Blog WSJ: Ambitious Autopilot Push Angered Tesla Engineers

Tesla engineers are reportedly jumping ship because they don’t believe Autopilot 2.0 hardware can meet the fully self-driving promise that Elon Musk proudly touts.

The Wall Street Journal says Sterling Anderson, previously the Autopilot director, decided to leave Tesla in December in part because he didn’t agree with the claims Musk was making about the vehicle’s potential for full autonomy.

According to the WSJ (paywall):

In a meeting after the announcement, someone asked Autopilot director Sterling Anderson how Tesla could brand the product “Full Self-Driving,” several employees recall. “This was Elon’s decision,” they said he responded. Two months later, Mr. Anderson resigned.

The Autopilot division has lost some 10 employees and four managers recently, according to the report. Satish Jeyachandran, the former director of hardware engineering for Tesla’s Autopilot team, and Berta Rodriguez-Hervas, a former machine learning manager also left the company in June. Anderson was succeeded by Chris Lattner, a former Apple developer, but he left in June after just six months on the job.

Tesla has declined to comment on the report.

 

[whitex]

Sorry, maybe missing the subtlety. what is your point?

My point was that Tesla didn't disable the suspension to save lives.

 

[tomas]

My point was that Tesla didn't disable the suspension to save lives.

I would say they did it to further reduce the already small probability of another road hazard fire (which could cause injury or loss of life) until they could devise another measure. Which they did. And then They restored it. Safety has always been a very high priority, and there's a lot of pride in the safety record. Did I miss something?

 

[whitex]

I would say they did it to further reduce the already small probability of another road hazard fire (which could cause injury or loss of life) until they could devise another measure. Which they did. And then They restored it. Safety has always been a very high priority, and there's a lot of pride in the safety record. Did I miss something?

They disabled the SAS to reduce a chance already astronomically low (as per Elon's blog) that the car will be damaged. Yes, you can stretch that into possibly causing injury or loss of life, but that would only apply to anyone stubborn enough to insist on staying in the car while the car is slowly catching on fire (all 3 fires that did occur had a minimum 10 minutes warning). So no, the primary purpose of the titanium shield was not human safety, it was to prevent very rare damage to the vehicle which caused bad PR when the media blew it out of proportion.

 

[tomas]

They disabled the SAS to reduce a chance already astronomically low (as per Elon's blog) that the car will be damaged. Yes, you can stretch that into possibly causing injury or loss of life, but that would only apply to anyone stubborn enough to insist on staying in the car while the car is slowly catching on fire (all 3 fires that did occur had a minimum 10 minutes warning). So no, the primary purpose of the titanium shield was not human safety, it was to prevent very rare damage to the vehicle which caused bad PR when the media blew it out of proportion.

Not sure we are saying anything different then. Just putting the accent on a different syllable maybe. Regardless, sounds as if we agree they did it for a positive reason. Nothing nefarious about it.

 

[whitex]

Not sure we are saying anything different then. Just putting the accent on a different syllable maybe. Regardless, sounds as if we agree they did it for a positive reason. Nothing nefarious about it.

Well, not a nefarious end goal, but disregard for customers who paid for the feature - especially considering that they sneaked the SAS disablement into an update without warning anyone, without putting it into release notes. I personally called them about SAS not working and even their own support people didn't know they did that. They could have made the disablement optional, but they didn't.

Consider another extreme example: FSD is supposed to be safer than humans, right? Ok, therefore, would you argue that if Tesla disabled all AP2 cars from driving at all until FSD is ready and safer than humans, would that be something nefarious or for the customer's good (after all, they end up with a safer car in the end)? I bet there would be plenty of upset customers unable to start their cars until FSD is ready for prime time, no?

 

[Electroman]

Monday is hyperbole day.
Tuesday is exaggeration day
Wednesday is sky is falling day
Thursday is 'I am being cheated' day

 

[oktane]

Monday is hyperbole day.
Tuesday is exaggeration day
Wednesday is sky is falling day
Thursday is 'I am being cheated' day

Friday is Tesla failed us day

 

[tomas]

Well, not a nefarious end goal, but disregard for customers who paid for the feature - especially considering that they sneaked the SAS disablement into an update without warning anyone, without putting it into release notes. I personally called them about SAS not working and even their own support people didn't know they did that. They could have made the disablement optional, but they didn't.

Consider another extreme example: FSD is supposed to be safer than humans, right? Ok, therefore, would you argue that if Tesla disabled all AP2 cars from driving at all until FSD is ready and safer than humans, would that be something nefarious or for the customer's good (after all, they end up with a safer car in the end)? I bet there would be plenty of upset customers unable to start their cars until FSD is ready for prime time, no?

That's where I thought you were going. I did not see anything nefarious about temporary removal of feature to deal with safety issue and risk of loss of property. Sure, Tesla also had a reputational dog in the hunt, but that's fair- they did not want to fail due to public perception issues. That would have been worse for customers than temporary loss of a non crucial function.

 

[MSDadMich]

When you guys mention the self-driving feature, are you generally referring to level 3, 4 or 5 self driving?

 

[SlicedBr3ad]

When you guys mention the self-driving feature, are you generally referring to level 3, 4 or 5 self driving?

Fsd, when done and dusted will be lvl 4 for Tesla. Until Then, assume lvl 2-3.

To answer other threads. If someone sold me ap2 after test drive using ap1, I'd be upset, at myself. I agree that at least a discussion should be had with customer on how to resolve this. My only recommendation to those in this situation is hang in there a while longer and before the end of (2018/2019 tops) you'll see ap2 become eap.

 

[AWDtsla]

I miss the old days when people argued incessantly when Tesla took away the adjustable suspension option for a period of time. Oh, the discourse of Tesla protecting their customer's cars and lives. Then when Tesla gave everyone a free Titanium under shield on their car and gave back adjustable suspension height - crickets from the unhappy mass and forgotten like it never happened.

This too shall pass.

Protected 0 lives.

I still don't have "very low" setting in my car, which is in the firmware, just not available to us plebs. So it was never fully re-enabled.

 

[tomas]

I miss the old days when people argued incessantly when Tesla took away the adjustable suspension option for a period of time. Oh, the discourse of Tesla protecting their customer's cars and lives. Then when Tesla gave everyone a free Titanium under shield on their car and gave back adjustable suspension height - crickets from the unhappy mass and forgotten like it never happened.

This too shall pass.

Oh no, you said beetlejuice 3 times

 

[Barklikeadog]

That's a true statement but Tesla does not think it's impossible.

Tesla has proven that it has never been unethical over and over again!

People said you can't convince any one to buy an EV because there's no market.

It's as plain as daylight.

But Tesla sold Roadster as an EV.

People said it's impossible to bring the price down for the next version but Tesla did with Model S and X.

Panasonic personally thought that it's impossible to sell that many cells so why even thinking about Gigafactory but now it admits it's wrong and yes it can.

People even said Model 3 was a fraud, but here now it is.

Just because some engineers who don't believe it's possible, and yes there have been many many problems and obstacles, but Tesla has proven again and again that it is possible.

that is quite some political spin there. Don't look at the man behind the curtain...

 

[whitex]

That's where I thought you were going. I did not see anything nefarious about temporary removal of feature to deal with safety issue and risk of loss of property. Sure, Tesla also had a reputational dog in the hunt, but that's fair- they did not want to fail due to public perception issues. That would have been worse for customers than temporary loss of a non crucial function.

I think where you and I disconnect is on the following:

1. What is the threshold of risk that you are willing to disable non-crucial features for? In this case Elon stated that the chance of of battery puncture was astronomically low and the risk to human life even if that happens was additionally astronomically low. You say that is sufficient risk, so where is your threshold of acceptable risk?
2. What is considered a non-crucial feature?

Given your mindset, I suppose you'd be ok for all Tesla's to have their acceleration limited to 0-60mph in 10.0 seconds until safety features of AP2 are fully implemented (including using all 8 cameras, being able to prevent ALL accidents due to user accelerating too fast)? Acceleration being the non-crucial feature you say it's ok to disable, and the risk being getting into accidents, however minor and however astronomically low the chances of them occurring are.

While you may be ok with such nanny moves, a lot of customers are not, myself included. If the risk is astronomically low of an event that isn't even likely to cause bodily harm, I want the feature I paid for. Life is full of risks, and living means taking calculated risks. ~100 people die on average in car accidents EVERY DAY in just the USA - does that mean we should all stay home?

 

[tomas]

I think where you and I disconnect is on the following:

1. What is the threshold of risk that you are willing to disable non-crucial features for? In this case Elon stated that the chance of of battery puncture was astronomically low and the risk to human life even if that happens was additionally astronomically low. You say that is sufficient risk, so where is your threshold of acceptable risk?
2. What is considered a non-crucial feature?

Given your mindset, I suppose you'd be ok for all Tesla's to have their acceleration limited to 0-60mph in 10.0 seconds until safety features of AP2 are fully implemented (including using all 8 cameras, being able to prevent ALL accidents due to user accelerating too fast)? Acceleration being the non-crucial feature you say it's ok to disable, and the risk being getting into accidents, however minor and however astronomically low the chances of them occurring are.

While you may be ok with such nanny moves, a lot of customers are not, myself included. If the risk is astronomically low of an event that isn't even likely to cause bodily harm, I want the feature I paid for. Life is full of risks, and living means taking calculated risks. ~100 people die on average in car accidents EVERY DAY in just the USA - does that mean we should all stay home?

Please do not put words in my mouth. How do you read my "mindset"? Are you Carnak?

I personally see lowering and AP as two different things. One more road hazard fire and Tesla may have been dead. Not rational, but public perception isn't. AP is not a standard feature as lowering was. It is a beta where you acknowledge your responsibility in order to use it. And Tesla has logs. So what you do is transparent. I see no reason to change that.

 

[whitex]

Please do not put words in my mouth. How do you read my "mindset"? Are you Carnak?

I personally see lowering and AP as two different things. One more road hazard fire and Tesla may have been dead. Not rational, but public perception isn't. AP is not a standard feature as lowering was. It is a beta where you acknowledge your responsibility in order to use it. And Tesla has logs. So what you do is transparent. I see no reason to change that.

Ok, so why didn't they just call SAS "Beta", add an acknowledgement like for AP, and allow people to use it at their own risk, rather than sneak in a disablement in a firmware without warning even their own service folks?

Second, before you said Tesla disabled SAS for customer's safety, now you say it was only perception of safety, therefore not really for the safety of customers but to keep Tesla from having PR issues.

Third, you never answered my questions, what do you consider a non-crucial function that Tesla should be able to disable without warning? We know SAS is one of them, how about quick acceleration? Is acceleration more or less crucial than SAS? Lastly, what kind of risk to personal property is sufficient to justify disabling such non-crucial features? One in a million chance of $100K damage, one in a billion chance of $1,000 damage? Where do you draw the line?

 

[conman]

As I posted elsewhere, relevant:
Andrej Karpathy on Twitter

We're hiring strong ML/CV/Roboticists for the Tesla Autopilot Vision team. We ship autonomy at scale. Join us: [email protected]

 

[tomas]

Ok, so why didn't they just call SAS "Beta", add an acknowledgement like for AP, and allow people to use it at their own risk, rather than sneak in a disablement in a firmware without warning even their own service folks?

Second, before you said Tesla disabled SAS for customer's safety, now you say it was only perception of safety, therefore not really for the safety of customers but to keep Tesla from having PR issues.

Third, you never answered my questions, what do you consider a non-crucial function that Tesla should be able to disable without warning? We know SAS is one of them, how about quick acceleration? Is acceleration more or less crucial than SAS? Lastly, what kind of risk to personal property is sufficient to justify disabling such non-crucial features? One in a million chance of $100K damage, one in a billion chance of $1,000 damage? Where do you draw the line?

1) suspension was already on cars. Could not retroactively declare it beta.
2) I did not. That was someone else. I said it was for safety and reputation.
3) I'm not jumping through hoops for your academic questions. It is very clear that you are intent on being combative. I won't play that game. Bye.

 

[bhzmark]

1) suspension was already on cars. Could not retroactively declare it beta.
2) I did not. That was someone else. I said it was for safety and reputation.
3) I'm not jumping through hoops for your academic questions. It is very clear that you are intent on being combative. I won't play that game. Bye.

Put him on the ignore list and then don't respond so I don't have to read your responses either. Is somebody really going on about the titanium shield and sas changes on a thread about autopilot? The forums ignore function is its best feature.

 

[tomas]

Put him on the ignore list and then don't respond so I don't have to read your responses either. Is somebody really going on about the titanium shield and sas changes on a thread about autopilot? The forums ignore function is its best feature.

Yep. I did half of that. "Bye" meant ignored.