I can offer a somewhat unusual perspective on Tesla and SpaceX, from the manufacturing POV, because I never did sign their production NDAs, although I have made parts for them.

Producing for Musk companies is a fucking shit show. Specifications change mid-production on a constant basis, to a ridiculously comical extent. I honestly can’t imagine what the hell is wrong with these people that they change hole locations and part geometry for every single order, knowing full well that 1000 of 1000 units have already been produced and boxed and loaded onto pallets.

As an Elon Musk fan, it was rude surprise that their shit is so fucked.

I tried holding off on production until the last possible nanosecond, but no. No good. Changes happened the day orders are due.

The only Musk work I do at this point is emergency corrections for other local outfits who made parts with rapidly evolving and mutating requirements.

I thought that the designers would settle down as Tesla and SpaceX get through production hell, but they seem committed to mass production of no-two-things-ever-the-same.

I drive a Model S and I figure probably most things people make and do are objectively retarded, I just happen to know more than I want about this particular retardation.

Pretty sure calling things retarded is a nogo in 2021.

Yoų can say the seven words freely tho.

Shh retard

The big question is, would you buy a model S again?

Did you know what their manufacturing process was like _before_ buying your model S?

That's modern day software development/management for you. Thank the MBAs

Software/design issues requiring hardware recalls are routine in the industry, though. There's really nothing notable about this particular issue at all, except that it involves "consumer-like" devices that posters here understand and feel expert on. We all think that "we'd never have done something this dumb", so therefore it's an easy shot to take where esoterica about engine control PCBs isn't something we have the confidence to second guess.

(Needless to say, we totally would have made the same kind of mistake, and probably do so every day.)

The only meaningful ding on Tesla here is #4 above: clearly this was a design flaw and should have been a recall from the start, not a customer-financed repair.

>There's really nothing notable about this particular issue at all, except that it involves "consumer-like" devices that posters here understand and feel expert on

I think you're off base here; it's the very fact of Tesla using "consumer grade" hardware that is the (notable) problem. The people who "would have made the same mistake" are qualified to pass judgment if they know they are not qualified to develop automotive grade systems.

I don't think that holds. In fact while longer-wear flash devices do exist in the market, the design mistake here would have driven any feasible part to failure. While you can argue that choosing consumer parts was "a" mistake, it wasn't the one that caused this failure, which was a software/integration thing.

If you had twice the writes per cell, the life would be 10 years instead of 5. Still wasteful, but...

And if you stepped up to something with effective wear levelling and a bit more capacity, instead, this bad design could have lasted forever.

Or if you socketed the eMMC, it would have been no big deal.

I think the biggest mistake here is the drivetrain depending on a device which is known to fail (eMMC) even without the logging problem (why not using a separate device for the logging?) it would eventually fail, other manufacturers have similar problems, (Mazda, Ford, Mercedes) but you just lose the audio system.

> drivetrain depending on a device which is known to fail

Uh... the drive train itself is "known to fail". Everything wears out. Everything breaks. You just aren't surprised when your transmission needs rebuilding or your brakes or shocks need replacement because those are "car things" you've been conditioned are normal. Well, electromigration is just another kind of normal wear and tear to be designed into the device.

And it's the design here that was at fault, having incorrectly planned for the replacement rate. But there's absolutely nothing weird about building cars out of parts that fail over time!

The drivetrain doesn't fail when eMMC does. A whole lot of stuff that you'd normally do on the touchscreen is inaccessible, though.

I certainly have shipped dodgy stuff knowingly for lack of time to do better. "This is a prototype, we'll apply the Chinese warranty, if it's broken we'll replace the unit". I would ship non-UV cables to outdoors stuff, because I had no time to select and order a better one, I shipped an outdoor temperature sensor cover that was not open to the air enough and was just tracking the sun. The only red lines were thunder and fire. I don't want to start a fire, and I don't want to conduct the thunder.

But I did that way earlier than Tesla in the process.

Yeah, but it's not a single person or single decision. They went with consumer eMMC, and probably had a justification showing that they'd never need replacement for 30 years of use or whatever. But then someone else overfilled the partition, leaving not enough space for the hardware wear leveling to work in, which over-wears the available space. (No idea what the filesystem choice was, some are more susceptible to this than others). And then a third party decided to ship with the logging settings set to dump all sorts of junk to storage that wasn't needed on a production device.

None of those are particularly boneheaded in isolation. It's just a bug. Bugs just happen.

(But again: design bugs should be recalls and not repairs)

> Chinese warranty, if it's broken we'll replace the unit"

That's the mass production warranty for integrated objects. Calling it "Chinese" is just weird unless you are calling attention to a claim that only china does mass production anymore.

For cars, we have the "extended warranty" which is just a bet that the car won't fail to much on average.

Same for hard drives with 3 vs 5 yr warranties.

> Always thought of this as a fascinating race to watch. What will happen first: (1) Tesla figuring out how to really build & manufacture cars or (2) the legacy automotive companies really figuring out build & deploy software?

Yes! And (3) is potentially: A company who's even better at building and deploying software and hardware enters the market.

You forgot the much more cynical (4) Tesla lowers the bar for success and quality diminishes across the industry in a race-to-the-bottom on price. ;)

A Dacia Sandero Stepway (5 seater small SUV/crossover) costs € 13 000 brand new. How much is the cheapest Tesla again?

Apples to oranges if it's not electric

The Chevy Spark EV was supposedly nearly as cheap after rebates, I think. But I believe it was discontinued, and I assume that means it probably was being produced at a loss, so it doesn't really count.

Should have compared it to a Dacia Spring then: 17.500 euro with >200km range.

I believe that is before subsidies and including VAT, so it will be more like € 11 000 after subsidies. The point still stands: there are many other manufacturers with long experience in making cheap cars.

Nothing that Tesla has built is anywhere near the bottom of the industry though.

If you are referring to Apple, I have a question.

I see a lot of speculation that apple will produce a car.

But apple at the moment just designs hardware, doesn't really make it (as in they don't have their own factories), so presumably they would either need to build factory for their cars or hire some other factory(or factories) to build cars for them ?

As far as I understand most cars sold in USA are build in USA (I am not sure if that's only because of shipping costs or if there are any import taxes or something).

Even if most parts were from China finally assembly would probably have to be done in USA, and factory quite big and expensive, and have lots of trained people (to have apple standards).

Given that its apple that we are talking about car would almost certainly be an instant hit, so they would need to produce a lot of them, which again means big factory with lots of people.

How likely is it that everybody missed such factory ? Especially with press so obsessed with apple. Even if some other company like Toyota were building it for them, they would likely need a lot more additional capacity than they have now ?

And we have seen with Tesla, that scaling out is really the big problem here.

I genuinely interested how likely is it that apple cars will be on streets anytime soon.

There was speculation that they were partnering with Kia/Hyundai to do the manufacturing. This was big news a week ago, but Kia/Hyundai have backpedaled since.

Tesla seems to have a clean slate advantage.

Every established company I've worked for seems to have as many processes that get in the way as those that help.

also - nobody mentioned the amount of data collected. Is it good diagnostics or is it hoarding?

I remember the story of the guy who tried to return his tesla under the lemon law thing because of some weird failure. Tesla diagnosed it and found he would pop the trunk before every failure happened, and they used that to prove he was doing it. After that, I can see an engineering push to log more, not less.

Gm has this covered I believe. I think they have a less totalizijg view of over the air undates than Tesla. Tesla uses it as a marketing tool to attract “tech savvy” consumers. Gm has good tech look at Onstar.

It ain't tesla