Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

FSD Beta 10.11

This site may earn commission on affiliate links.
First early impressions. It’s handling 3 and 4-way stop signs better. It used to start off and then slow slightly and then pick up speed again which was annoying and sometimes confused other drivers. That’s gone. I just went through a dozen intersections with stop signs in SF with a variety of traffic mixtures and it handled pulling away from the stop smoothly.

Another data point: 10.11 was a little drunken with lane changing and signaling plans and would regularly begin signaling a left or right turn while driving for a click or two and then discontinue. In my brief drive so far today it feels like this may have been fixed but I’m not sure yet. This issue was laughably bad while drive (west?) on Lincoln Way along the south side of Golden Gate Park. I drove a dozen blocks in a straight line along Lincoln Way recently and at almost every intersection I drove through the car would begin momentarily signaling a left turn or lane change as the car approached and drove through the intersection. I plan to try that route later today to see if it has improved.
 
Let’s see if this will be the end of this beta subversion rollout….

0C95FFCC-5B44-4A04-B052-9D4AB28E6CA3.png
 
First early impressions. It’s handling 3 and 4-way stop signs better. It used to start off and then slow slightly and then pick up speed again which was annoying and sometimes confused other drivers. That’s gone. I just went through a dozen intersections with stop signs in SF with a variety of traffic mixtures and it handled pulling away from the stop smoothly.

Another data point: 10.11 was a little drunken with lane changing and signaling plans and would regularly begin signaling a left or right turn while driving for a click or two and then discontinue. In my brief drive so far today it feels like this may have been fixed but I’m not sure yet. This issue was laughably bad while drive (west?) on Lincoln Way along the south side of Golden Gate Park. I drove a dozen blocks in a straight line along Lincoln Way recently and at almost every intersection I drove through the car would begin momentarily signaling a left turn or lane change as the car approached and drove through the intersection. I plan to try that route later today to see if it has improved.
does it still do the stupidity on the highway, driving along happily in the right lane 1.5 miles from our exit, then signals left with intended lane three (yes, 3) lanes over on the left, then after starting to fight its way over it gives up after two lane changes before figuring out it should have stayed in the right lane all along :)
After a few attempts at not disengaging I gave up and just take over as it starts warning of an upcoming lane change.
Thats what makes me disbelieve folks who claim a high number of disengagement free drives - what is the tolerance for stupid driving.
 
I'm not a peasant. I'm a valuable source of data. That's the deal for being a beta user--early access for me, giving up privacy to be a data source for Tesla.

Start from the premise that the people deciding what versions of FSD beta to roll out to the fleet are motivated solely by what's in the interests of Tesla. It's not a product yet, they're not trying to deliver it yet, so as a FSD beta user I shouldn't have any expectation or feel any entitlement to the latest and greatest version.

Given those premises, ask yourself: what is in the interests of Tesla?

First, as a software developer, I want to know whether or not my new version of software is actually an improvement over my old version. The best way to do that is A/B Testing (go look it up). You move a small number of users to the new version and then watch carefully whether the new version is better than the old version. You can't just move everyone to B after a small period of time before you know the A version is better, because in a dynamic environment (to include things like changing seasons) you want to hold constant as many variables as possible.

So what we're seeing is consistent with 10.10.2 being a pretty good release, with a known and reasonable rate of disconnects. 10.11 throws a lot of new tech into the stack, plus adds Canada. It's not too surprising that 10.11.1 and 10.11.2 are being pushed to the 10.11.0 users, as that's consistent with the 10.11 stack needing some serious refinement before it's better than 10.10.2 by metrics that (likely) include disconnects, collision alerts, and so forth.

My interpretation (as a 10.10.2 user) is that I'm in the B (scientific control) group. I'll only get upgraded after the 10.11.x branch both stabilizes and is demonstrated to Tesla management through objective measurements to be better than 10.10.2. That might happen next week if all goes well, but they might not reach that bar until 10.11.3, 10.11.4, etc. My continued use of 10.10.2 is still useful to Tesla to provide that solid baseline, so it doesn't do me any good to sit and stare at the app waiting for upgrade notification for at least a week.
Excellent post - a lot of people should read it.

The majority of people seem to view the FSD beta program as a way to get FSD early and/or their ‘right’ because they signed up and ‘drove like a grandma’ for x days/miles. In reality, Tesla is giving some drivers early access in exchange for them testing the code and giving data. Tesla really gains nothing from letting more people into the program than they need to get data and there are a lot of potential downsides.
 
does it still do the stupidity on the highway, driving along happily in the right lane 1.5 miles from our exit, then signals left with intended lane three (yes, 3) lanes over on the left, then after starting to fight its way over it gives up after two lane changes before figuring out it should have stayed in the right lane all along :)
After a few attempts at not disengaging I gave up and just take over as it starts warning of an upcoming lane change.
Thats what makes me disbelieve folks who claim a high number of disengagement free drives - what is the tolerance for stupid driving.
I’ve seen a lot of ‘stupid’ lame selection in town, too. I was driving in the right lane of a 4 lane 45 mph road last week. I needed to turn right in a mile and a half and there were no cars in my lane at all. Suddenly it turns on the left signal and starts changing into the left lane, nearly cutting someone off. I took over and cancelled the move, but it’s a baffling phenomenon.
 
Last edited:
I’ve seen a lot of ‘stupid’ lame selection in town, too. I was driving in the right lane of a 4 lane 45 mph road last week. I needed to turn right in. Mike and a half and there were no cars in my lane at all. Suddenly it turns on the left signal and starts changing into the left lane, nearly cutting someone off. I took over and cancelled the move, but it’s a baffling phenomenon.
Yes, it’s been pretty dumb about lane change planning for a long time. I haven’t driven it enough yet to know about 10.11.2, but I’m doubting this has changed yet.
 
Excellent post - a lot of people should read it.

The majority of people seem to view the FSD beta program as a way to get FSD early and/or their ‘right’ because they signed up and ‘drove like a grandma’ for x days/miles. In reality, Tesla is giving some drivers early access in exchange for them testing the code and giving data. Tesla really gains nothing from letting more people into the program than they need to get data and there are a lot of potential downsides.
The majority of people view it that way is because Elon said back in September everyone who's driving score qualified would be allowed in the program.
 
  • Like
Reactions: bizi
It looks like 10.11 was released as a 10% sampling of FSD Beta population (roughly 6k vehicles).
What percentage of FSD Beta cars do you think are reporting through TeslaFi?

TeslaFi is currently reporting (as current installs):

2021.44.30.21 (10.10.2): 1843
2022.4.5.15 (10.11): 41
2022.4.5.20 (10.11.1): 88
2022.4.5.21 (10.11.2) 480

That's a total of 2,452 FSD Beta cars, out of an estimated 60,000, so about 4% ? Or am I missing something?

By that math, the 480 at 10.11.2 would represent 480 / 2452 of the population, or about 19.57%.
 
  • Like
Reactions: FSDtester#1