Hacker News Reader

Comment by happytiger | Hacker News Reader

I’m confused by sf. They did this for Uber in 2016, Amazon’s zoox, and a host of other companies, etc.

Tesla has a safety record that’s not unimpressive compared to the 36k people who die every year even though their failures have been more public and newsworthy. Their stats have been relatively objective.

We also have a framework for this: https://www.nhtsa.gov/technology-innovation/automated-vehicl...

As there is a framework in place already, shouldn’t we push forward? As long as we are reducing the statistical risk of driving below the risk of human drivers, shouldn’t that be the litmus for whether these programs are allowed to progress? I know I am being simplistic, and I understand that no statistical model will ever make up for the loss of a family member, the generally accepted foundation for the morality of self-driving car technologies hinges on whether they can beat the human-driven statistical equivalent.

If there is concern about that they shouldn’t continue. But if there isn’t, these experiments should progress as they could ultimately reduce mortality.

There were 3,723 deaths in California in 2020. The lifetime odds of dying in a motor-vehicle crash for a person born in 2019 were 1 in 107. These odds suck. Is that really acceptable? Can we do better? We know we can. If we can reduce that number we should.

If we can’t, we should restrict this technology’s rollout until it can.

Am I overly objective?

I’m not an expert.

I mean ultimately I want legislation built on objective data and loving values with human compassion at the core. I feel like sometimes we see these decisions based on how fast the system can adapt the ocean of existing automotive and roadway regulatory compliance to the technology changes, rather than new realities. People have no idea how much the books and legal cases that many cal trans engineers have spent their lives developing stand in the way of self-driving vehicles. I honestly think the biggest issue is probably the legal a s regulatory overhead of responding to the technology rather than the technology itself, but since it’s so hard to quantify I don’t often bring this up. But the biggest passion I have seen if from those same engineering minds, who truly understand the complexity, but also want to see it happen. Their timeframes are just long. I’ve heard 2045 and 2050.

Anyone more expert than I care to weigh in? I’d love to learn more. I am not trying to weigh in with any merit — I am a nubile entrant and simply sharing the output of conversation with people who will always be more qualified and I, but I’d love to learn more about this.

View on HN

>Tesla has a safety record that’s not unimpressive compared to the 36k people who die every year even though their failures have been more public and newsworthy. Their stats have been relatively objective.

They haven't released a detailed deep dive on their statistics, other than some definitions of what counts as a crash. This is from the company headed by the "funding secured" guy. Elon is clearly willing to lie if he believes that it will serve his interests or mission.

Even if we accept their safety stats at face value, Tesla is getting ready to launch a new feature - "Full Self Driving" - which behaves substantially different from autopilot and has a much different operating design domain and intended functionality. They've had close to 2k internal testers and O(100) external testers of "Full Self Driving" for about a year - nowhere near enough time and data to get the same level of confidence they have for autopilot. Add on to this that it will operate in environments that are substantially more accident prone than highway driving or clearly marked single lane roads, and you can see why regulators might be worried by a larger rollout.

>This is from the company headed by the "funding secured" guy.

Even worse he's the guy willing to falsely accuse someone of being a child rapist because that person pointed out he had simply tried a publicity stunt that failed, badly, and his tech solution to a cave rescue with a huge media focus was actually worse than useless.

In other words when correctly and honestly called out for sprouting bullshit Musk goes fully delusional in the nastiest way possible. It's pretty worrying behaviour. He may well go on to do amazing work, i don't know either way on that and I don't have to have an opinion on it either. Surely nobody sane would never trust his word on anything for any purpose ever again after that display. I'd expect from a five year old.

> In other words when correctly and honestly called out for sprouting bullshit Musk goes fully delusional in the nastiest way possible.

It's called narcissistic injury, a behavior associated with narcissistic personality disorder.

Or perhaps.. aspergers? Which Elon does have. Yall are placing way too much emphasis on an outburst of an autist with a medical disorder.

There’s a pattern of deception on his part that is way more regular and extends beyond that insult that can’t be attributed to Asperger’s - I didn’t event mention the pedophile insult in my parent comment.

> Surely nobody sane would never trust his word on anything for any purpose ever again after that display.

TSLA is one of the most widely held retail stocks. Why do you think this is? Might you be assigning too much weight to this?

> In the short term the market is a popularity contest; in the long term it is a weighing machine

Warren Buffett

Elon is undoubtedly skilled at driving up hype - the popularity contest part. As for long term - Tesla themselves claim they never plan to pay dividends.

> Even worse he's the guy willing to falsely accuse someone of being a child rapist because that person pointed out he had simply tried a publicity stunt that failed, badly, and his tech solution to a cave rescue with a huge media focus was actually worse than useless.

The impression I get is that was just a silly attempt at tit for tat. E.g. guy says "You're doing X for Y bad reasons!" while not having any evidence that you're doing things for Y reasons, so you shoot back with something in the same vein. Y being either "getting publicity" and "being a pedo" respectively.

Yeah, they're not really the same severity of accusation/insult (particularly as Musk has a bigger megaphone), but raising the stakes is common if someone starts a fight with you.

Also note: "Richard Stanton, leader of the international rescue diving team, urged Musk to continue construction of the mini-submarine as a back-up, in case flooding worsened."

"So what if I burnt him alive for calling my hairstyle stupid? It was just tit for tat!"

> There were 3,723 deaths in California in 2020. The lifetime odds of dying in a motor-vehicle crash for a person born in 2019 were 1 in 107. These odds suck. Is that really acceptable? Can we do better? We know we can. If we can reduce that number we should.

This is the crux of the argument. However, it falls flat when you consider that many other simple "can reduce that number" measures have not been adopted – like fitting speed governors to cars limiting them to some speed limit, putting tracking beacons in cars to making them automatically stop at red lights, or putting in devices in cars to limit drunk driving.

I don't see a reason why a much more complicated technology should leapfrog these simpler ones and be put into production so quickly, potentially risking innocent lives in the process. In fact, I believe the reason why it is being rushed so quickly is arguments like yours coupled with a low understanding of the technology involved. People are viscerally more willing to accept the idea that an "artificial intelligence" is going to drive the car, rather than the idea that their car is a multi-ton hunk of metal that needs to be limited and monitored carefully.

> I honestly think the biggest issue is probably the legal a s regulatory overhead of responding to the technology rather than the technology itself, but since it’s so hard to quantify I don’t often bring this up. But the biggest passion I have seen if from those same engineering minds, who truly understand the complexity, but also want to see it happen. Their timeframes are just long. I’ve heard 2045 and 2050.

Those timeframes seem normal to me. Rollout dates should not be decided by the whims of corporate shareholders. Let's put simple speed limiters on cars first, everyone can understand that technology – and see how that goes!

This is too simplistic thinking. For example a basic speed limiter can be very dangerous when overtaking slower cars, where it might be dangerous in some situations to not be able to go above the speed limit for a short moment.

If there’s no way to safely overtake a slower car without breaking the speed limit, then just don’t. A self driving car shouldn’t decide to do that either. Also, a speed limiter could easily be designed to allow exceeding the speed limit for a short moment.

Ideally true, but not true in practice. When you're on the open road and the speed limit is 55mph, but everyone is going 80mph+, then anyone going the speed limit is a hazard. The speed limit is just an abstract notion at that point. Sure you may be violating the law, but I'm more worried about laws of physics at that velocity.

You'll have people passing you on the left and behind you at +30mph. The cars behind you in your lane will form a queue of waiting to pass you, and they'll be getting into the right lane needing to accelerate from 55mph to 80mph (hard for a lot of cars). This is going to cause the cars in the fast lane to brake, which will cascade through the lane and cause a traffic slowdown. If someone isn't paying attention, it's going to result in a rear end collision and an hours-long traffic jam for hundreds of people.

This is one of my worries with self driving cars: They'll follow the letter of the law even when it's not practically safe.

Which is exactly why they are proposing to install speed limiters in all cars, which would address the first part of your argument.

what are the logistics for this? There were 263.6 million cars in the US in 2015 (source: https://en.wikipedia.org/wiki/Passenger_vehicles_in_the_Unit...), so I'm sure there's even more today. I could see having manufacturers install speed governers in new cars, but how would you coordinate installing in all the "old" cars? There'd be a period (probably years) where some, but not all cars don't have that limitation.

> When you're on the open road and the speed limit is 55mph, but everyone is going 80mph+, then anyone going the speed limit is a hazard

Surely it's the ones doing 80mph that are the hazzard

No it's not. https://en.wikipedia.org/wiki/Solomon_curve

Any moving vehicle is a hazard, the faster it's moving (relative to the road) the greater the hazard

A stopped car won't crash. A moving car will.

No, it's the opposite in a situation where people are mostly going 80 and only a few are doing the speed limit. Try traveling down interstate 95, this is an every day situation.

Look at traffic as a fluid: A stream is calm when the water flows at the same speed and direction, when there is nothing to get in the way. Put a rock into the stream and now water has to move around it, causing flow patterns that collide. Traffic can be looked at as a fluid in the same way. If you want it to flow without friction, it all needs to go the same speed in one direction. Accidents and traffic slow-downs are caused when people merge and people have to slow down unexpectedly, which is going to happen when you're going 30mph under the prevailing traffic speed.

Yes, the person going 55 is following the law, but the law was written in a faraway room, a long time ago, by people not driving on the road in that instant. Sometimes in the instant, following the law is not the safe thing to do. It's the same reason it's not safe to go 25 in a 55, and that trucks are required to put their flashers on if they're going under a certain speed. Because they hare a hazard if they are going too slow.

> trucks are required to put their flashers on if they're going under a certain speed.

This is not correct. Many states in the US do not allow the use of hazard lights while moving.

See? You can visualize flaws with these kind of simple systems fairly quickly, don't you think similar ones exist in the purportedly "self-driving" cars?

There are literally millions of situations where a Tesla might do the wrong thing, and some of them have been caught on camera.

If we can't put speed limiters on cars, we definitely shouldn't be putting highly complicated computers in charge with no humans at the wheel.

> If we can't put speed limiters on cars, we definitely shouldn't be putting highly complicated computers in charge with no humans at the wheel.

This argument makes no sense.

How do obvious edgecases invalidating a simple solution have any bearing on the feasibility of a complex solution being able to handle as many edgecases as a human driver?

I think the point is more of a bikeshedding problem: people are more willing to accept “AI” because they are not knowledgeable enough to come up with a counter-example, whereas given a simple automation like “don’t go above the speed limit” any joe on the street can think of why that’s a problem.

But AI is not magic, its just more rules.

Trucks have been running with governors for decades. That has reduced truck accidents, not increased them. It makes for some idiotic overtaking behavior every now and then (when one truck that goes 89 km/h overtakes a truck that goes 88.5, and it takes forever and then some), but that's a nuisance, not a danger.

> That has reduced truck accidents, not increased them.

How did you determine cause and effect here? There have been a LOT of changes to trucks, roads, signage etc.

One of my first cars 'lost' the highest gear (5th) after a couple of years.

You adapt and do less takeovers and only when it is absolutely safe.

I'm not sure in what scenario you would use your highest (ie slowest accelerating) gear if wanting to do a fast overtake...

My car has a natural speed limiter (the engine is only 56 hp), and it just means I don't try to do overtaking maneuvers that would be dangerous. If I had a supercar, I would try many more dangerous overtaking maneuvers. Which is safer?

> For example a basic speed limiter can be very dangerous when overtaking slower cars

Eh I have a speed limiter I use quite often just to make driving through average speed areas more chill, and it doesn’t stop me pushing the throttle and getting up to full speed if I need to.

It’s a bit smarter than “never go over the speed” - if you push the accelerator in a definite way the car will still respond, while if you hold your foot fairly still and only accelerate gently it will cap out at the max.

This is the absolute state of this debate. You can not overtake if doing so requires exceeding the speed limit.

The best part is the same ignorant people develop the driverless cars, so we get Google cars that turn right across the bike lane instead of merging into it.

You can not overtake if doing so requires exceeding the speed limit.

Unless you are driving in the state of WA:

https://app.leg.wa.gov/rcw/default.aspx?cite=46.61.425

But people want to buy self-driving cars. And nobody wants self—limiting cars. So it’s not really a case of leapfrogging anything

It’s not like manufacturers are waiting on the government to let them sell limited cars

Or get this, we could walk and never get into a car crash!

I take it you don't live in an American suburb then?

I think it was sarcasm

It’s baffling to me that robots are allowed to break the speed limit.

> I’m confused by sf. They did this for Uber in 2016, Amazon’s zoox, and a host of other companies, etc.

My understanding is that those companies had trained test drivers, but Telsa is pushing this out as a beta feature update and allowing the consumers be their test drivers?

That is a critical difference for sure.

> trained test drivers

Are normal people paid minimum wage who become complacent and occasionally fall asleep (e.g. Uber accident), I would argue non-paid Tesla customers are at least as good as "trained drivers". There are 2K drivers using FSD beta for almost a year, w/o any accidents.

100 times this. Tesla Beta testers are people driving their own cars, and are - by selection of being able to afford a Tesla - in average more qualified then the minimum wage test driver.

The word "trained test driver" is misleading. It sound like "test pilot", which are usually the best in their field. But paid test drivers are just people that applied for a low pay job. Heck, they maybe even be tired all the time because they have to have two jobs.

> by selection of being able to afford a Tesla - in average more qualified then the minimum wage test driver.

How does “income received” translate into “qualified for driving”? A good lawyer, doctor, it consultant, … is probably on average a less safe driver than someone that has been through a training, despite earning more.

If your worry is that the test drivers are tired because they need to work multiple jobs - mandate better pay. And from what I know from my doctor acquaintances, being tired from overwork is part of the job description.

One could argue that young drivers are disproportionately likely to get into accidents (inexperience, bravado...) and are also less likely to be able to afford a Tesla.

But even that is not really sufficient for supporting the statement either.

> are - by selection of being able to afford a Tesla - in average more qualified then the minimum wage test driver.

You think having money qualifies someone for being a good test driver? This is one of the most classist comments I've ever read.

A responsible company would hire people based on some criteria of competence before putting them in the "supervisor" seat for an AI driving assist. Even an irresponsible company like Tesla or Uber will still do more due diligence for the competency of their test drivers than for their customers, whose only qualification is a wallet.

> A responsible company would hire people based on some criteria of competence before putting them in the "supervisor" seat for an AI driving assist.

Yeah, but this does not happen.

Source: How much is a test driver paid?

Exact wages vary at different companies. Cruise, the self-driving arm of General Motors (GENERAL MOTORS), has advertised positions that pay $23 an hour. Meanwhile, Waymo, the self-driving entity of Google's parent company, has posted jobs offering $20 an hour.

Uber declined to disclose how much it pays its safety drivers.

https://money.cnn.com/2018/03/21/technology/uber-test-driver...

Why does the salary indicate they are not chosen for their ability to do this job well? Do you also believe people who own a Tesla would be better at serving food than most waiters because they have higher income?

I just bought a car that costs more than the highest optioned Tesla.

According to you, Tesla should be hiring me for FSD beta testing, according to "selection".

I don't even know how to respond to your assertion.

>> I know I am being simplistic, and I understand that no statistical model will ever make up for the loss of a family member, the generally accepted foundation for the morality of self-driving car technologies hinges on whether they can beat the human-driven statistical equivalent.

It's not about emotions; it's about statistics.

The problem that needs to be overcome is not the emotions of people losing loved ones, but the difficulty of accurately estimating safety by test-driving autonomous cars.

The following is from a Rand Corporation study titled: "Driving to Safety"; subtitle "How Many Miles of Driving Would It Take to Demonstrate Autonomous Vehicle Reliability?"

Key Findings

Autonomous vehicles would have to be driven hundreds of millions of miles and sometimes hundreds of billions of miles to demonstrate their reliability in terms of fatalities and injuries.

Under even aggressive testing assumptions, existing fleets would take tens and sometimes hundreds of years to drive these miles — an impossible proposition if the aim is to demonstrate their performance prior to releasing them on the roads for consumer use.

Therefore, at least for fatalities and injuries, test-driving alone cannot provide sufficient evidence for demonstrating autonomous vehicle safety.

Developers of this technology and third-party testers will need to develop innovative methods of demonstrating safety and reliability.

Even with these methods, it may not be possible to establish with certainty the safety of autonomous vehicles. Uncertainty will remain.

In parallel to developing new testing methods, it is imperative to develop adaptive regulations that are designed from the outset to evolve with the technology so that society can better harness the benefits and manage the risks of these rapidly evolving and potentially transformative technologies.

https://www.rand.org/pubs/research_reports/RR1478.html

Bottom line, Tesla's (and everybody else's) "safety record" isn't and it will take a lot more time to know whether self-driving cars are as safe as human-driven cars. The stats you quote are not sufficiently informative.

Fully agree, there are simply too many factors, all I know, a fsd shouldn't hit stationary vehicles.

Besides, i don't even see the value proposition for fsd other than commercial trucks maybe.

If people want to be driven around, maybe a taxi or public transport would be more economical and better for congestion and the environment in general. Not sure what people want to do while being driven in a fsd car. A human to human discussion is already possible. Watching a movie? Read a book? Sleep? All these are not practical in a driver's seat.

That's silly. We'd never release any product with ridiculous standards like "Test for 100 years".

Original horseless carriages didn't have any rules like that. We liked them anyway. That seems a more realistic model of adoption, than ivory tower perfect-safety pipe dreams.

>> We'd never release any product with ridiculous standards like "Test for 100 years".

The same timeframe does not necessarily apply to other kinds of systems. The study is about self-driving cars and the numbers quoted are about self-driving cars, in particular self-driving cars compared to human-driven cars.

The study starts by defining various industry-standard measures of reliability, for example scroll down to page 3 and look at equation 1 and equation 2. You can see that those are not about "perfect safety" but about measuring the probability of an accident and so on.

As a general rule, if you are confused by the text of a study, check out the maths. They should be unambiguous.

Sure, but the conclusion of the post was something like "we can't test these things enough". Which seems like a deliberate misrepresentation of the math as well.

I'm sorry, which post do you mean? Who said 'something like "we can't test these things enough"'?

Original horseless carriages had enough weird rules: https://en.m.wikipedia.org/wiki/Locomotive_Acts

Edit: it took almost 4 decades to improve them. We are starting with driverless vehicles just right now.

Tesla is resolving this by releasing a great many vehicles with the FSD systems installed and running and feeding data back for review, while not actually operating the controls - result is to observe how the FSD would have behaved under billions of miles of driving, noting where real drivers differ from automation and why, without any real accidents attributable to on-parole systems.

> The lifetime odds of dying in a motor-vehicle crash for a person born in 2019 were 1 in 107. These odds suck. Is that really acceptable? Can we do better? We know we can. If we can reduce that number we should.

This is like the old shitty political saying “Something must be done! And this is something, therefore this must be done!” It sucks that driving is so dangerous. That doesn’t mean letting tesla do whatever they want will help improve things. As others have commented, the data on tesla’s safety is apples-to-oranges. It doesn’t mean much of anything.

> the generally accepted foundation for the morality of self-driving car technologies hinges on whether they can beat the human-driven statistical equivalent.

On a side note, legally if a Full Self-Driving system kills someone ... who gets sued? The manufacturer? Is Tesla going to tank every single lawsuit from its Full Self-Driving system injuring/killing someone?

I don’t think this is a side note. In the article they appear to have issue with Tesla’s marketing of their feature and the reality.

On Monday, California's state regulator said: "Based on information Tesla has provided the DMV, the feature does not make the vehicle an autonomous vehicle per California regulations."

So the safety concern seems justified. Tesla said it’s full self driving. In reality it’s just assisted driving. Maybe they aren’t prepared to deal with the risk???

But Full self Driving isn’t the same as saying fully autonomous. The beta makes it clear you are still the primary driver of the vehicle.

Really we are deciding car company marketing terms are suddenly dubious when someone is legitimately on the cusp of a breakthrough.

VW Blue Motion… there is no blue in that motion.

Audi Quattro Engine… is it four cylinders? 4 litres?

Ford Mach-E — Mach has a defined meaning… seems deceptive to me.

Those other marketing terms are not trying to mislead people into behaving dangerously in an environment where people can get killed. Perhaps I'm a minority, but I absolutely don't want companies to use marketing weasel words when it comes to self-driving capabilities.

Let's not beat about the bush, it's clear that some drivers are already abusing the current "self-driving but not really-self driving" autopilot features. This steps up the bullshit saying "oh this time it is full self driving (but not the total absolute hands off autonomous driving". So which is it? If it's not fully autonomous, don't bullshit saying it's full self driving.

In other words, if there's no blue in Blue Motion, no count of 4 in a quattro and no mach in Mach-E, that's not likely to get people killed. But if there's no full self driving in something called "Full Self Driving", that can and will get people killed.

How will it kill people exactly? You mean the people that (despite the countless warnings and explanations) still think that they can completely ignore their surroundings… then what? The very cautious auto driving car that follows the speed limit and is criticised for being “too conservative” will… do what exactly?

You seem to think Tesla’s system is perfectly safe. I’ve seen nothing that indicates that—just the opposite, in fact.

Could you reference an FSD safety incident?

I know there have been some autopilot accidents but that’s not this system.

Do a web search for “Tesla fsd beta scary” for video examples.

The only examples I could find were of people uncomfortable with the automation, but not scared of an accident. Also no actual accident. Do you have a direct link?

My contention is only indirectly with the safety of the system, and mostly with the labeling and messaging. A cruise control system that is unsafe to leave unattended but isn't called "Fully Self-Driving" or even "Autopilot" is fine. There's no ambiguity. Everyone knows that cruise control is a dumb system that will happily crash you into the wall.

However, I have issues with calling and marketing something "Fully Self-Driving" and then adding paragraphs of text to explain that it's actually not fully autonomous, in fact it's not supposed to be autonomous at all and the driver has to have their full attention on the road and hands on the wheel and any other operation is dangerous. If this was something benign, I'd just consider it disingenuous and shady marketing (that seems par for the course for corporations), but in case of a 2 ton vehicle it's not just their own customers at risk, but everyone around them.

Yes, you and me know what this system really is and can approach it responsibly. But we see so much evidence that there are many people who have no scruples about ignoring this (willfully or not) and assuming that the system must be good enough if it's called "Fully Self-Driving" to unpack a lunch on their lap.

there is no “quattro engine”

It's marketing term for AWD. In which case it really is perfectly good name standing for number four.

“quattro” is that term. there is still no “quattro engine”

Why don’t they call or 4WD - that is the accepted term for a car that drives on all 4 wheels?

Why do we name things? Why do we have names for anything?

Agree. So FSD for something that drives a car for you. Is as valid as Quattro for a car that drives on all four wheels.

Full self driving means something. FULL SELF and driving something that fully self drives. Quattro, is just word for four... And there is 4 things that do things.

Other one is currently absolutely total lie which should result termination of company, other is not in anyway misleading.

Quattro was a rally car model. They used the name to describe their 4WD drivetrain. They constantly advertise it using the shots of that rally car racing in that (now illegal) rally race.

But sure, FSD is the irresponsible one.

Alright, let me just go to market with my “Cure For Cancer”. It doesn’t actually cure anything, but what’s that got to do with anything?

If it cures most instances of cancer, nobody will complain that the moniker isn’t totally correct in all cases.

If FSD can handle >90% of my driving, I’ll not quibble.

By definition, the driver is in control of the vehicle 100% of the time when using a level 2 driver assistance system. These systems assist the driver with some of the steering wheel, accelerator, and brake functions, but ultimately, the human driver is always driving.

You won't get to market. The Susan G Komen Foundation will sue you into oblivion, first. Apropos of anything else, they have trademarks on anything resembling "for the cure" when it comes to cancer.

Source: https://en.wikipedia.org/wiki/Susan_G._Komen_for_the_Cure#Co...

Tesla has nothing to do with the person? Not founded by anyone named such, clearly dubious marketing too?

Sure. My point exactly was… no one seems to care about the countless meaningless terms automotive industry uses. But suddenly we are dissecting the possible Meanings of a feature that actually is pretty close to what it says.

Autopilot keeps your heading and velocity and is aware of objects in its path… so that’s extremely similar to autopilot in a plane.

Full self driving drives the car on its own. It still monitors driver awareness. But the car does most of the driving. I don’t see the issue. It does what it says.

> It does what it says.

> Full self driving drives the car on its own. It still monitors driver awareness. But the car does most of the driving.

Full directly contradicts most

Again, it’s closer to what it does than Blue Motion. Or Toyotas latest “Safety Sense” (even though it sense hazards aka forward collision detection…)

I can reel out examples all day, that no one bats an eye at. But Tesla so something… it’s a whole different ball game.

Uh neither of those even suggest what it does while Tesla's says it does something it doesn't do

So esoteric meaningless names are good.

Descriptive names are bad, because: I don’t think full self driving is full if there are any interventions.

(Despite there being a sliding scale of autonomous driving from levels 1 to 5 with FSD easily falling into 3 if not 4).

Teslas official literature classes it at level 2. But yep, super misleading compared to “Safe Sense” (aka forward hazard detection). Do you know what Tesla call that feature: “forward hazard detection” those bastards with their descriptive names!

You're either being facetious or ignorant. The marketing name of a model, especially nonsensical terms such as "blue motion", are clearly recognized by anyone as being pure marketing terms chosen for aesthetic reasons.

By contrast, car features generally have non-misleading names - "anti-lock breaking system", "cruise [speed] control" etc. "Full self driving" goes beyond the pail any way you look at it. It is a very explicit name: it means very specifically that a car can drive itself, with no human intervention whatsoever. Even more, Elon Musk, the CEO of the company, has repeatedly claimed to buyers that FSD will allow the cars to work as robo-taxis that can be allowed to work for their owners while their owners are otherwise occupied. This again very explicitly says that FSD doesn't require any human intervention whatsoever.

The fact that, after realizing that it obviously is nowhere near close to such a thing*, they started putting in warning that Full Self Driving is in fact not full self driving in any way shouldn't excuse them. They have obviously repeatedly lied and misled buying customers, and the name has to be changed.

*(and won't be in the next 50 years, most likely, until we actually start seeing AGI or something)

You think there are consumers out there who are being tricked by Tesla? Like they buy FSD and are like “what!?” This isn’t what I expected!? You lied to me!

Also FSD is offered as a subscription so you can “try it out” if you are cautious. Do people really see this as an issue?

There are MUCH worse things on the road than FSD driven Teslas - what consequence do people think will occur?

How long are you going to hold to your disingenuous argument?

"Full self driving" is an explicit term - "self driving" means the car drives itself. "Full" means it does it fully. It couldn't be any clearer what the term means.

Terms like "blue motion", "quattro", "mach-e" are just made up branding terms that would mean nothing to anyone without knowing about them before hand.

I’m not being disingenuous. I really don’t think what features are named on cars are important. The disingenuous position I see is that everyone agrees with me… except if Tesla do it.

FSD doesnt mean anything to anyone either, as it’s never been done before.

> except if Tesla do it.

Its because only Telsa does it. Name an example of another manufacturer using a term as explicit and misleading as "full self driving".

You don't see other manufacturers calling ABS "fully tailgating safe".

Mercedes call their version: Intelligent Drive System

BMW: BMW Personal CoPilot systems

Audi: Traffic Jam Pilot (which they class as Level 3 autonomous driving, higher than Tesla class both autopilot and FSD).

This isn’t an opinion… this is fact. People hate on Tesla out of ignorance of the automotive industry.

Some more facts, Tesla make: - the safest cars (no qualification required) - most efficient production cars - most advanced “auto pilot” but they class it the lowest level.

The nonsense in HN threads talking about safety issues in Tesla or FSD need to go look up cruise control crashes in any other manufacture.

(I don’t own a Tesla nor am I invested. I do work with almost all the major OEMs in my day job)

So explain to me why Tesla themselves put out a video that featured the text:

> The driver is only in the seat for legal reasons. The car is driving itself.

and talk to me about "it's only marketing, you should understand".

Oh, and as the owner of three Audis over my life, there is NO material from Audi that calls it a "Quattro Engine".

It's the "Quattro all wheel drive system".

So it’s Audi using a confusing name for 4WD (the accepted term).

What’s the accepted term for level 5 autonomy?

What’s the accepted term for level 2?

Do you think there are tesla consumers being tricked by FSD capabilities? Right now it highlights that it auto drives on highways but not on streets… which is pretty accurate in my opinion. Where is the dishonesty?

You’re really grasping at straws with the Audi thing. First it was something that has never been said, and now a word for “four” is confusing for naming a four wheel drive system.

Full Self Driving would be acceptable and accurate for level 5.

Now for level 2 advanced driver assistance.

“The driver is only in the seat for legal reasons. The car is driving itself” is a current Tesla statement. That’s inaccurate.

“The car is driving itself until it can’t, or it does something it shouldn’t. But you’re in charge, and we legally disclaim any responsibility for failures of the car to drive itself” doesn’t have quite the same ring to it, I admit.

Not clutching at straws. Using a direct comparison to a competing car company on marketing terms. If we are okay with companies calling features whatever they want. Then FSD is fine. If we aren’t, then let’s get the outrage train going on the entire car industry.

Your original phrasing was "misleading" and "confusing". "Quattro" may be vague, at worst, but I challenge you to define how it is either of those things.

Misleading: it originally referred to a rally car model made by Audi but is used to describe 4WD. It is usually advertised with shots of said rally car driving in a now illegal rally race.

Not misleading or confusing at all. I’m glad people focused on Audi’s Quattro out of my examples because it’s honestly the worst one.

Too many are caught up in word-thinking, flustered about “full” when nobody buying it thinks it a 100% safe tech. People want to bitch about it, and do so with “but but but it says ‘full’ when it isn’t!”

Why do you think Tesla chooses to call it this way versus "assisted driving" or something similar?

How do you have autonomous robotaxies if the "full" in "full self driving" isn't "full" in the sense that every native english speaker assumes?

So ”full self driving” in the car world is just marketing for “assisted driving”?

This is why we’re never going to have self driving cars. If everything that they say in the industry has to be taken with a grain of salt, why would any city agree to take on that liability?

Doesn’t even matter if the tech saves lives. You can’t even be straight about what you can deliver.

When the system is registered as a level 2 system, even if it's fully autonomous, human driver is responsible.

I'm pretty much 100% certain it'll end up being the driver who engaged the self-driving mode in a situation where it wasn't safe.

Think of it like firing a gun into the air. You didn't aim to kill anyone, but if someone is unlucky enough to be hit when the bullet comes back down you are still responsible.

If someone is unlucky enough to be struck by your FSD vehicle it'll still be your fault, regardless of whether you could have known it'd happen.

Drivers will need to weigh trusting their car's software against the risk of being responsibly for what it does.

> But if there isn’t, these experiments should progress as they could ultimately reduce mortality.

Another aspect that is important but easy to miss - deaths caused by an autonomous vehicle are learning experiences - once a couple have happened there will be an engineering response and they will stop happening.

Compare that to ordinary drivers, where each death is just a pointless waste of a life. There is an argument in favour of putting these cars on the road even if they were more dangerous than humans. The faster we push through the painful learning experience the sooner the payoffs of superhuman driving will start being felt in the death statistics.

That’s fine and all, but who’s going to volunteer to die for no fault of their own, just to progress science? Heck, what if it isn’t even someone who signed up for the FSD beta but just a random bystander the car decided to run over.

You’re missing the point: Random bystanders are run over by human drivers all the time. But when a human driver makes a mistake, we can’t analyze what happened and push out an update to everyone else.

That’s fine, but you can’t sign people up for medical trials just because people die from diseases anyway. You don’t get to force other people into becoming Guinea pigs for science.

Unfortunately or not, this is exactly what Deaths per X miles driven statistics actually are. And that’s the statistic set at the base of automotive policy making.

Here: https://www.iihs.org/topics/fatality-statistics/detail/state...

Unfortunately or not, this is exactly what Deaths per X miles driven statistics actually are. And that’s the statistic set at the base of automotive safety policy decision making.

https://www.iihs.org/topics/fatality-statistics/detail/state...

> once a couple have happened there will be an engineering response and they will stop happening

That's the mother of all assumptions. They might get worse as a result of the engineering response. You are assuming that there is a perfect understanding of the problem space. There isn't. This is not a 'mere matter of engineering' at this stage.

It isn't that big an assumption given how the engineering responses to incidents include incorporating those incidents into their simulated testing processes.

The point is that the understanding of the problem space increases with each incident. This doesn't mean that regressions are impossible (they are probably inevitable), but each regression also increases understanding of the problem space.

> Another aspect that is important but easy to miss - deaths caused by an autonomous vehicle are learning experiences - once a couple have happened there will be an engineering response and they will stop happening.

A death isn't the same as running into a stationary emergency vehicle of course, but AFAIK Tesla hasn't taken any meaningful engineering response to that issue. Assuming they will to other issues seems overly generous.

OTOH, Waymo seems to have only crashed into one bus, assuming it would move over.

> I’m confused by sf. They did this for Uber in 2016, Amazon’s zoox, and a host of other companies, etc.

Each of those companies, and Waymo, etc., all agreed to accept liability for any damages from their efforts.

Tesla is sticking with "the driver is responsible, not us". So yeah, that's why there's a discrepancy.

Why is it every time self-driving comes up, proponents act like there's nothing between forging ahead towards L5 and improving over an unaided driver?

Currently AP's "self-driving" features cannot save a life, they have killed people.

To understand that better, examine AP in finer detail:

Teslas have safety features that can intervene when the current situation seems to be leading to an accident. Let's call those correction. Those can (and have) save lives.

Tesla also has features that are supposed to be more convenient for the driver. The claim is they take a load off for the driver. Call those the convenience features.

The convenience features have not evolved to the point where you are ever allowed to be less vigilant than you would be otherwise! By definition if AP is driving, you too must be fully attentive and "driving" hence the increasingly intrusive detection which has now evolved to detecting wheel input every 15 seconds...

Now realize that while for the sake of examination I'm describing convenience vs correction, these are all systems in one car. You can get the correction features without FSD, but you can't get the inverse.

Because of this, the convenience features alone cannot avoid an accident that the avoidance systems would not have.

This is tricky to think about, but consider a driver who would have drifted off the road without AP. As soon as the driver deviates too far from what AP would have done and starts touching a lane marker, the lane detection systems realize there's an issue and correct.

However, say there's a tricky lane marking where the correction system wrongly believes the driver input is wrong. They get an alert and an attempt at correcting steering inputs.

But because the driver is already inputting a direction with full intention, this wrong input is already actively being overridden by the human who plotted their turn and can clearly sense that they should fight the suggested input before they end up running off the road.

With the convenience features enabled, the driver's input will come after the wrong action has already been taken!

If the avoidance systems knew that the convenience feature was about to send you into a jersey barrier... the convenience feature wouldn't be "allowed" to send you into a jersey barrier!

I want to see studies that focus on the safety of driving with full driving aids but without convenience features. Claims that you are less tired from using them show that it's an inherently unsafe practice since you should be exactly as vigilant as normal.

Because of what's explained above, the expectation would be that there is no drop in safety if your car only corrects when there is a hazard... but there is a drop in safety when you are falsely lulled into a sense of safety.

To me WIP self-driving should be monitored by trained people. It shouldn't be a box you get to tick in your infotainment screen.

If Tesla wants to carefully pick who's allowed to run these betas more power to them, but the number of idiotic videos I can pull up right now of people intentionally testing FSD updates in situations they know are hairy and risking other people's safety is insane.

That is just setting up self-driving to be dead on arrival.

> Currently AP's "self-driving" features cannot save a life, they have killed people.

That is just not true. Distracted/impaired human driving is a real thing that kills people. Just two weeks ago there was a DUI event where a Tesla driver on AP literally passed out behind the wheel and the car was trivially brought to a stop anyway.

> This is tricky to think about, but consider

This is postulating a two-mode failure! That's just bad engineering. Yes, that can happen, but by definition (involving two rare events overlapping) it's an extremely rare condition and will show only a vanishing frequency in any final statistics.

Never compromise core features (in this case "don't let distracted drivers hit the thing in front of them") chasing this kind of thing. Work on the secondary/tertiary modes once you have everything else down.

> Just two weeks ago there was a DUI event where a Tesla driver on AP literally passed out behind the wheel and the car was trivially brought to a stop anyway.

Nope, and this is a perfect example!

AP didn't save that person's life, or more specifically, AP wasn't not needed to save this person's life

Tesla made Emergency Lane Departure Avoidance standard on all Teslas a while back.

Emergency Lane Departure Avoidance only intervenes if the driver deviates from expected behavior

So situations where AP would avoid a collision while being in control, the car still has a feature using the same sensor suite and the same capabilities as AP to intervene.

A Tesla without AP would have stayed in it's lane in the absence of driver input and slowed to a halt with the flashers on.

Honestly a better outcome...

It's as simple as, AP is half heartedly handling the happy path + the unhappy path.

It's ok to not have perfect handling of the unhappy path as it does currently, something is better than nothing in 99% of cases. But it's not ok to have this mode that falsely gives drivers the impression they're covered on the happy path.

Like your example could not be more perfect because a person driving under the influence relied on AP to handle the happy path for them. Now you can get into "well maybe they would have driven anyways if it didn't have AP"

And that's where emergency LKA and AEB come in. They wouldn't have given this driver a crutch to get on the road, but if they still chose to drive, they'd avoid an accident in the same manner AP did once things went off the rails.

This all goes back to the fact that the convenience features by definition cannot save a life. AP, as in the feature manually activated to assist the driver actively, cannot save a life, because it only works on the happy path. Other features like LKA and AEB are what save lives

> This is postulating a two-mode failure! That's just bad engineering. Yes, that can happen, but by definition (involving two rare events overlapping) it's an extremely rare condition and will show only a vanishing frequency in any final statistics.

> Never compromise core features (in this case "don't let distracted drivers hit the thing in front of them") chasing this kind of thing. Work on the secondary/tertiary modes once you have everything else down.

I have no idea at all where any of this is coming from... please quote where I suggested core features should be compromised for secondary features.

> A Tesla without AP would have stayed in it's lane in the absence of driver input and slowed to a halt with the flashers on.

That's not how those systems work. You're confused, I think. If you let Jesus take the wheel in a Tesla under manual control it will absolutely depart your lane and hit whatever you are aimed at. The emergency features are last minute corrections to avoid an imminent collision, and they work about as well as other car vendor's products do, they reduce the likelihood and velocity of an immiment collision. They don't drive the car for you. The feature that does is AP.

> please quote where I suggested core features should be compromised for secondary features.

You want to refuse autonomy solutions to everyone to save hypothetical bikers at sunset in Mendocino! (Edit: sorry, I got the wrong elaborate example and was paraphrasing another poster. You want to save drivers where they incorrectly countermand an incorrect autonomy decision in a circumstance where the incorrect autonomy would have been correct. Or something. But the logic is the same.)

> If you let Jesus take the wheel in a Tesla under manual control it will absolutely depart your lane and hit whatever you are aimed at.

You are wrong.

https://www.teslarati.com/tesla-lane-departure-avoidance-sav...

I understand your confusion though, Tesla is well behind other manufacturers here.

VWs have had this behavior for years now, almost as soon as they started dabbling in advanced safety assists, meanwhile Tesla got it just two years ago.

https://youtu.be/TITEf_taUto

Kind of shows where the priorities were early on, saving incapacitated drivers vs "look ma no hands!"

Also your second quote just leaves me with more questions.

What on earth are you talking about

> You want to save drivers where they incorrectly countermand an incorrect autonomy decision in a circumstance where the incorrect autonomy would have been correct.

What?

I don't know what to tell you. I have tested this feature on my own car, and its behavior matches the public documentation (c.f. https://www.tesla.com/blog/more-advanced-safety-tesla-owners). It will scream at you as it departs the lane, but it will depart the lane. It will correct only if it detects an object or barrier it needs to avoid. It does not do what you think it does. Tesla does offer a product to do that. It's called "Autopilot" and it absolutely works great. For example it prevented a collision (or even lane departure) entirely in that DUI case we're discussing.

Your test sounds like it failed then... did you read your own link?

> Emergency Lane Departure Avoidance is **designed to steer a Tesla vehicle back into the driving lane** if our system detects that it is departing its lane and **there could be a collision, or if the car is close to the edge of the road**.

And again, AP acted as a crutch that an intoxicated driver attempted to lean on.

Now an intoxicated driver could have decided to drive without the crutch... but emergency lane departure avoidance would have kicked in when they passed out and centered the vehicle to a stop instead of trying to maintain a certain speed.

(And let's head-off the straw-man here, no I'm not blaming AP for an intoxicated driver but I am stating the reality, that AP. Autopilot. Is marketed in a way that causes sober people to think of it as a co-pilot, let alone drunk people)

AP literally disengages when off the happy path, it's other safety features that help when things go wrong.

It's amazing that people still haven't caught onto this.

You just keep digging here, and it's all wrong. I don't know how else to reply but with facts:

> AP literally disengages when off the happy path, it's other safety features that help when things go wrong.

AP never disengages autonomously, ever. You turn it off via driver action, either via steering input that countermands it or via the stalk control. (Same deal for TACC -- it's on until you hit the brakes to kill it).

Really, go rent a Tesla for a few days and experiment. I think you'd find it enlightening.

Wrong again...

https://jalopnik.com/this-video-of-a-terrifying-high-speed-t...

https://www.techtimes.com/articles/254126/20201112/watch-tes...

> Autopilot disengaged about 40 seconds prior to impact due to the Tesla issuing a Forward Collision Warning (FCW) chime

Here's a reckless driver in a Tesla assuming AP could handle unsafe speeds on their behalf.

Once again AP is not the one to blame but... a driver tries to use it as a crutch and it disengages due to the FCW.

Due to the complete lack of input from the driver, many suspect the driver didn't realize AP would be disengaged by the FCW, and while AP disengaging would have given an alert, it'd be at the same time as the multiple FCW alerts they got.

So just like the DUI, the safety features would have worked *better* without AP. If they had been driving without AP they would have not have expected a now-disabled feature to respond to the FCW for them...

Seems like you didn't realize this was the case either with FCW and AP, so you're welcome....

Ps: My Model S test drive (remember those) was not going so hot when the headliner started peeling off the demo car, but then went cold when the not-a-salesperson panicked when I didn't disable AP fast enough (since we were approaching a section of lane markers in their test circuit where it was known to veer towards barriers)

I'm pretty familiar with Teslas though, I was a huge fan until gestures to everything from AP marketing to silent CPO->used change, etc.

Pps: Please don't go and try this on a public road to prove an internet stranger wrong.

You already admitted to trying to test ELKA (which won't intervene until things are going seriously wrong.)

You're not going to be able to test a FCW with AP without doing something silly...

It’s a setting. You can have a just a warning or a corrective maneuver applied.

Emergency LKA (which maneuvers) turns on every time the car is started regardless of the setting and has to be re-disabled unless something changed since the press release

>Tesla has a safety record that’s not unimpressive compared to the 36k...

But when Tesla cars were then compared to similarly priced cars; that record dropped through the floor.

What were the numbers, out of curiosity? Telsa is worse per mile for safety than others similarly priced cars?

I don't remember except that i found it significant, i tried to find the article i think it was a university that did the study, if i find it i will post it.

I’m not sure why we would care about controlling for price of the car…

There are many reasons why one could be interested in that.

On the personal decision making level you might want to know what car provides the best safety record for your budget. If you can afford a tesla you would be interested in the safety of similarly priced cars to compare with.

On the society wide statistics level Tesla drivers are not exactly like the whole driving population. If for no other reason the price selects a specific demographics with their associated behaviour patterns. By controlling for the price we can hope to better control for these differences too.

You’d also want to control for age of car. Recent expensive cars are more likely to have advanced safety features that older or cheaper cars lack. Maintenance can also be a factor.

Price indicates class of the vehicle (premium in Tesla’s case), so they can be compared with cars in the same class that have similar safety features.

Product safety is more nuanced than simple statistics in aggregates. Assuming that Covid-19 has mortality rate of 2%, imagine if there was a medication that cures Covid-19 with 100% efficacy but has a side effect of causing immediate death for 1% of the patient who take this medication. One could argue that we'd be better off in aggregate but I hope you can see why this hypothetical medication wouldn't be approved.

Why wouldn't it be approved? vaccine 'side effects' might not sound serious but they are absolutely deadly to some if we were to give them to millions of people. For the elderly, a 1% risk of death might be far better than the risk of covid since mortality shoots up once you get older. Obviously you shouldn't give this vaccine for anyone whose risk of death by covid is less than 1%

I'm not referring to vaccine. Assuming you've been infected, if you have 98% change of surviving this disease, would you really be okay if you were forced to take a medication that has 1% chance of killing you?

It wouldn't be framed that way. It would read: "Chance of recovery improves from 98 to 99 percent."

Chemotherapy regularily kills people. I would absolutely consider it. Taking a lot of pain to get a chance at extending my life.

Chemo is a good example. It's used in a situation of last resort where probability of dying is already very high and if I were in that situation, I would too.

If my options are to be forced to have the medication or prevented from having it, I’d rather be forced to have it. It reduces my risk by half so why not?

YouTuber Mahmood Hikmet analysed Tesla safety numbers in his recent video https://youtu.be/DxXiTf4sO54?t=1190 and it doesn't look good for Tesla.

"Tesla autopilot saves lives" is a bold claim.

Reply on news.ycombinator.com