product safety litigation and is a Tesla owner, immediately responded to this news that this was just Tesla trying to avoid criminal liability when a failing EMMC chip kills someone. Sure, they see everything in terms of liability :-)


> Perhaps there is not a good understanding about the differences in the legal ramifications of a vehicle failing and resulting in property damage or loss of life from the failure of something considered a 'wear' item versus something that is not classified as a wear item.

This was my first thought as well. It feels like they were trying to avoid some implication beyond mere replacement of storage modules. Perhaps they are aware of some other component that will also fail at a high rate before the useful life of the vehicle?


That seems logical, but expensive niche cars tend to be worse in many ways than the mass market equivalents. BMW’s more expensive cars for example are simply less reliable. At the extreme Bugatti Chiron skips a lot of modern car tech.

It’s always a trade off. New tech may mean better acceleration or whatnot, but you don’t have 30+ years of reliability data to work from to refine things. The early years of automatic transmissions for example are words apart from what’s available today.


> but you don’t have 30+ years of reliability data to work from to refine things

... Hold on, are you suggesting that when this car came out, under a decade ago, no-one knew about Flash wear? If anything, people would have been more conscious of it then than now.


No, I am saying when designing a new system their isn’t an automatic processes to discover faults.


the sort of persons who can be entrusted with designing a totally bespoke motherboard like that, and it is totally a custom job, absolutely should be aware of flash wear out.


Sure, and all code should be free from bugs.


totally free from bugs is a much higher and impossible standard to meet than the bare basics of "don't solder a SSD onto a motherboard that you know will wear out in a few years".

It's more fundamental, to continue the software analogy, it's more "how did they possibly miss that" error like shipping a device that's supposed to have mariadb listening only on localhost, but it actually listens with no authentication on all live network interfaces.


You’re picking a single mistake as somehow unusual. It’s easy for the original spec to have been fine, some change happens that nobody considers in that context and boom you get an issue. Depending on usage pattern a SSD can be expected to last a 20 years, but changing the usage pattern by say storing more data isn’t going to raise a red flag saying you now have an issue.

Someone turned on excessive Linux logging and boom an issue shows up early on some a Tesla Models. Chances are it wasn’t even intended to enter production that way.


> you don't have 30+ years of reliability data

Flash memory was invented in the 1980s, so yeah: we do.


Steel has been in mass production for well over 100 years and people still make the same mistakes today with new mechanical systems. Flaws often seem obvious after the fact, but that’s when your looking at actual failures. If it never occurs to you that something could fail, well good luck.


An embedded engineer which is unaware of flash wear is incompetent. It's the number one failure mode of flash and more often than not drives the design decisions around it.


Aware of and accounted for are different thing. I mean metal fatigue is hardly some deep secret, yet it’s constantly causing issues.


So something like:

If the metal structure holding the instrument cluster and infotainment unit fell apart after 6 years and the manufacturer charged the customer to fix it?

That also wouldn't be ok.


> BMW’s more expensive cars for example are simply less reliable.

Less reliable for the second or third owner.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: