$12K for FSD is insane

[COS Blue]

Yes, "data sharing" is the key here. From Privacy Notice | Tesla :

So tonight I thought I would verify that Data Sharing is turned off, because I don't ever recall opting in. To my horror, it was on! Ok, horror is a bit strong. I don't have an internal camera (thankfully) but apparently I've been "agreeing" to share footage with Tesla from my outboard cameras. So, PSA: if you care about such things, you can turn off Data Sharing at Safety & Security > Data Sharing.

I wonder if they got the footage of my car slamming on the brakes because it saw a shadow? And maybe they fixed phantom braking based on that footage?

 

[Yelobird]

I can tell you that some companies are forced to upload footage and audio (and other telemetry) in some countries. you can guess which countries those are. this is a fact although I wont be able or willing to prove it, for many reasons. those who work in the field will know this but no one else is supposed to (doh).

there are lots of things known in the industry that those who are outsiders dont know. this is how is it in most fields, though, right? so this should come as no surprise.

there are things that will raise eyebrows, but I'll hold off on that level of tinfoil-ism. suffice to say, there are many level of understanding in what happens behind the scenes and what vendors say is what they say, but its not what is really going on.

some things, they are gagged. some things, they self-gag. some things, they just dont want said for lots of reasons.

its like cell radios. the chipmakers know a lot more about what is going on. those that make chips for towers, even more. those that USE chips, know almost nothing of what's really happening.

the more you work in computers, data and the corp world - the more this is not surprising at all.

Please leave the conspiracy tin foil hat BS out of the forum. Unreal how people love to spin crap with no facts because they know but can’t say….

 

[linux-works]

I'll tell people what I can, because they should have some awareness about this whole thing.

they need to know that vendors are not legally required to disclose a thing to you in terms of their true data privacy. there's no law forcing them and they wont simply volunteer it.

 

[Knightshade]

I'll tell people what I can, because they should have some awareness about this whole thing.

they need to know that vendors are not legally required to disclose a thing to you in terms of their true data privacy. there's no law forcing them and they wont simply volunteer it.

This is absolutely false in a ton of countries in which Tesla operates dude.

Stop making up nonsense and trying to pass it off like fact.

 

[dbldwn02]

I've driven 4 different brands of TACC, including Teslas AP3/EAP.
My wife's Mitsubishi Outlander PHEV TACC is by far the best I've ever driven. Never a false stop, and always a full stop for things like firetrucks.

Take their radar/camera setup and throw it on a Tesla and you got yourself a perfect level 2 system.

 

[JustJohn]

https://www.washingtonpost.com/technology/2022/02/02/tesla-phantom-braking/

 

[Knightshade]

https://www.washingtonpost.com/technology/2022/02/02/tesla-phantom-braking/

ROTFL


"surged" to 107 complaints in the past three months, on a fleet of almost 2 million cars.

Even funnier about half are from November, the month Tesla issued a SW updating with a braking bug that was fixed via OTA just a day later.
Remove that data and complaints end up below average for the other 2 months.

For context BTW when NHTSA investigated Nissan for unintended braking they'd received more complaints than Tesla had, on a number of vehicles about 1/4 the size of Teslas fleet.

Folks who keep pretending this is a Tesla-specific issue and not inherent to speed-adjust cruise have to do a lot of mental gymnastics to hold that narrative together.

 

[terranx]

Folks who keep pretending this is a Tesla-specific issue and not inherent to speed-adjust cruise have to do a lot of mental gymnastics to hold that narrative together.

A major difference between Tesla and everyone else is you are not forced to use traffic aware cruise control in other cars. You can just set standard cruise.

 

[Knightshade]

A major difference between Tesla and everyone else is you are not forced to use traffic aware cruise control in other cars. You can just set standard cruise.

You're not "forced" to use it here either- you can just drive manually like most people did for most of the past 100+ years.


Note too the story mentions "They also commonly referred to issues on two-lane highways" which again sounds like a lot of folks using AP someplace the manual specifically tells you not to.

 

[terranx]

You're not "forced" to use it here either- you can just drive manually like most people did for most of the past 100+ years.


Note too the story mentions "They also commonly referred to issues on two-lane highways" which again sounds like a lot of folks using AP someplace the manual specifically tells you not to.

Sure, but who the **** wants to buy a car without cruise control? I personally find cruise control more necessary in an electric car, since you don't get that feedback of engine noise to keep your speed steady.

 

[Knightshade]

Sure, but who the **** wants to buy a car without cruise control? I personally find cruise control more necessary in an electric car, since you don't get that feedback of engine noise to keep your speed steady.

Ok but then we are back to all cars with active cruise have these issues- some worse than Tesla (I cited such from numerous brands earlier in the thread- as recently as a couple posts ago for Nissan for example).

This "surge" was only about 100 people, out of a fleet of 2 million, and half of those were when a known buggy SW was released that was fixed a day later.

The vast vast vast majority of owners use TACC and AP a ton, and it continues to work fine 99.99% of the time for them, and nobody ever seems to be able to cite any case where the 0.01% of bad behavior actually caused an accident, it just annoyed them.

 

[backinthelab]

https://www.washingtonpost.com/technology/2022/02/02/tesla-phantom-braking/

If our voice-command "bug reports" actually went to Tesla for review then no one would need to contact the NHTSA in the first place. They're welcoming government intervention by not reviewing and remedying these reports internally.

 

[momo3605]

Ok but then we are back to all cars with active cruise have these issues- some worse than Tesla (I cited such from numerous brands earlier in the thread- as recently as a couple posts ago for Nissan for example).

This "surge" was only about 100 people, out of a fleet of 2 million, and half of those were when a known buggy SW was released that was fixed a day later.

The vast vast vast majority of owners use TACC and AP a ton, and it continues to work fine 99.99% of the time for them, and nobody ever seems to be able to cite any case where the 0.01% of bad behavior actually caused an accident, it just annoyed them.

The braking bug was only on FSD Beta. At the time (10.3) there were only people with 100 score. So no. Half of those people were not on The buggy firmware.

 

[Knightshade]

The braking bug was only on FSD Beta. At the time (10.3) there were only people with 100 score. So no. Half of those people were not on The buggy firmware.

I never claimed half were on the bad firmware.

I suggested that half the total complaints in the 3 month period cited (that's what "half of those" means in my post in case you were unclear) came during the single month folks would have been reporting on the bad firmware.

Which, BTW, impacted roughly 12,000 people per the actual recall notice.

So it's hardly a stretch to think most of the bump in braking complaints (an increase from 13 total complaints to 51 total complaints) that month came from that group. That's 38 more people. Out of 12,000. A mere ~0.3% of the total in the beta at the time.

That's why the SURGE OF PROBLEMS headline is so hilarious. The highest reported month in the entire thing is 51 total people complaining, in a fleet of roughly 2 million.

And if you remove the post-recall month the highest reported is 32 people complaining, out of a fleet of 2 million cars, for the entire month (december--- by January the # had already dropped to 24 people)

That's only a "surge" in the context of starting with an already tiny # of people and ending with an only slightly less tiny #.

If our voice-command "bug reports" actually went to Tesla for review then no one would need to contact the NHTSA in the first place. They're welcoming government intervention by not reviewing and remedying these reports internally.

It'd be physically impossible to review every voice report from a fleet of millions of cars.


Bug report is there to bookmark the logs locally, so if you have a genuine problem requiring a service ticket the service folks can reference what the car was doing at the time.

 

[JustJohn]

I never claimed half were on the bad firmware.

I suggested that half the total complaints in the 3 month period cited (that's what "half of those" means in my post in case you were unclear) came during the single month folks would have been reporting on the bad firmware.

Which, BTW, impacted roughly 12,000 people per the actual recall notice.

So it's hardly a stretch to think most of the bump in braking complaints (an increase from 13 total complaints to 51 total complaints) that month came from that group. That's 38 more people. Out of 12,000. A mere ~0.3% of the total in the beta at the time.

That's why the SURGE OF PROBLEMS headline is so hilarious. The highest reported month in the entire thing is 51 total people complaining, in a fleet of roughly 2 million.

And if you remove the post-recall month the highest reported is 32 people complaining, out of a fleet of 2 million cars, for the entire month (december--- by January the # had already dropped to 24 people)

That's only a "surge" in the context of starting with an already tiny # of people and ending with an only slightly less tiny #.








It'd be physically impossible to review every voice report from a fleet of millions of cars.

Automakers measure defects in PPM so 0.3% is a total s*** storm. It doesn't work. It also doesn't work much worse than other mfrs.

Physically impossible...post up some stats on that.

 

[JustJohn]

You're not "forced" to use it here either- you can just drive manually like most people did for most of the past 100+ years.


Note too the story mentions "They also commonly referred to issues on two-lane highways" which again sounds like a lot of folks using AP someplace the manual specifically tells you not to.

The manual looks an awful lot like it was written to explain the shortcomings of something that probably doesn't work as expected.

 

[AndreP]

Does anyone have a snip of the blurb in the manual about not using Autopilot on two-lane highways?

We can forgive a few people missing these notes about Autopilot's limitations, but this isn't a few people. If so many are unaware of this, it seems to suggest that Tesla needs to better educate users or needs to disable the system outside of its ODD to avoid people misusing or abusing it.

 

[Knightshade]

Automakers measure defects in PPM so 0.3% is a total s*** storm.

Probably why the rollout was stopped within hours, and everyone rolled back, or forward, to a new version quickly.


Though even then if you look fleet-wide, it was far less complaints than other car makers had on their systems that had active cruise issues.

As I already pointed out to you.

Nissan as one example had more complaints, on a fleet 1/4 the size.

Physically impossible...post up some stats on that.

On what, the fact you can't manually listen to millions of individual bug reports in a timely fashion?

Are you that out of worthwhile arguments?

The manual looks an awful lot like it was written to explain the shortcomings of something that probably doesn't work as expected.

It must be exhausting being wrong this often my dude.

It's written the same as every other car company manual with the same types of warnings.

https://www.fordservicecontent.com/Ford_Content/Catalog/owner_information/2021-Ford-Mustang-Mach-E-Owners-Manual-Job-2_version-1_om_EN-US_05_2021.pdf

Ford Mach E manual.

Page 222

There's 15 different warnings called out where the system may not work right. Including.

:On rare occasions,
detection issues can occur due to the
road infrastructures, for example bridges,
tunnels and safety barriers. In these
cases, the system may brake late or
unexpectedly

Lest you think oh maybe Ford just sucks too-- nope. EVERYONE has the same warnings.

https://www.cadillac.com/bypass/pcf/gma-content-api/resources/sites/GMA/content/staging/MANUALS/5000/MA5002/en_US/2.0/20_CAD_XT5_OM_en_US_U_84290261B_2019OCT29_2P.pdf

That's Caddy. Page 199 begins their laundry list of WARNINGS when using adaptive cruise...including

Do not use ACC when:
On winding and hilly road

and

On curves, ACC may respond to a vehicle in another lane, or may not have time to react to a vehicle in your lane

ACC may detect a vehicle that is not in your lane and apply the brakes
ACC may occasionally provide an alert and/or braking that is considered unnecessary... this is normal operation. The vehicle does not need service




because every adaptive cruise system by every company has these same issues.

 

[Knightshade]

Does anyone have a snip of the blurb in the manual about not using Autopilot on two-lane highways?

We can forgive a few people missing these notes about Autopilot's limitations, but this isn't a few people. If so many are unaware of this, it seems to suggest that Tesla needs to better educate users or needs to disable the system outside of its ODD to avoid people misusing or abusing it.

This thread is a great example of the fact many never read the manual at all, regardless of how well written or not.

Here's one example from the current Model 3 manual though:

Autosteer is intended for use on controlled-access highways with a fully attentive driver. When using Autosteer, hold the steering wheel and be mindful of road conditions and surrounding traffic. Do not use Autosteer in construction zones, or in areas where bicyclists or pedestrians may be present.

Controlled access highways- which are those with on/off ramps, and no at-grade intersections-- that is traditional divided highways where there can never be oncoming, or cross, traffic.

 

[AndreP]

This thread is a great example of the fact many never read the manual at all, regardless of how well written or not.

Here's one example from the current Model 3 manual though:

Controlled access highways- which are those with on/off ramps, and no at-grade intersections-- that is traditional divided highways where there can never be oncoming, or cross, traffic.

How many people do you think can drive around and determine on the fly whether a road is a "controlled access highway"? Like "nope that one has an at-grade intersection, I better turn off (or not use) Autopilot".

Is that seriously what it says in the manual?