I would hate to be a product safety attorney at Tesla
-
There is some more information on potential Tesla touchscreen failures in the news this morning. Autoblog has information on what the failure is and the Seattle Times has some additional information.
https://www.autoblog.com/2021/01/14/tesla-nhtsa-recall-touchscreen-failure/
https://www.seattletimes.com/business/tesla-balks-at-touch-screen-recall-us-agency-takes-action/My day job is as a product safety engineer, and I work closely with my company's safety attorneys. My thoughts are based on my experience only and I do not have specific information on discussions within Tesla.
The basic concern of the NHTSA is that the defroster can only be accessed through the touch screen, and the touch screen uses a memory type that will reach its maximum number of read/write cycles in 5 - 6 years. This could disable the defroster which is considered a critical safety feature.
Tesla has 159,000 cars to fix. At best, it will cost them $1,000 to fix each car based on a rough estimate on the cost of parts, labor, logistics and the overhead of managing a recall. It could cost them as much as $2,000. That means that running a recall would cost $160M - $320M. This would wipe out profits for quite a few consecutive quarters and would have a dramatic effect on the stock price.
However, there is a risk if they don't run a formal recall. If a single Tesla owner gets in an accident because the defroster failed due to the touch screen failure, they will sue Tesla. The plaintiff's attorneys will argue that Tesla was aware of a defect and they ignored it. They will argue that offering a software update for a fee is not an acceptable remedy for a known defect, especially if the software update only delays the issue and doesn't fix it. Tesla's attorneys will have to argue that the defroster is not a safety system, or they will need some other argument to indicate that the driver was at fault for not manually clearing their windshield when the defroster failed.
The risk is that this could become a class action suit. The plaintiff's attorneys may argue that the basic remedy would be to refund owners the cost of their vehicle, raising the potential cost to Tesla by a factor of 50. This doesn't even include possible punitive damages.
This is all supposition. I don't know any internal discussions within Tesla. I can't even state if the NHTSA position is correct. I can only estimate what the potential risks to Tesla are.
One thing to remember is that the plaintiff's attorneys will sway the jury based on emotion and the Tesla attorneys will try to use logic. History shows that emotion is often more of an influence on juries.
GM faced significant financial impacts from their ignition switch problem and only survived through bankruptcy and the creation of the new GM that didn't have to cover all of the financial obligations of the old GM. A jury may see Tesla as worse for knowing about the problem for a long time and charging customers to fix it, especially if that remedy only delays the issue.
This is high stakes poker. The financial risks to Tesla could be significant.
-
@roadkilled said in I would hate to be a product safety attorney at Tesla:
memory type that will reach its maximum number of read/write cycles in 5 - 6 years
This seems incredibly short sighted. I can't imagine a competent software engineer would overlook it.
-
@roadkilled So the touchscreens begin to fail after 5-6 years of normal use? That would be a decent lifespan for a laptop but not for a car. I'm not convinced a rear defroster is a critical safety feature but a front defroster seems even more important. Wouldn't that be under the same menu?
That gets me thinking, many Ford products only a couple years old have chronically unreliable backup cameras, which are a federally mandated safety feature. Would the same risk of legal action (although likely greater financial resources making it less risky) apply to them?
Color enthusiast, Citroën fanatic, and captain of a 1985 Mercedes 300D
-
By affect their stock price you mean it will go up because "Elon is doing the right thing with his company", right?
Because logic does not work with Tesla stock
2020 Gladiator, 1956 F100 project, 1993 Eagle Summit race van, Mystery car?
-
@zaphod-s-heart-of-gold So open short positions or no?
-
@mastermario said in I would hate to be a product safety attorney at Tesla:
I can't imagine a competent software engineer would overlook it.
OTOH, a software engineer often must work with whatever runtime environment she is given and may have no say in what the hardware looks like. In this case IIRC they run Linux with some low-level graphics library, so they might be developing and performing unit testing on regular PCs and doing integration testing in emulators or dev boards with final validation in real hardware, thus maintaining several layers of abstraction between the dev environment and the real thing.
"I drank what?" -- Socrates
-
@mastermario The software engineers aren't the ones speccing the storage. What the software engineers have done is made an update that greatly decreases local logging so they don't burn through that write limit as quickly. But it's still an issue, especially for people who are at or close to the limit now. It could have been easily and cheaply fixed initially by using better storage, larger storage, and/or removable storage, but now that these cars are out there, it becomes an expensive fix.
-
@aremmes @facw I have no experience with any sort of programming or hardware development, my point was, as a system, however it was designed, this seems to have been made with a glaring mistake. I would assume this should have been caught at least in final validation testing.
-
@roadkilled Their stock is up 6.5% as we speak. Tesla is a cult, no doubt about that. Any other automaker and their stock would have dropped 10%. I found this on some weird automotive site that I've never heard of:
https://jalopnik.com/tesla-would-take-nearly-1-600-years-to-make-the-amount-1846044574
Edit: HOLY CRAP! Tesla stock is actually NEGATIVE!!!! (11:26 am)
-
@roadkilled Leave it to TSLA to make Y2K a reality in 2025! lol
-
@roadkilled LMAO
This is why FAA has such strict rules for certifying flight control computer hardware.
-
@mastermario They probably should be doing long-term endurance testing that would catch this sort of thing. I haven't actually read if their testing missed it, or if they decided that it was fine as long as it happens out of warranty in most cases.
-
You're right, and most (all?) normal automakers would do a recall - and also would have not made such a design decision along the way I'd think...
Elon can tweet some dumbass meme and bump the stock price billions, then raise more capital by selling some more equity (like last Sept). He doesn't seem like the type to care about that stuff. I can't imagine working for him... I should just make tons of money off the stock, but I just refuse. And I refuse to buy his cars.
-
Shit like this and all the crashes due to the inappropriate use of autopilot makes me wish they’d just crack down on them. Just cheap short sighted gimmick crap that will eventually fail a decade down the line,
-
@roadkilled said in I would hate to be a product safety attorney at Tesla:
One thing to remember is that the plaintiff's attorneys will sway the jury based on emotion and the Tesla attorneys will try to use logic. History shows that emotion is often more of an influence on juries.
Emotion certainly seems to be an influence on other Tesla matters, such as valuation... I wonder how many traditional cars' defrosters and defroster switches survive as many cycles as you give here.
But my bigger question is, how many other peculiar problems and defects will come up in the future? I've made a bunch of money from a small TSLA investment, but I think buying one of their cars would be a poor investment.
-
@whoistheleader Apparently the backup camera is another issue the NHTSA is looking at for this issue.
I don't think it was initially overlooked. If I had to guess, it was a lack of foresight. I'm assuming that the original design of the system had fewer parameters that had to be stored in the memory. The Autoblog article said that the memory can handle 3000 erase/write cycles. The estimate is that it might only take 5 - 6 years to reach that point. Let me use 5.5 years to simplify the math since that is 2000 days.
I'm guessing that the original design only wrote to this memory when a major setting was changed. The original Tesla engineers may have been looking at settings that get updated less than once a week, possibly only once a month. I don't know what the settings would be, but let me use something like radio presets as an example. I would be surprised if I change the presets on my radio more than a dozen times over the life of my car. An active user may do it 100 times. You could have a few settings of this type that get written without coming close to 3000 cycles over 30 years. The original designer made sure that the only settings that went into this memory were settings that you wanted to keep if the battery died or had to be replaced.
Now, what happens if there is some turnover in the design department. The people who knows what settings are stored in which memory leave and new engineers come in. The new people are asked to implement new features which become standard. The new programmers either don't know which memory the settings are going into, or they weren't told of the limited erase/write cycles of the memory. As more features are added, the frequency of write/erase cycles increases until it gets to 1 - 1.5 times a day. At that point, you reach the limit in 5 - 7.5 years.
There are problems with this theory, and it is pure supposition. I'm not proposing an intentional design or even an incompetent design. My theory is based on a lack of documentation or communication. Let me again reference the GM ignition switch issue. The people who found out that there was an ignition switch issue didn't realize that the switch could disable the airbags. It wasn't a willful decision, just a lack of communication between groups.
I'm not anti-Tesla. I don't own one for financial reasons, not philosophical reasons. I bought a 2016 Mazda 6, but I could have waited a year or two to get a Tesla Model 3. The Mazda cost a little more than $27K without taxes and fees. The Model 3 at launch would have cost $15K more at a minimum. (Has Tesla ever delivered the promised $35K Model 3?) I would need to add the cost of adding a charging port in my garage. In theory, maintenance costs would be lower over the long term, but in 5 years my car has only needed scheduled maintenance costing $2K. Much of that is brakes and tires which would still be an expense for a Tesla. I can't see myself spending an extra $15K in maintenance over what a Tesla would cost. The more expensive car would also have more expensive insurance. I would also have likely needed to rent a car for a number of long trips. For example, my wife and I have driven from Seattle to Jasper and Banff a few times, and I don't think it would be easy with the current Supercharger network.
-
@whoistheleader one of the articles linked specifies it affects both the defogger and the defroster so I assume that means front and rear.
-
Sounds like the NHTSA is made up of shorters and pedos....
-
@zaphod-s-heart-of-gold Beat me to it. These things could become mobile incinerators, and stock would double.
-
@roadkilled they will pass it off as "regular maintenance" and gaslight you into thinking negative consumer practices (just like apple) are beneficial. Or that you are just looking for something to whine about.
Call me when the money runs out.
-
Serious question, what would it take to retrofit this into a Tesla? Would it be doable?
2022 Ioniq SE
2008 Crown Victoria Police Interceptor
1964 Corvair Monza -
@roadkilled said in I would hate to be a product safety attorney at Tesla:
The estimate is that it might only take 5 - 6 years to reach that point.
I think the word "might" is a big word. Kind of like saying a tire might wear out in 18 months. It all depends on how often you use the car and how you are using it.
If I were Tesla, I would implement a memory life monitor based on actual usage and provide a no charge replacement up to 100,000 miles if used up. Similar to what they do with pollution controls I believe.
-
@racinbob said in I would hate to be a product safety attorney at Tesla:
@roadkilled said in I would hate to be a product safety attorney at Tesla:
The estimate is that it might only take 5 - 6 years to reach that point.
I think the word "might" is a big word. Kind of like saying a tire might wear out in 18 months. It all depends on how often you use the car and how you are using it.
If I were Tesla, I would implement a memory life monitor based on actual usage and provide a no charge replacement up to 100,000 miles if used up. Similar to what they do with pollution controls I believe.
The problem isn't that easy to dismiss. There are 159,000 vehicles being considered, and memory usage will probably follow a typical distribution. That may mean that 1% have a problem at 5 years, 3% at 6 years and 8% at 7 years. The NHTSA isn't going to say that it's acceptable for a failure at 5 years because it's only 1%. That is still 1600 vehicles. You have to assume that some fraction of owners won't get their car serviced on a given schedule even if they get notice from Tesla saying "bring your car in for service or your screen will go blank."
Again, the GM ignition lock failure happened only after many cycles of the ignition cylinder, and it only happened to an extremely small fraction of the vehicles. No manufacturer gets to make the argument that a potential safety issue only affects a small number of products. The law of large numbers will catch up to you eventually. If you build enough cars, and there is a defect, something will happen regardless of how hard you try to avoid it.
I'm more familiar with the CPSC than NHTSA. The CPSC requires manufacturers to go to great lengths to reach product owners, and the CPSC does monitor recall effectiveness. How many defective products have been fixed or removed from the market? If the recall is going too slowly, the CPSC will require the manufacturer to spend more money to reach more consumers.
The problem with the memory life monitor is that, in order for it to be accurate and not lost, it needs to go into the same type of memory that is the problem in the first place. This could significantly increase erase/write cycles leading to an earlier failure.
My guess is that Tesla engineers have already tried to find ways to eliminate the problem or at least monitor it. They may have looked at telemetry using the existing over-the-air update system. My guess is that they haven't found a good solution. Otherwise, they would have submitted it to the NHTSA and might have received some form of relief from a recall mandate.
-
@ranwhenparked Honestly, I'd say no. The car's OS is baked into the same chips that run the MFD, so I'd think you'd have to come up with a whole new control system for the motors and battery, then interface it into that dashboard.
Unless someone's got an easier solution?
-
@roadkilled Well said. I guess we will have to see what Tesla and NHTSA's next steps are.
