Software is Eating the World-Tesla Edition

Last week Consumer Reports refused to recommend Tesla’s Model 3 because it discovered lengthy braking distances. This week Consumer Reports changed their review to recommend after Tesla improved braking distance by nearly 20 feet with an over the air software update!

Last week, after CR’s road test was published, Tesla CEO Elon Musk vowed that the automaker would get a fix out within days.

Until now, that type of remote improvement to a car’s basic functionality had been unheard of. “I’ve been at CR for 19 years and tested more than 1,000 cars,” says Jake Fisher, director of auto testing at Consumer Reports, “and I’ve never seen a car that could improve its track performance with an over-the-air update.”

…In retesting after the software update was downloaded, the sedan stopped in 133 feet from 60 mph, an improvement of 19 feet.

…The improved braking distances raised the Model 3’s Overall Score enough for the car to be recommended by CR

Tesla is also responding to other concerns raised by Consumer Reports. It’s quite astounding that Tesla is able to improve something as physical as braking distance with a software update and also astounding that they are able to update so quickly–even pure software firms don’t respond this quickly! Quite the win for Tesla.

The larger economic issue is that every durable good is becoming a service. When you buy a car, a refrigerator, a house you will be buying a stream of future services, updates, corrections, improvements. That is going to change the industrial organization of firms and potentially increase monopoly power for two reasons. First, reputation will increase in importance as consumers will want to buy from firms they perceive as being well-backed and long-lasting and second durable goods will be rented more than bought which makes it easier for durable goods producers not to compete with themselves thus solving Coase’s durable good monopoly problem.

Comments

If braking could be "improved" by software, that means they had "software limited braking" in place. That itself is a cautionary tale. A complicated one. Was this limit for anti-lock, anti-slip, anti-skid? Is the change in settings for a dry, clean, test track suitable for general release? Optimal for general release?

Fixing with updates means the poosibility of breaking with updates. How much is Tesla testing?

Is Consumer Reports going to test how easily a hacker can change the braking distance on your car?

Improve it's braking with OTA update.

The fact that a production car had a basic function-- like ABS, which is what this was, delivered with such poor performance is very troubling.

It indicates there was no design to a requirement, and/or testing to validate the performance had been achieved. The praise for being able to do OTA, is overlooking some pretty serious flaws in the design/manufacturing, validation and verification.

Ahh. It occurs to me now that an electric car has two modes of breaking: regenerative and traditional.

If you favor regenerative braking you don't stop as well but you recharge your battery better. If you favor traditional brake pads you break quickly but you capture less energy.

It seems CR got Tesla to favor traditional breaking in hard conditions and capture less power.

Presumably CR tested stopping distance by slamming the brake pedal to the floor.

In this scenario, the car should assume that stopping is the most important feature. Energy recovery is nice, but a completely floored pedal means it's a secondary concern.

The Model 3 has permanent magnet motors so the fastest stopping is with controlled electric stop, which can consume power. It will have brakes for emergency and parking, but electric drive will be better at stopping in that the motor is designed to handle the heat anyway, and it will have no wear. Plus, the control electronics will know if a wheel is slipping faster.

All Model 3s so far are dual motor 4 wheel drive.

And CR tests braking with multiple rapid panic stops, intentionally testing brake fade from overheating the brakes. This is a controlled way to test for braking going down hills. From back in the days before highways cut through hills and valleys were filled to eliminate steep roads.

CR claims to let the brakes cool fully between tests

"In this scenario, the car should assume that stopping is the most important feature."

Well, as we saw with Uber's killer self-driving, the programmers don't emphasize safety but rather suppress safety to avoid false positive braking.

The programmers have to make some tradeoff there, or the car will never move. Human drivers also make a tradeoff between false-positive braking and hitting people, since we are also affected by glare, fog, shadows, etc.

The obvious human example of false-positive braking as a strategy is slamming on the brakes when you're driving through a residential neighborhood and a ball rolls in front of your car. Most of the time, that's a false positive; once in awhile, there's a five year old kid running after the ball and you've just avoided running over him.

On the other hand, slamming on the brakes when you come over a hill and are momentarily blinded by glare from the sun is a terrible idea in most cases, because the guy behind you is *also* going to be blinded by the glare.

Speaking of software failures, shouldn't voice acquisition have gotten those brake and breaking things figured out better?

It is all AI, I'm told.

If (holy.shit.yelled == true) set brake.pressure = stop.right.now

+1

Just one highlight of the funniest and most entertaining MR comment section in some time

If (brake.pressure=stop.right.now) set windshield = marsh.mallow

From what I can gather it's more like braking is a complicated function involving tradeoffs with wear and tear, regeneration, stopping distance and steering ability. Tesla's update involved a reoptimization and improvement to their algorithms.

I think we agree on this page that software optimization is good. We are just a little suspicious of fast turnaround.

They probably had several alternate configurations already developed but had not decided which to implement.

If that is the case, isn't this more of a "small steps toward a much better world" story than a "software eating the world" story?

How much for someone to jailbreak a Tesla?

Why don't you let me try and optimize myself?

Surely Tesla's update represented a "reoptimization and improvement" to the algorithms ... knowing that the only part of braking that mattered was the specific one of stopping hard from 60mph. Has nobody heard of Goodhart's Law?

As others have said: In an emergency "brake to the floor" stop (which is what CR tests), I don't want a "complex trade-off with wear & tear, regeneration, and steering ability." I want my vehicle to stop.

My next-of-kin won't much care that the wreck around my corpse has infinitesimally more battery charge or more intact brake pads.

That Tesla didn't get this right the first time (but was quickly shamed into fixing it by adverse publicity) is less "software eating the world" than "bugs eating the world".

I think your larger "durable goods becoming service" is more on point.

I don't know that this is really true. Imagine that there is a stopping profile which involves burning your motors, frying your breaks, and bricking your car, permanently.

If you instruct your car - by slamming on those brakes as hard as possible - to kill itself, should it comply? Without confirmation and without compromise?

It seems like different people would have different answers to this questions.

A stopping profile which bricks your car is a defect. There is no reason any stopping profile should do so.

I find it hard to believe that any stopping profile would burn the motors. Regenerative braking is great. Relying on it in an emergency is...poor design at best.

A stopping profile "frying your brakes"? Any brakes that'd overheat so drastically in a single application are either mis-designed, or so ridiculously over-engineered that other component (like the tires) would fail long before the brakes did.

Hypothetically, I suppose it's possible to design hardware so fragile it needs to be protected from by software lest it be used at 100% of capacity. I assume and hope Tesla's engineers did not do that. That suggestion that maybe they did does not sound to me like much of a defense.

Sounds like putting a bow on a turd. Just cause the software can force the mechanics to do something doesn't mean those mechanics were designed to do that thing.....wouldn't be surprised a year later we hear that the braking mechanics give out because they were not engineered to tolerate what the software is now pushing them to do. The technology cheerleaders really need to take a chill pill sometimes.

We have pain and other sensations so we can develop the reflexes and counter-measures that protect from or mitigate harm. Non-tactile technology will probably end up killing us.

I'm half-joking of course, but only by half.

And then I just read this about a Tesla car hitting a parked police car while in autopilot. Tesla responded thusly "Tesla's semi-autonomous Autopilot mode has come under scrutiny following other recent crashes. The carmaker says the function is not designed to avoid a collision and warns drivers not to rely on it entirely." So their autopilot system IS NOT DESIGNED TO AVOID A COLLISION? What exactly is it designed to do then? Surely part of driving is avoiding hitting other stuff?

When "autopilot" had its first fatality I was critical of Tesla for having over-promised. If not in the fine print, at the high level, and in the "autopilot" name itself.

This has led to a tech-generation of "fans" who took the top level message as a promise, if not of "self-driving AI" now, at least soon.

And yet we here things like this:

"Looking at last year California DMV disengagement reports Nvidia car could not drive literally ten miles without a disengagement."

That's bad.

It's from a good article: AI Winter Is Well On Its Way

Self-driving cars as shitty robots.

Good link 10/10

(I personally would have been more confident had this change been regression tested for a thousand hours on a variety of surface conditions. Closed course. Professional driver.)

They are just sloppy. They intentionally limited the braking power before because they knew it could cause suspension damage. But they only have to survive to the next capital raise buy misleading the media. The rental car was made by the only Tesla employee who knows how to but the sound reducing material in the correct way.

Everything is insane. I work in software and it's ending up in everything and you can no longer tell what anything is supposed to do or fix it yourself.

The Tesla, at least, is actually made better by software. Does my fridge need to be smart?

Stallman is a crazy crank, but he's right about everything in our life being monitored.

Edit: here is the link.
http://nymag.com/selectall/2018/04/richard-stallman-rms-on-privacy-data-and-free-software.html

As much as I think GDPR will be a clusterfuck, it may be less of a clusterfuck than letting every single thing in the world spy on us / become part of a botnet / be subject to ransomware if we want to open the fridge.

Speaking of GDPR, a bunch of lightbulbs went out because the company couldn't get compliant in time: https://www.independent.co.uk/life-style/gadgets-and-tech/news/gdpr-latest-smart-home-devices-broken-internet-of-things-iot-a8369401.html

> Even pure software firms don’t respond this quickly! Quite the win for Tesla.

No, this shows either a) they had already tested this code but deliberately left the braking distance long for some reason (see e.g. Dan Lavatan-Jeltz above), or b) they just pushed new code to prod that affects a physical system with AT MOST perhaps a week or two of testing.

There's a reason software firms don't iterate this quickly-- insufficient testing is how inadvertent bugs appear!

If braking performance can be improved over the air, it can be degraded over the air. Let's hope Tesla are better at security than most car companies.

'they just pushed new code'

Version control is another fascinating subject, because as seen from some of the original reporting concerning the testing, it just might be that different cars had different software (package/optimization/etc.) installed. The results do seem to be all over the place, assuming that the Tesla self-reporting can be assumed to be accurate.

As a programmer, this did make me wonder. How on earth do you put something like this in your CI pipeline?

I suspect the issue is that all electrical vehicles and hybrids have a combination of regenerative braking and standard disc braking. The more that you use the discs, the more you're turning energy into waste heat instead of recharging the battery. Which is bad. The software would need to decide how much of which type of braking to use in different circumstances. It sounds like they dialed in more disc braking with the brake pedal mashed to the floor, which will generate shorter stopping distances but will waste some energy and hurt range a bit (which could be why they didn't do it in the first place).

Severely compromised braking, say 100 yards to achieve a full stop from 40 mph, is a small price to pay for a cleaner greener planet with less energy wasted on overheated disc pads.

You seem to have missed the point about hurting range, which is not a trivial consideration for a full electric vehicle.

Regardless of what the driver who no longer has the charge to reach their destination thinks about a cleaner greener planet.

A separate point is how good are the Tesla's brakes in another sense - since a regenerative braking system is used, it is quite possible that the braking system that is not regenerative is more sensitive to fading due to heat than a car with no regenerative braking at all, as those brakes are designed with such heat build up being part of normal operation.

Tabarrok: "The larger economic issue is that every durable good is becoming a service." Yea, right. Making a reliable car, washing machine, dishwasher, etc. is hard, it's easier just to redefine the poorly made durable good as something it isn't. Silly Con Valley is making fools of everyone.

Amen. I don’t need my fridge to be smart or to be sold as some SaaS con job.

I understand this doesn't address the "larger economic issue" but the "win for Tesla" seems anything but. Here's a company that quickly caved into a CR review, in order to get its seal of approval. Sure, Tesla responded. But does it truly make its cars safer/better? If so, why wasn't it in the first place, and if not, go to Lavatan-Jeltz: Tesla and Musk just need to survive. Gosh I wish, Tabarrok et al. could be more than just mere cheer-leaders to "technology."

So, just as my computer knows when I'm in a hurry to get some deadline hit, and decides to auto-update right then, can I look forward to my car being in the middle of a download when I am lot for a job interview?

Car Updating In Progress, File 2 of 354 downloaded, do not touch brake or steering...

Yes, I'm sure Tesla hasn't figured out that one yet

It wouldn't be the regeneration vs friction braking tradeoff, it would be steering control vs breaking rate. For example, if you go into a full skid, you have no control and if you going into a partial skid like most older anti-lock systems you lose some control.

If you make the system smart enough with information on direction changes and steering commands along with traction on all wheels, you can pick the optimal depending upon the specific conditions. When you take evasive actions, the breaking rate slows down allowing faster turning.

'it would be steering control vs breaking rate'

OK, time to actually look, because when this first came out, there was some actual information that would not seem to support this. Here it is - 'The magazine says that it let the brakes cool down between tests, but they couldn’t get a decent stopping distance after the first try:

In our testing of the Model 3, the first stop we recorded was significantly shorter (around 130 feet, similar to Tesla’s findings), but that distance was not repeated, even after we let the brakes cool overnight. Consumer Reports publishes a distance based on all the stops we record in our test, not just the shortest individual stop.

Tesla claims that based on its own test with the same tires, the average stopping distance is a significant 20-ft shorter. Car and Driver, meanwhile, gave the Model 3 a much better 70-0 braking report card although they noted:

Tesla barely ekes out a win in this category with its stop from 70 mph, although we did notice a bizarre amount of variation in our test, which involves six consecutive panic stops—the third of these stops took an interminable 196 feet.

Car and Driver also had a long max braking results from their own test – though still better than some competition:

A Tesla spokesperson sent us the following statement:

Tesla’s own testing has found braking distances with an average of 133 feet when conducting the 60-0 mph stops using the 18” Michelin all season tire and as low as 126 feet with all tires currently available. Stopping distance results are affected by variables such as road surface, weather conditions, tire temperature, brake conditioning, outside temperature, and past driving behavior that may have affected the brake system. Unlike other vehicles, Tesla is uniquely positioned to address more corner cases over time through over-the-air software updates, and it continually does so to improve factors such as stopping distance.' https://electrek.co/2018/05/21/tesla-model-3-braking-weakness-cr-tests-tesla-denies/

"It’s quite astounding that Tesla is able to improve something as physical as braking distance with a software update"

For some reason, it did not receive nearly as much coverage in the press, but I'll have you know that vee at Volkswagen long ago figured out how to improve something as physical as our diesel engines' emissions controls with some nifty software. Tesla is only now learning software tricks that vee at Volkswagen have long forgotten (and we're doing our best to make you forget it too).

This is what I was thinking. Tesla wrote a fix to match the testing conditions at the magazine.

"and second durable goods will be rented more than bought which makes it easier for durable goods producers not to compete with themselves" - except that in the presence of the threat of entry, selling creates an enforceable threat to vigorously compete with an entrant. It's not a priori obvious the Coase durable goods problem is a problem in the entry context.

Engineers understand, I think, that much of engineering involves tradeoffs. Meaning that most "improvements" are not free because most often getting more of "this" is obtained by reducing the amount of "that." And since Tesla's software is proprietary, I'd assume no one outside of Tesla really knows what was traded for that improved minimum stopping distance.

Apple famously (infamously?) traded reduced performance to improve battery life on some of its old phones, and received negative publicity for doing so. The question, perhaps, is the extent to which customers actually want "durable goods as a service," especially if they not only receive little or no knowledge of the tradeoffs involved but absolutely no choice on whether to accept the "upgrade."

Overall durable goods mostly seem less durable than they once were in that they not only wear out faster but are more costly to repair (if they're repairable at all). Which isn't necessarily nefarious, as there's little demand for a washing machines or refrigerators that last 25 years (at least if newer ones work better and/or cost less to use), and what would you do with that twenty-year old laptop anyway, even if it did still work?

Nonetheless, durable goods that can be "improved" at the whim of the manufacturer will inevitably produce conflicts of interest, even while raising that most basic of questions, "Who owns this thing, anyway?" Is the time ripe for appliances that require an annual license/maintenance fee if one is to continue using the appliance and, would the maker dare turn off the cold in my (or "my") refrigerator if I don't pay it?

Well, OK, perhaps it will start with an "extra cold and lower operating cost" option which I can decline, but, will this eventually lead to a world in which everything from a front door lock to an automobile ignition incurs a fee each and every time it is used?

as there's little demand for a washing machines or refrigerators that last 25 years

My parents just recently got rid of an old Maytag washing machine that was bought around 1971. The Apollo-era missions used washing machine timers in them so it could have been the same kinds of parts that went into space.

But there is an incentive for refrigerators, HVAC, washers, dryers, and dishwashers to last a long time: because they are super inconvenient to live without and super annoying to replace. I don't like losing my house for a day while workmen take doors off of my house to wheel in the new fridge and hook it up. Losing that day more than once a decade robs me of my life.

A laptop is easy to replace in comparison.

Durable goods as a service has already arrived in farm machinery. And it's a huge expensive problem.

https://www.forbes.com/sites/jasonbloomberg/2017/04/30/john-deeres-digital-transformation-runs-afoul-of-right-to-repair-movement/#7c5fdba25ab9

Indeed, I wonder what's the trade-off for the used car market. If you buy a 5 year old Tesla..... will you be allowed to repair it yourself or intellectual property laws will interfere?

And you have to accept all upgrades, for warranty and liability purposes, even if you don't like any changes in functionality that might accompany it.

PS. The answer is NO. You will not be able to repair it yourself. Never.

Actually, according to the current U.S. framework, the answer is yes - for a certain value of 'yourself.'

You can see this tug of war going on with John Deere, by the way - 'John Deere, Apple, Microsoft, Samsung, AT&T, Tesla, and the vast majority of big tech firms have spent the last decade monopolizing repair: “Authorized service providers” who pay money to these companies and the companies themselves are the only ones who have access to replacement parts, tools, and service manuals to fix broken machines; they are also the only ones who have software that can circumvent encryption locks that artificially prevent people like Schwarting from working on equipment. So people like Schwarting find enterprising ways around these locks by finding unauthorized versions of software or by hacking through firmware altogether.

But what started as hacking out of necessity has quickly transformed into a bonafide political movement.
Schwarting examines his combine. Image: Jason Koebler

Schwarting and other farmers across the country have found themselves on the front lines of the right to repair movement, the biggest people-versus-big-tech revolt in in recent memory. The goal of this movement is to ultimately get a law passed that will allow farmers, independent repair people, and average consumers to take back ownership of their tractors, their tablets, their cell phones, their air conditioners.

I met with farmers in Nebraska who are leading this movement, and the push is showing considerable momentum: 18 states are currently considering “fair repair” bills, which would require manufacturers to sell repair parts and tools to the masses, would require them to make repair manuals available to the public, and would require them to provide circumvention tools for software locks that are specifically designed to prevent third party repair.

“The Fair Repair act gives an individual the ability—you’ve always had the right—to purchase the diagnostic tools or to take their equipment somewhere local, or to try and repair the equipment yourself,” Lydia Brasch, a state senator who is sponsoring the bill in Nebraska, told me.' https://motherboard.vice.com/en_us/article/kzp7ny/tractor-hacking-right-to-repair

I come at this from an industrial software controls point of view.

Good design assumes that Tesla couldn't just change the setpoint of the braking parameters at random (nor could some hacker). Normally a design parameter will have an engineering high and low setpoint and the parameter can be changed between those two parameters. To change the actual setpoints would require a much more invasive firmware update.

Generally speaking the high and low setpoints should have undergone engineering analysis and testing and any parameter within that range should have been safe to some predetermined threshold.

And here is a Tesla that does not bother with an over the air update to end up doing this, instead of braking - 'A Tesla Model S that crashed into a fire-department vehicle in South Jordan, Utah, while operating on Autopilot earlier this month accelerated for three to five seconds before the crash, according to data from the vehicle cited by the Associated Press on Thursday evening.

That is a new development amid the investigation of the May 11 crash, in which the driver of the Tesla suffered a broken ankle. The National Highway Traffic Safety Administration is investigating.

According to the AP's report, the police suggested that the Tesla, which had been traveling at a preset speed of 60 mph, was behind another car that had slowed down to 55 mph. The Tesla slowed down as well, the report says, but when the leading car changed lanes, the Tesla accelerated to regain its speed shortly before impact.' https://www.businessinsider.de/tesla-model-s-sped-up-before-crashing-in-utah-autopilot-2018-5?r=US&IR=T

Maybe braking is not the most important concern for Tesla? Particularly as the car seems to have a certain propensity for hitting stopped emergency vehicles - 'If confirmed, the incident will mark at least the third time a Tesla in autopilot has crashed into a stationary emergency vehicle since January. This month, a Tesla Model S crashed into a stopped firetruck in Utah, echoing a similar incident in January, when the same model collided with a stopped firetruck on the 405 freeway in Los Angeles county.' https://www.theguardian.com/technology/2018/may/29/tesla-crash-autopilot-california-police-car

I personally am shocked that someone got a broken ankle on the way to fully autonomous vehicles. That would never happen in a manually operated automobile.

You left out the fire truck, which was apparently out on a call, with the firefighters having something else to do with their time than deal with the totalled Tesla and its luckily lightly injured driver.

The pattern is how often Teslas seem to hit vehicles that are already engaged in dealing with another problem - which is why emergency vehicles tend to be stopped at the side of the road in the first place.

If this helps incels get laid, then I'm all for it.

Interesting. Aside generic cheer-leading for a fast turnaround, there's very little good to say or speculate about this story. And even the fast turnaround itself is particularly suspect on several levels.

"It’s quite astounding that Tesla is able to improve something as physical as braking distance with a software update and also astounding that they are able to update so quickly–even pure software firms don’t respond this quickly! Quite the win for Tesla."

Sort of. The thing is, stuff like stopping distance is determined by weighing numerous trade-offs in design. So, given that there were trade-offs, what were they that lead Tesla to picking the original stopping distance, and what are they now with the new one?

Nothing is free - otherwise Tesla would have set this stopping distance to start with. So what are we losing to get it? And, given that its an OTA update, and with knowing the trade-offs, why couldn't owners be allowed to pick what they want (within the limits of the hardware and legal mandates)?

Good point. CR will probably simply report "problem solved" without examining what was traded-off for this newly acceptable result. Unless the old software was just blatantly faulty, which wouldn't be that surprising given that Tesla is basically doing its beta testing on new car buyers, much more so than most other carmakers.
As for the point about consumers wanting to buy durable goods backed by long lasting firms providing ongoing service - it seems to me that consumers are taking software updates for granted, but it's not always the old names that provide it. Oppo broke into Blu-ray players providing better service and updates than say Sony. The cost of adding wifi and some software programmers has been going down over the years. But I would hate to wait around as much for my car to download and install new software updates as often as I have to put up with such updates on my computer and phone.

It seems like most of this is an attempt to lock people in place, married to a single source model where you never really own the thing, but can't escape to another either.
Not that it will necessarily be successful, just that seems to be the intention.

What this means is that a hacker, disgruntled Tesla employee or Telsa itself, can change the safety handling characteristics of the vehicle on the fly, leaving you the victim at their mercy.

Tesla makes a crappy car. Eh. What's amazing is how easily Tesla could shift the attention from a poor quality car to software. We've suspected that Americans are stupid, but they seem to go out of their way to prove they are stupid. A car is a car. Duh.

It seems to me that a car manufacturer should be more stubborn than this when it comes to something as fundamental as braking. After all, who employs more automotive engineers, Consumer Reports or Tesla?

"The car should stop in an acceptable distance at highway speeds" is not exactly an exotic requirement. So what does it mean that 152 feet was fine yesterday, but today it's 133 feet or bust?

Did Tesla not do their own testing? Was there nobody in charge of making sure that the car had acceptable braking performance? Did nobody define what acceptable braking performance was?

" It’s quite astounding that Tesla is able to improve something as physical as braking distance with a software update"

The only thing that is astonishing here is that Tesla does not optimize for braking distance in the first place. They are not a car company but a software company. There is an old joke asking if you really would wanna buy a car from Microsoft or another software company given their track record to build stuff that crashes twice a day.

If software takes all of our jobs, we have nothing left to do except watch Roseanne.

The software update concerned the ABS (anti-lock braking) system. On modern cars the ABS system monitors the speeds of the wheels and pulses the brakes on/off to keep the wheels from locking up. Tesla's system can adapt on a time scale of about 2 milliseconds, most other cars are in the 100 millisecond range. Changing the pulse frequency and/or pulse pattern can have a pronounced effect on braking distance.

ABS explained:
https://en.wikipedia.org/wiki/Anti-lock_braking_system
Tesla manual, ABS page:
https://www.manualslib.com/manual/913720/Tesla-Model-S.html?page=52

also:
(1) regenerative braking is not relevant to emergency braking
(2) all Model 3 sold to date are RWD
(3) Motor Trend got 119 ft for the Model 3, about the same as they get for an A4 or an Accord
http://www.motortrend.com/cars/audi/a4/2018/2018-audi-a4-ultra-premium-vs-2018-honda-accord-20t-touring-comparison-review/

Braking distance? What braking distance. As the recent case showed that some company will rate ride comfort above pedestrian life and jerky braking, in fact no braking at all, even in emergencies.

https://techcrunch.com/2018/05/24/uber-in-fatal-crash-detected-pedestrian-but-had-emergency-braking-disabled/

If the design engineers are smart they would have written memos and had the bean counting managers to sign off the factory settings.

Most comments don't talk about the scariest portion of Alex's comments: the long term consequences on durability. Manufacturers can enforce obsolesence by guaranteeing your car only works for x years. And if for example, they say that safety or environmental concerns justify the choice of x, they can make it so that you can't keep your car longer even if you want to since they can stop updates, etc.

"they can make it so that you can't keep your car longer even if you want to since they can stop updates, etc."

That's always been true. Indeed, there's an urban myth that auto manufacturers must produce spare parts for X number of years. Usually, it's stated as 10 years. But it's not true. Manufacturers must meet their warranty agreement and they don't have any obligations beyond that.

Is there a significant difference between not making spare parts for a model and not providing updates?

Of course 3rd parties will step in and start producing spare parts for older vehicles and I would assume 3rd parties would step in and start producing critical software updates for vehicles in the future.

It's certainly not true that a vehicle manufacturer could just intentionally "brick" a model of cars without facing repercussions. They would immediately be subject to a class action lawsuit. Just as if a software seller tried to brick an older piece of software to force a user to pay for upgrades.

Tesla's problems are many but its 'wins' are few. Eventually, Musk will run out of cash that he can borrow at favourable rates and will discover that free cash flow matters no matter how he chooses to update the software in his vehicles. Tesla is doomed and will default on its loans. Why economists don't discuss that instead of playing at the periphery is puzzling to this individual.

"The larger economic issue is that every durable good is becoming a service. " No need for a recall, just a simple software upgrade. This is really amazing. We are living through another phase of the industrial revolution, like the assembly line.

Comments for this post are closed