OFFICIAL BUTTON WATCH

[kaffine]

You're aware Tesla just recently pushed back hard on a recall from NHTSA, almost making NTSA make it an involuntary recall? So maybe Tesla isn't so magic or special.

While I don't fully agree all your examples are the same as a company purposefully, knowingly killing uninvolved, uniformed people to accelerate their R&D, I will point out that you believe it to be unacceptable by the tone you use describing other companies doing it. So we're all in agreement, it is not acceptable for Tesla to release software they know will kill people at a higher rate than the baseline, just because it will make them money, right?

Companies continuing to pollute is a very similar situation I will agree. It's a good example. It's not OK, just like it wouldn't be OK for Tesla to do it or something similar just because Elon has a long term vision.

Accidents and design mistakes are very different than knowingly operating an unsafe system in order to collect data. The tobacco argument is a very interesting one in this context, given the number of supporters here claiming that Tesla drivers know what they are getting in to and have the right to make that decision for themselves.

No I wasn't aware of the recall. I'm also not claiming Tesla is better than other companies in that regard just that they really aren't any worse than other companies either.

Is it ethical to delay R&D until you can reduce risk if it means 400k more people will die due to the delay? To me this is much less about ethics and much more about public perception. What the public is willing to accept does not mean it is the ethical choice. Does it make a difference if the deaths are easy to point to and say that is what caused it vs the issue coming to light years later? It is easy to point to autonomous driving failures that cause a death it is much harder to show the lives that autonomous driving saves from avoid an accident or reducing the serverity.

Is it ethical for a medical company to offer something that will save a persons life if they know it will fail 4 years later killing them? The public seems to be more accepting of a person dying of natural causes the first time than of dying of a medical drug or device failing years later. When the medical device fails it will suddenly be the fault of the company that made it even though if it wasn't for that the person would have died 4 years earlier if it wasn't for that device.

 

[Badinfluenz]

Haha.Read between the lines

 

[mikes_fsd]

Saturday meaning 2 weeks?

I'll take the over

I'm telling you guys. Now its set up for August, and the next tweet will be that DOJO is so close to coming online that there isn't any value in pushing it out to more people before then.

I’ll believe it when see it. I don’t believe it will drop this week.

Well, do you believe it now?
He delivered, on time...
Now, lets see the pivot to complaining about something else ... oh wait, the pivot already happened 6 hrs ago with the first set of videos released.

 

[Ptheven]

Well, do you believe it now?
He delivered, on time...
Now, lets see the pivot to complaining about something else ... oh wait, the pivot already happened 6 hrs ago with the first set of videos released.

Don't see why you included microterf's post in there since it's clearly about how they will keep having excuses to only release to Elon's core group of leg humpers, not about the release last night.

 

[mikes_fsd]

Don't see why you included microterf's post in there since it's clearly about how they will keep having excuses to only release to Elon's core group of leg jumpers, not about the release last night.

Easy, low hanging target for you to pivot on.

 

[Ptheven]

Easy, low hanging target for you to pivot on.

Well, irrespective, Elon getting a software release date not-late (it wasn't right, technically) isn't impossible, it's just an irregularity. When someone pisses on their own balls 99 out of every 100 times they use a urinal, the one-off doesn't mean they don't deserve the reputation of being a ball-pisser.

 

[mikes_fsd]

When someone pisses on their own balls 99 out of every 100 times they use a urinal,

Dude, you might wanna have that checked out!

 

[Knightshade]

Sorry, can't use these numbers found at carinsurance.org. Which auto makes get in the most fatal crashes? (2021 Results)

If you don't have better data, of course I can use that data.

They don't control for car age and then also safety equipment like DSC, ABS, pyrotech seatbelts and airbags. Tesla are so new, 2014, and all cars have that equipment as standard. A 1985 Dodge or Ford doesn't, probably not a 1995 or 2005 either.

While I agree they don't control for age, the % of any given brand that is 20 year old cars is pretty tiny relative to late model ones..

If Tesla were BARELY better than Dodge that'd matter, but they're massively better.

Hell, nearly EVERYONE is a lot better than dodge per 1k sales, including companies around at least as long or longer.

BTW they list Saab and Saturn, brands that is out of business.

Saab made cars as recently as 2014, Saturn 2010... and the data in the study goes back 2017. So again you don't need 20 year old, or even 10 year old, cars from either name to be included here.

 

[Knightshade]

You're aware Tesla just recently pushed back hard on a recall from NHTSA, almost making NTSA make it an involuntary recall? So maybe Tesla isn't so magic or special.

... do you mean the recall because the large touch screens on old S/X vehicles could fail after years of use?

Because, especially given those models have a secondary, non-failing, dash display, that's... not exactly a THEY DON'T CARE IF IT KILLS YOU kind of recall, now is it?

Meanwhile you've got companies like GM and Ford who've been caught, repeatedly, LITERALLY AND ACTIVELY deciding to keep paying off wrongful death lawsuits- for years- because it's cheaper than fixing a $1 part across millions of cars.

 

[rxlawdude]

Don't see why you included microterf's post in there since it's clearly about how they will keep having excuses to only release to Elon's core group of leg humpers, not about the release last night.

Nostradamus? That you? Or Dionne Warwick?

 

[Ptheven]

Nostradamus? That you? Or Dionne Warwick?

I'm not the one making the prediction. I think microterf is right though, based on Elon's well-established history of complete BS on the subject.

 

[daktari]

If you don't have better data, of course I can use that data.






While I agree they don't control for age, the % of any given brand that is 20 year old cars is pretty tiny relative to late model ones..

If Tesla were BARELY better than Dodge that'd matter, but they're massively better.

Hell, nearly EVERYONE is a lot better than dodge per 1k sales, including companies around at least as long or longer.




Saab made cars as recently as 2014, Saturn 2010... and the data in the study goes back 2017. So again you don't need 20 year old, or even 10 year old, cars from either name to be included here.

The data is crashes last 3 years correct, but all cars, whatever age, are included! You will have a majority of 15 year old cars (average car age in US?) making some brands look worse.

What they should have done is compare only new cars sold the last 3 years. That would be interesting.

You just can't compare crash worthiness in a 20 year old Corolla or F-150 without airbags with a 6 year old Tesla.
Example of 1997 Rover 100 vs 2017 Honda Jazz:

Just have a look at EuroNCAP and see how similar new cars are. Almost all new cars, except a few cheap econo-boxes, have excellent and similar results. Tesla are just similar to the other best cars in their class, and no reason to believe a passenger in the 10 best will have different chance of survival.
The European New Car Assessment Programme | Euro NCAP

 

[nepenthe]

Yep, we're fully in agreement. Like I said yesterday:


I'm glad to see that @mikes_fsd has come around from downvoting it when I said it to loving it when someone else said it. I wonder what made him change his mind.


Testing in a car on public roads is more complex than that. The failure of your system may injure a completely uninvolved person in another car or a pedestrian. This is not the same as an individual deciding to eat unhealthy food or using a kitchen knife, which do not harm other uninvolved people. Tesla has a responsibility to all of society, not just it's own customers. It needs to in fact evaluate how well it's customers are able to interact with the system they created and the risk that places all of society at. This is also a highly complex system that no consumer is fully able to evaluate the risk of, due to incomplete information about the system. This is why we don't just allow aircraft manufacturers to make airplanes with no oversight and sell tickets to customers with a disclaimer saying "you're an adult, you evaluate the risk."

Driving a car and adjusting the radio, could injure a completely uninvolved person in another car, or a pedestrian. This is not me arguing that some people are not abusing the very clear responsibility as the driver/operator of Tesla's "beta" feature-set, but when done so as intended, it is hard to see the risk as anything different than the average driver spending a couple seconds adjusting the radio.

But yes, I am sadly very aware of how far we have moved away from individual responsibility as a society.

 

[Knightshade]

The data is crashes last 3 years correct, but all cars, whatever age, are included! You will have a majority of 15 year old cars (average car age in US?) making some brands look worse.

That's...now how averages work though.

Since the # of new cars sold stays (very roughly) similar year to year- the MAJORITY on the road wouldn't be 15 years old.

Also it's only 12.1 years, not 15.

Just have a look at EuroNCAP and see how similar new cars are. Almost all new cars, except a few cheap econo-boxes, have excellent and similar results. Tesla are just similar to the other best cars in their class, and no reason to believe a passenger in the 10 best will have different chance of survival.
The European New Car Assessment Programme | Euro NCAP

NHTSA data found the 3 models with the lowest change of passenger injury ever tested are all made by Tesla. (S, 3, X)

NHTSA threatened legal action when Tesla pointed this fact out, because officially NHTSA does not rate cars by that measure (they just do 1-5 stars)- but it's exactly what their own actual data shows.... (which is why Tesla never had to retract the claim- since it's factually accurate).

Not "Teslas are among the best" but 'If you made a list of the 3 best, they'd ALL be Teslas"

(honestly it's almost certainly 4 now with the Y, but I haven't seen the same analysis done for the Y)

 

[gearchruncher]

but when done so as intended, it is hard to see the risk as anything different than the average driver spending a couple seconds adjusting the radio.

FSD actively moves the controls of the car, and can suddenly change path, requiring immediate, decisive control. Look at the videos that drive the car right into a post.

No properly done Functional Hazard Analysis would rate an FSD error on the same level as changing the radio controls.

Controls moving in a non-expected way is basically what happened to the 737-Max and pilots did not react the way the designers intended. Are you saying that was a failure of personal responsibility and was no bigger deal than when they change the frequency on the radio? The FAA generally disallows any failure that can cause harm if not dealt with within 3 seconds, as human factors studies show this is unrealistic.

 

[powertoold]

Level 5 coming soon!

https://twitter.com/x/status/1414435086548733952

 

[Krash]

It will soon capture turn signals, hazards, ambulance/police lights & even hand gestures…​

So what hand hestures are we talking about?

 

[dipper]

FSD cannot do most humans can do with "human eyes" as Elon Musk said himself.

I look at the other drivers (could be either direct or through side mirror) and predict their intent. No FSD can do that. Looking at cars turn signals and brake lights can be a false sense of indicator when the other drivers may not use them.

 

[gearchruncher]

"Capture" is a convenient word that does not mean the system will do anything with it.
Technically, it already captures all of these photons.

 

[Cheburashka]

It will soon capture turn signals, hazards, ambulance/police lights & even hand gestures…​

So what hand hestures are we talking about?

FSD will hit the accelerator when someone gives you the bird.

I vaguely recollect he said it will recognize emergency vehicles a while ago, probably around a year or more, so this is not a new promise.