The existing software is to be removed, replaced, recalled - that Telsa is choosing to simply promise an (as yet) undefined new version that will magically solve what Musk has been failing to deliver for ten years now does not change the fact that the current software must be taken off the cars as it appears to be unsafe. That is a recall by any other name.

It'll be interesting to see what the response is if Musk simply releases a new version of the same software that (like every previous release) merely makes incremental changes rather than fundamentally solving the problems that 'Full' self driving actually has.

As a defense "no-one has died" seems to be a depressingly low bar.

Yet no-one's got around to fixing that...

It used to be standard on BMWs, until the wetware got fed up with the complaints and migrated to Audis. Good luck with recalling that!

"That is a recall by any other name."

Recall = bring it back. To change it, replace it or refund it.

Update = it stays with the customer, and will be changed in situ .

Ergo not a recall. Musk is correct here.

Disgree. In the industry and among the customer base:

Recall = we've found something unsafe in the product and we need to fix it quickly.

Update = we're making some improvements to functionality and it ought to be fixed when convenient.

Updates are done on the hoof, often without the customer knowing or as part of a service. I recently took my car to the garage to have the boot catch fixed and they asked me to empty the door pockets because they were going to change some door wiring as part of an upgrade. It needed doing, but it wasn't dangerous and didn't merit a recall or even a letter to ask me to bring it in at my convenience. It would have been done at the next service if I hadn't brough it in for the boot.

Musk might be correct pedantically, but I bet he's more concerned about customers' (correct) interpretation of the word "recall".

Absolutely this - while 'recall' has historically meant physically bringing things back to the supplier, the real core of the issue is that the product was defective and had to be fixed or replaced. If the software has to be fixed or replaced because of safety reasons, it's a recall even if the cars don't need to be physically returned.

An update is merely functional

Context is everything or, if not, at least substantially more important than dictionary definition.

A vehicle that, for some reason, could not handle an OTA update would *have* to be returned. Similarly, a "recalled" vehicle could conceivably be remedied by a visit from a mobile mechanic.

It is for the relevant governmental organization which mandates "recalls" to decide whether something is a "recall" or not, not any given manufacturer or vendor.

" travelling straight through an intersection while in a turn-only lane, entering a stop sign-controlled intersection without coming to a complete stop, or proceeding into an intersection during a steady yellow traffic signal without due caution. [...] the system may respond insufficiently to changes in posted speed limits "

These being utterly elementary examples of hopelessly lousy driving, how can anyone assert that "autonomous" vehicles are safer than human-driven ones (even at the low level of "autonomy" currently implemented)? A software update ain't going to fix the fundamental problem -- that the automaton doesn't understand anything: it's purely reactive to stimuli that actually mean nothing to it. That's the antithesis of what Chesley Sullenberger described as the situational awareness essential for safe piloting in the much less crowded sky.

Self driving cars (could) be safer than humans because they don't get distracted texting, are never tired, and are never drunk.

However I'm not sure if the Tesla FSD system is safer than the average driver, because Tesla are very coy about the data they do release. The standard lane hold autopilot system *does* seem to be safer when comparing highway only driving to highway only AP.

If it's "safer than a human driver", then why does it quickly need to hand over control to a human when things get dicey (i.e. just before things go to sh*t)?

Oh, wait, I know that one: so when an accident happens it's not FSD that gets the blame, even though it may have caused the situation in the first place.

As for releasing statistics and data, what I would *really* want to see is the 15 seconds of data and video before and after an accident. I suspect we won't get that because it would show just how brutally those statistics that Musk is so fond quoting were massaged.

Wouldn't that be Tesla's approach to what they call Autopilot/FSD? It's pretty obvious it's already got a huge amount of technical debt, they're having great difficulty adding more options, and they're trying to make up for it with manual mapping (employees look at images taken by the car and tell the software what each sign is). And Musk is still too stubborn to go back to Lidar and Mobileye because he prefers to promise the moon when it comes to self-driving.

There's a rumour v4 hardware will at least now incorporate radar. Still not LIDAR, though, because I think the manchild has a personal problem there. Screw safety, Musk is unhappy. Quick, promote some more tweets!

Instead of quibbling about what 'recall' really means, maybe Musk should first understand what 'full self-driving' means. Hint - It doesn't mean what Musk/Tesla say!

I have a Tesla. Great car but they are a million miles away from full self driving and should say so. Doesn't even know the speed limit for a significant proportion of the time.

"entering a stop sign-controlled intersection without coming to a complete stop"

As a European, I have never understood this 'come to a complete stop' malarkey. We have stop signs of course but usually linked to traffic lights, while most junctions use 'give way', not 'Stop'. Arriving at a 'give way' junction requires a driver to slow down to a speed slow enough that they can safely come to a complete stop if they have to give way to another vehicle at the junction. If there is no other vehicle at the junction and the driver can safely proceed, it is a waste of both time and energy* to come to a complete stop and then immediately accelerate away.

My (to be fair quite limited) experience of intersections in US roads is that there are vast open spaces around and it's quite easy to see oncoming traffic, so the widespread use of stop instead of give way (and the almost complete lack of roundabouts) makes US roads weird. What I would 'import' to Europe is allowing right (left n the UK) turns on red.

*It might seem trivial as a couple of seconds and a few grams worth of fuel, but multiply that by the number of stop signs on a journey and the number of vehicles and it builds up to a significant amount.

"What I would 'import' to Europe is allowing right (left n the UK) turns on red."

I wouldn't. It's already dangerous enough trying to cross roads in this country even at crossings. There's already enough bad driving habits around red lights, and allowing drivers to go through on a red when pedestrians are supposed to be crossing is a fundamentally bad idea.

(Completely agree on the rest of it, though. I never got the hang of 4-way stops...)

Agreed, free right turns on red are extremely dangerous to pedestrians and only make sense in the notoriously pedestrian hating US. Anywhere civilized it's a (usually) terrible idea. The exceptional cases where it makes sense, most countries have signs or lights to indicated right turns on red light are allowed or a slip lane is put in.

Americans just overuse stop signs. They came about in the 20's when cars had poor brakes and tyres so the thought was that it would never be okay to 'roll' at a low speed through a stop sign because it could mean you couldn't stop in time for that jaywalking pedestrian. And in typical American exceptionalism they refuse to change with the times. Roundabouts and give-way junctions are far better solutions to low traffic crossings, but these are rare in the US (more common nowadays, but still much rarer than the n-way stop).

You'd think that would have changed with the invention of screenwipers..

:)

There are junctions with very limited visibility, so you have to stop, listen and then slowly proceed to gain more visuals and then go if it is clear.

There are junctions where you think you can see clearly and there is no point to stop, but this may be an optical illusion.

Stop signs are not placed because someone had a bad day at planning centre. It's because many people had a bad day at the junction.

4-way stop sign intersections are the default in the US in my experience. Whether or not there is good visibility, anywhere there's a 4-way intersection that doesn't have traffic lights it's nearly always a 4-way stop-signed intersection. Because screw having to actually do any traffic planning. Just make it all 4 way stops and it's good.

It all just smacks of arrogance on Elon Musk's part. He thinks that because he decrees it so, Tesla's self-driving cars will be the way forward. Doesn't anyone realise that by removing tasks from the driver to the point where he or she is merely a supervisor will mean that their driving skills will actually atrophy?

In the same way that critical thinking skills do when you're an idiot who sacks everyone who has their own opinion?

Yup.

Let's not confuse the issue, which is all about the preposterous slander imposed by the word "recall".

That is what must be changed and Musk will die on that hill if necessary (aka send unlimited amounts of salesdrones to die in his place until the issue is snowed over by whatever important news is churned up next - look, squirrel !).

Go park your TechnoKing arse down at the Tesla software lab and apply some of your famed TechoKing software engineering skill to fixing the problems. The buck (and the body count) stops with you

Even Tesla’s basic AutoPilot still fails to lower the vehicle speed to the new limit before reaching the sign.