FSD rewrite will go out on Oct 20 to limited beta

[Dutchie]

Just for fun.

I predict the FSD beta next week will do turns at intersections, will do a full route on NOA, will have new visualizations like what we saw on Autonomy Day, and will be more reliable than our current AP but still require active driver supervision.

Duh?! Why do you think Elon said that it will go out to careful and safe drivers?

 

[boonedocks]

Huh?! You are part of the super super secret tester who now already has the rewrite? I thought the first batch of the rewrite FSD would go out next week.

Huh? Nope.

https://twitter.com/elonmusk/status/1314287894962622466?s=20

 

[Dutchie]

Huh? Nope.

https://twitter.com/elonmusk/status/1314287894962622466?s=20

no, the question was “So I'm wondering if this rewrite fixed the phantom braking that has appeared in recent builds?...”. This rewrite has not been made public yet, so you would not know..

 

[MorrisonHiker]

no, the question was “So I'm wondering if this rewrite fixed the phantom braking that has appeared in recent builds?...”. This rewrite has not been made public yet, so you would not know..

Elon stated:


The latest wide release is 2020.40.x. It won't need to wait for the FSD rewrite release according to Elon's response.

 

[NHK X]

Just for fun.

I predict the FSD beta next week will do turns at intersections, will do a full route on NOA, will have new visualizations like what we saw on Autonomy Day, and will be more reliable than our current AP but still require active driver supervision.

Agree. Will definitely be active supervision for the time being. Don’t think Elon recently has been saying with wouldn’t be. Feature complete but not unsupervised.

 

[boonedocks]

Elon stated:
View attachment 598132

The latest wide release is 2020.40.x. It won't need to wait for the FSD rewrite release according to Elon's response.

He doesn’t understand ¯\_(ツ)_/¯

 

[mikes_fsd]

He doesn’t understand ¯\_(ツ)_/¯

He understands fine, his point is/was that you are posting your observation of phantom braking being not fixed -- on a non post-rewrite build -- to the "FSD rewrite" thread.


Edit: ¯\_(ツ)_/¯

 

[Redbrick]

I am not making stuff up just to cast doubt because I don't like Tesla's FSD approach.

As Powertoold just reminded us, Elon said that Tesla would be responsible for a robotaxi related accident. So he said Tesla would assume responsibility.

Also autonomous driving means the human is not driving. If an automaker sells an autonomous driving system and says the human does not need to pay attention to the driving, then I don't see how the human would be held responsible unless it was an accident that had nothing to do with the driving.

exactly....you could apply his elevator analogy to scenario easily..

 

[Redbrick]

I don't follow what you're saying. Why would people that did not buy FSD receive a different version (or any version*) of FSD? And even if they did, why would that matter anyway, given that they don't have FSD?

* Okay, we do need to consider the fact that in reality, FSD is running in shadow mode on cars that do not have FSD, so I would expect that all vehicles capable of running the rewritten FSD NN/software will likely get the update once it goes into wide release**

** And now we need to consider that pre-HW3 cars will not be able to run the latest FSD software. So in THAT sense, there may be a subset of vehicles that have not bought FSD that are running a subset of the code (or maybe even fork of the old (current) software. Is this what you are talking about? And if so, again, why is that significant? I guess you could say that future active safety features may work better on HW3 equipped vehicles, so there may be some "have-nots" on HW2, but you can also make the statement that whatever active safety features they have are probably going to exceed what the car had when they first bought it.

I was simply speculating on the resources required to maintain distinct versions of the software...better to focus all efforts on one with missing features than to maintain two separate variants like what we have now. For example I once asked why tesla continues to send updates when fsd is just around the corner? I wonder if we're going down that path again.....or is it like a 'home' version vs a 'pro' version which would seem to be preferable from a resource standpoint.

 

[Redbrick]

Elon has made the FSD race quite entertaining for the last few years though. Other FSD companies aren't making ridiculous statements and then busting their ass to live up to what they said lol.

I don't think any of us actually lost anything as consumers. Tesla's AP is and has been the best driver assistance available, so nothing to complain about there.

...as much as I poop poop tesla....the guy is trying....that's admirable... The others are just at the sidelines watching with popcorn...

 

[phil4791eng]

I'm sure they need a UK tester, right? I'm always happy to help out.

Definitely has NOT fixed phantom braking. I had 3 phantom braking events this am drive on my daily commute.

I’ve only had one “phantom brake” event ever and I’ve had my MY since August and I drive a lot.

 

[powertoold]

I think the Tesla AP team is the most productive and motivated FSD team in the world because their work is directly tied to the user experience and safety every day. Tesla is able to send out AP updates to thousands of cars with a push of a button. You also have Elon making bold statements about the capabilities of the software. Anyone's who's worked on large software projects (or anything big in life) knows the importance and difficulty of staying motivated and productive. The Tesla AP team gets to see and feel the fruits of their labor in the most direct and satisfying way. We see this in the form of the many YouTubers who constantly test every new update. Karpathy emphasizes this constantly whenever he does his recruiting spiel. I think that's why they've been able to meet so many of the challenges of FSD development. Just my 2 cents as an aging and less motivated software developer anyway

Just imagine yourself as an AP developer with access to the AP fails where someone is hurt or even killed. The amount of immediate responsibility is immense. The drive to improve is proportionately strong.

 

[boonedocks]

He understands fine, his point is/was that you are posting your observation of phantom braking being not fixed -- on a non post-rewrite build -- to the "FSD rewrite" thread.


Edit: ¯\_(ツ)_/¯

Latest wide realest does not future wide release¯\_(ツ)_/¯ it means currently available wide release. ¯\_(ツ)_/¯ which would be 2020.40.

 

[mikes_fsd]

Latest wide realest does not future wide release¯\_(ツ)_/¯ it means currently available wide release. ¯\_(ツ)_/¯

You don't seem to understand that in this thread we are not discussing phantom breaking in the "last wide release" but the "FSD rewrite release" and now phantom braking "fixed with the FSD rewrite".
There is a separate thread that discusses Musk's comments and how it does not seem to have fixed the issue.
I know you know this, because you've already commented on that thread as well.

¯\_(ツ)_/¯ ¯\_(ツ)_/¯

This is what you replied to:

So I'm wondering if this rewrite fixed the phantom braking that has appeared in recent builds? I've heard from numerous co-workers that have been experiencing it (me too), on both HW2.5 and 3.0 cars. So much for Tesla's regression test suites - if it's not catching this new bug in current releases, it won't catch it in this new FSD beta.

And what @Dutchie commented about "you having access to FSD rewrite".

 

[whitex]

At the time you paid for FSD, other premium driver assistance packages were just as costly, were not as good, and had nowhere as much potential.

The reason I said that we didn't lose anything as consumers because Tesla's driver assistance system was and is currently the best you can buy for the money. And at this point, it may become even better than we expected with the rewrite.

As a consumer you can consider your money lost if you got absolutely zero functionality for it, which people who leased a car with FSD in 2016 did. It's not that it wasn't as good as advertised, it literally did nothing, zip, nada. If you paid for EAP in 2016, you enjoyed 100% the same set of features as someone who paid thousands more for EAP+FSD. Perhaps you are thinking today's definition of FSD - you might not be aware of the fact that in 2019 Tesla moved a bunch of features from EAP to FSD - that was the only way they could deliver anything of FSD. IIRC traffic lights the very fist feature which anyone who paid for FSD in 2016 would be getting, everything else (e.g. NoA) was part of EAP back then.

 

[ZeApelido]

*Other* companies don't need to do rewrites because they use the right approach from the start.

This is utterly ridiculous.

 

[diplomat33]

This is utterly ridiculous.

It was tongue in cheek.

 

[powertoold]

As a consumer you can consider your money lost if you got absolutely zero functionality for it, which people who leased a car with FSD in 2016 did. It's not that it wasn't as good as advertised, it literally did nothing, zip, nada. If you paid for EAP in 2016, you enjoyed 100% the same set of features as someone who paid thousands more for EAP+FSD. Perhaps you are thinking today's definition of FSD - you might not be aware of the fact that in 2019 Tesla moved a bunch of features from EAP to FSD - that was the only way they could deliver anything of FSD. IIRC traffic lights the very fist feature which anyone who paid for FSD in 2016 would be getting, everything else (e.g. NoA) was part of EAP back then.

I see. Yea, doing a 3 year lease with FSD in 2016 didn't provide any benefit. It is and was difficult to predict when Tesla would provide useful FSD features, and I guess the people who took that risk didn't end up getting anything within their lease.

What I was trying to emphasize was that that price of EAP + FSD ($5k + $3k) was comparable to other premium driver assistance packages available at the time (which often required you to buy the high end trims and unnecessary extras). And with FSD, at least you paid for the potential for more exciting features, which unfortunately didn't arrive within the 3 year lease and would have arrived within a 5 year lease

And honestly, the price of FSD at the time $3k was very low to reflect the likelihood of actual FSD coming out any time soon.

 

[whitex]

I see. Yea, doing a 3 year lease with FSD in 2016 didn't provide any benefit. It is and was difficult to predict when Tesla would provide useful FSD features, and I guess the people who took that risk didn't end up getting anything within their lease.

In 2016 Tesla showed a video of FSD in action, navigating intersections and all, and stated that it's only pending software validation and potential regulatory challenges. Are you saying it was reasonable for people to assume it would take more than 3 years to validate a system they claimed they already had?

 

[Chickenlittle]

Big news! FSD going out on Oct 20 to limited beta.

If we don't get in the limited beta, I am sure there will be lots of leaked videos of the FSD.

Should go out to the worlds worst drivers first. Just like vaccine testing do highest likelihood to get infection to see largest change. If worst drivers see remarkable improvement-success