I don’t understand the downvotes. This was actually the same reason for me that changed my perception of the company and of musk, from something fantastic to something dangerous.

When they went from saying « we’re aiming at self driving car in the future » to « our car are able to do it right now », leaving all kinds of ambiguity just for the sake of hyping the brand, they crossed the red line.

Humans driving cars cause preventable deaths.

I think it's like saying "I don't like surgery, because mistakes during surgeries cause preventable deaths"

It’s not a monetised feature, though.

Driving a car is indeed a monetized feature of cars.

He's saying even if a self driving car is more dangerous than a human, a human driven car is still dangerous and we all accept that. So its not about whether the deaths were preventable. We accept human driven cars therefore we accept preventable death.

We're now forced to consider the fact that this is an argument about the fuzzy math of risk vs reward as we have crossed the threshold of preventable deaths long ago.

What is the point of this comment ? Humans doing anything cause preventable deaths.

That has nothing to do with whether or not a new, potentially life threatening technology should be on the market.

I guess OP means preventable by others who are knowledgeable of the preventability?

If we arrive at a working solution faster (i.e. something that prevents more deaths than it causes), it makes sense to speed up its development to get to that state faster.

The question remains if we accept a death by software bug (i.e. human software developer error) in the same way we accept a death by human driving error.

> The question remains if we accept a death by software bug (i.e. human software developer error) in the same way we accept a death by human driving error.

The closest thing Hacker News has to consensus is "certainly not if it's Boeing, probably not if it's Uber but definitely if it's Tesla..."

Musk’s hubris is annoying, but how many lives did he save by not waiting until the autopilot was “ready”? It is probably in the dozens at this point.

I'm not sure why you think that. While this is a very dense document, it details a lot about the fatality statistics around Teslas and Autopilot. The short version is that Teslas are actually not very safe cars (relatively speaking), and Autopilot is probably much more dangerous than manual driving (although Tesla won't release the data, so we don't know for sure).

https://docs.google.com/spreadsheets/d/1ESnyJ4b7m96OCjs3GSQ6...

That's a list of all fatalities involving Teslas reported in the media, including ones caused by human Tesla drivers or the drivers of other vehicles involved in the collision, and speculation trying to associate that with a number of miles driven.

Then they have "miles per death" numbers in the 150M-240M range, but the US average is less than 100M per death (typically stated as the inverse, deaths per 100M miles[1]), so according to your data they're apparently better than average, though collecting it through media reports this way is questionable anyway.

[1] https://en.wikipedia.org/wiki/Motor_vehicle_fatality_rate_in...

I see a column for deaths attributed to autopilot, but it sums to 4 in the US, which is too small to be a good statistical sample even if we knew how many miles per death that was, but I don't see any data for miles driven on autopilot regardless.

The thing about comparisons is you need to compare like to like as much as possible. Demographics are an essential part of actuarial statistics. I think the best comparison is between Tesla and luxury vehicles, and here is an article showing Teslas are much more dangerous:

https://medium.com/@MidwesternHedgi/teslas-driver-fatality-r...

I'm not sure if you noticed (I didn't the first time I saw that document), but there are multiple tabs and that's under the 'Resources' section.

That said, the data is obviously incomplete. Only Tesla really knows. While it's obviously a guess, I think the most damning evidence is that Tesla isn't releasing their own data to the contrary, because I think Elon Musk of all people would be ecstatic to shutup the haters such as myself.

The demographic point is true, but it really invalidates the whole thing, because you have to be assuming that the same people who buy a Tesla would have otherwise bought a Mercedes S class and not e.g. a Corvette or Porsche. If the Telsa owners skew younger or have other demographic differences from ordinary luxury car purchasers then the comparison is meaningless without factoring that in because you're measuring the driver behavior as much as the cars. (And choosing "luxury cars" rather than "sports cars" or just "cars" as the comparison is maximally unfavorable, since luxury vehicles are generally the safest available regardless of driver.)

The comparison between Tesla and other luxury cars isn't that far of. It is even the one Tesla is using, or was using, to show success compared to high priced Mercedes, Audi, Porsche and BMW models.

One thing I find annoying is that Tesla and Musk seem to always move the goal post to point where it suits them best. For me stuff like self-driving taxis and sattelite internet are prime examples. It looks, at least to me, like running away from rwality faster than it can catch up.

One last point regarding luxury cars and safety. Tesla is proud of being the top performer at the NCAP crash test. And all of a sudden that comparison is unfair just because the numbers aren't positive for Tesla?

That's a valid criticism, but I still think it's the best analysis with the available data.

FWIW, and this is completely anecdotal, I find Tesla drivers to be the most obnoxious on the road - which would lend credence to your argument that it attracts a certain kind of person and is not the car itself. That's ignoring the entire Autopilot X factor though, which is a whole separate conversation with even more missing data.

‾\_(ツ)_/‾

the author is a tesla short and proud of it . Now the matter of data he sites its his blog which has no published paper or peer review. He quotes IIHS stats but says that they dont have enough data on Tesla ?? and he extrapolates tesla accident rates with correlation with news articles ?? and claims all death rates are not reported cause there may be data entry issues not realizing that it goes both ways?? . Being Short on tesla is different than being short on reality , the apples to apples comparison should be made with driver assist features . If its a matter of structural safety i believe that there should be no doubt that tesla make the safest cars by any standard euro NCAP / us NTHS .

Valid criticism of the source. And exactly the reason why I personally find Tesla, and SpaceX, so annoying. You just can't seem to get reliable unbiased data beyond Tesla's SEC fillings. It is either pro-Musk or anti-Musk to a degree where all neutral analysis are just drowned out. And that sucks, especially since Musk and Tesla ans SpaceX are intentionally fueling that cycle. Which in itself is somewhat worrying.

Do you dislike drunk-drivers and phone-drivers with the same passion and focus? Do you expect them to gradually become better and better at driving through natural evolution?

No car company implies that drunk driver would be safer than myself behind the wheel, and I should let her drive instead.

I hate drunk drivers way more than I hate Tesla. That stuff never shows up on HN though, whereas Tesla is on the front page all the time.

Autopilot (and by extension Tesla) is also a lot more controversial than drunk driving, and by extension more interesting to discuss.