Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register
This site may earn commission on affiliate links.
Crazy that people are still sacrificing their wheels! I just disengage before every right turn (and some left ones).

Perhaps Elon was trying to make Alan feel bette.

True, but it is the unknown bugs that should worry you. ;-)

Imagine Alan won't have to disengage on every right turn any more :) . Alan you don't actually do that do you?
 
  • Like
Reactions: FSDtester#1
You need to have something first to lose it. He is just pissed that @Tronguy
just used years of his own words against him. I like it when he takes shots at you.
He was warned, but doesn't care about looking like the Joker....
Thing is, @AlanSubie4Life actually does have good comments from time to time.

I mean: I'm a believer in Not Being Surprised when driving a two-ton car around the landscape, especially when it's using software "assist" (or whatever-you-want-to-call-it) that, shall we say, isn't guaranteed to Not Make Mistakes. The less surprised I am, the safer it is for me and everybody else on the road.

So, showing up here and hearing everybody's tales of woe and success and dumping mine into the pot as well is one of those survival-positive activities that humans have been doing since, probably, there's been humans. And tips on how to successfully sneak up on mastondons without getting flattened weren't just a nice story: It was, literally, life, death, and whether or not one and one's loved ones starved or not. There's reasons that when one has a good or bad experience one feels an urge to go tell it all to a crowd; and, when someone is telling tales like that, it tends to rivet the attention of the listeners, especially when one is doing something moderately risky (or worse). Both of those reactions are part of what got us all here; those that didn't have those reactions tended to get flushed out of the gene pool over the eons.

So, having said all that, one might imagine that there's a kind of innate, "You'd better play by the rules, Or Else." to all this. Telling believable, but false tales about How To Kill A Mastodon might earn one a rather rough visit by the surviving relatives of somebody who tried that method and failed. And, sometimes, it doesn't have to follow common sense: If memory serves, there were reports of societies back in the long-ago who were in the habit, if there was a drought or something, that would rise up and kill the king.

The reaction to Not Following The Rules is, I suspect, rather instinctive. Hence, Trolls, who Love Reaction, and are safely physically distant from their victims. And if money is involved, then all bets are off.

So, when I show up here and give a trip report or two on a new release, I make a conscious effort to tell not just the good stuff, but the bad stuff as well, warts and all. And I follow an olde-tyme method: "First you tell 'em what you're going to tell 'em, then you tell 'em, then you tell 'em what you told 'em." Just like 3rd grade and how to write a report. And one has to be very, very conscious of human failings: Malleable memory, wishful thinking, whatever-the-heck-the-opposite-of-wishful-thinking is, and so on. And to attempt, however imperfectly, to separate actual facts from evaluation. And, when doing an evaluation on something that's changing like FSD is, to put it in context as much as possible.

With Alan, I've come to the conclusion that his trip reports regarding the facts are real and accurate. He's the one that pointed out to (a disbelieving) me that under FSDS city streets that the car wasn't coming to a nice, steady halt when braking, but rather that it was oscillating in braking force. Next time I was out, the behavior he described (and I think he was the only one describing it) I found to be accurate. Tesla eventually fixed that, plus or minus a minor regression or two. And I took Alan off my block list.

The objective physical facts are one thing. It's Alan's evaluation of those facts that leave something to be desired. As I said before, one has to have a sense of proportion on these things. It's one thing to (rightly) complain when, under 10.x, the car makes an attempt to merge straight into a car in the adjacent lane; it's another, when the car struggles to merge into a line of cars that a human would have a bad time with.

And differing points of view around here are definitely welcome. FSDS is probably out of Junk status, but it sure isn't ready for Robotaxi yet. So a dim view of FSDS isn't exactly wrong.. but going over the top on the evaluation isn't exactly right, either.
 
Speaking of fixing bugs, now that Telsa has done the work getting FSD 12.3.6 to play nice with the base code in 2024.14, perhaps they will now move the group I call "testers" (stuck on 2024.3.25) up to 2024.14.x. The longer they put off the merge, the harder it will be.
I imagine we would have already been moved up to 2024.14 if that was going to happen, especially considering the recent large deployment. 2024.9.x still looks like where we are going to land. The hot weather improvements announced for 2024.20 would be nice to have also, but we'll likely get those in the fall just in time for when it no longer matters.
 
  • Like
Reactions: edseloh
True, but it is the unknown bugs that should worry you. ;-)

Remember that once an unknown bug shows it is known. So unknown bugs are also known bugs according to this Tweet.


Alan you don't actually do that do you?

I disengage routinely on many specific right turns. It’s pretty deterministic. Some are no issue so I can let it go. Others it does the same thing every time so I disengage every time.

Anything which is reasonably tight with no extra room with a curb involved I’ll disengage, as anyone would do. For example I disengage here every time, or press the accelerator vigorously to push it wide though that is a bit less effective these days.
 
Last edited:
I found to be accurate. Tesla eventually fixed that, plus or minus a minor regression or two. And I took Alan off my block list.
Thanks!

But they have not fixed this.

Any user of FSD can observe the regen bar and see (and of course feel) very small oscillations on many slowdowns with the most up-to-date release of 12.3.x. It’s not on every slow down but we can all observe it - you can see it on the many videos of high quality provided, as well, I assume. I haven’t checked but I will leave that as an official exercise for the reader and assume it is not some problem with my car (since people here have helpfully confirmed my observations I think that is reasonable).

This is of course one of the minor details of the various issues with stopping and is not the entire scope.

Hopefully it is fixed in 12.4.
 
Agreed, anyone that hasn't been pushed to .14 branch and is still on 3.25 are considered the testers... they will be 1st to get 12.4. Probably in the same order rollout of the previous v12 releases.
My wife's on 3.25 and hasn't activated her FSD in like two years...
 
I don't have FSD, only tried free FSDS for a month. But I am curious here. When FSD messes up like this, is there risk of an accident? Does FSD act impatient, get increasingly aggressive and try to make the exit? Or remain safe and patient, miss the exit and reroute?
Um. The "free FSDS for a month" is the, no kidding, Real FSDS that all of on this thread have been working/living/whatever with.

First off: No offense, but you've got to change your mindset about Who's Driving. Without diving into the terminology weeds, you are.

At this place, at this time, FSDS might very well be accelerating/braking, moving the steering wheel around, taking turns, doing merges, and all that jazz - but you are monitoring it, continuously. If it does something Not Appropriate, it's Your Job to immediately take over and Drive The Car.

Doing so is called an, "intervention". (There's some terminology arguments these days about "disengagement" or "intervention", but let's not go there.) I find that with FSDS over, say, ten drives or so, with 12.3.6, in my area, I intervene roughly three or so times. Others, with different personal sensitivity to Things Going Wrong might intervene once, or fifteen times. But if the car's about to whang into a railroad crossing or something like a student driver might do, it's very definitely Your Job to take over.

--Edit--
Forgot to put this in. In my experience, when the car's under FSDS control, there's always been time to intervene. No joke, there's plenty of time to blink. There's not enough time to play patty-cake with a texting phone. The driver monitoring (currently, torquing the steering wheels and eyes out the windshield) attempts to enforce this, but, if one is monitoring what the car is doing, then there's always been plenty of time to take over. Without scaring the passengers, even. There are even reports from Tesla that they pay attention to this.
--Edit--

As far as merging and that: FSDS has been getting better at that over the years. There's this one on-ramp near my place onto a busy three-lane local road where the correct move is to merge one lane to the left. The problems are that (1) the on-ramp is also an off-ramp for the busy three-lane road, and lots of traffic is trying to move off the busy three-laner onto the on-ramp when one is trying to get off onto the busy three-laner and (2) this ramp was made when the DMV had lost their sanity and is Too Short for all this activity. Humans occasionally get stuck and have to pull straight ahead onto the breakdown lane when trying to move left, not to mention people on the busy three-laner not being able to get over to the right on time, getting onto the breakdown lane, then backing up to get onto the offramp part of it. (Saner people would take the next off-ramp, turn left, then a right on-ramp, and get back to the interstate they wanted, only a mile or so out of their way.)

12.3.6 takes this ramp fairly well, meaning that it Makes It Over about eight times out of ten. If it didn't make it over, then it'd get stuck on an Interstate Going Elsewhere, taking it miles out of its way. Version 11.x would make it - but pretty much if there was no other traffic out there, so I'd guess at making it two times out of ten.

My experience with generic off/on-ramps is that FSDS does a pretty decent job, most of the time, and, if you're interested, you can see it highlighting cars a bright blue when it decides where it wants to duck in. Mostly it will slide in closely behind someone, braking to do so; sometimes it will zip ahead and duck into a gap; as a guess, about 5%-15% of the time it'll fail, or look like it's failing (usually, during heavy moving traffic), and it's Your Job to take over.
 
Last edited:
Just try and remember the order you received 3.25... if you got it early, you will probably stay in that same sequence.

I have no clue when I got 3.25 (first time I have actually checked my software version) but I did get the one month free trial of V12.3 right away, before anyone else I knew. Not sure if that means anything.
 
  • Like
Reactions: FSDtester#1
Remember that once an unknown bug shows it is known. So unknown bugs are also known bugs according to this Tweet.

I disengage routinely on many specific right turns. It’s pretty deterministic. Some are no issue so I can let it go. Others it does the same thing every time so I disengage every time.

Anything which is reasonably tight with no extra room with a curb involved I’ll disengage, as anyone would do.
Last time I disengaged for a tight corner was months ago. No curb rash on any tires yet. Glad I don't drive on your roads although you can have my low visibility intersections!
 
First off: No offense, but you've got to change your mindset about Who's Driving. Without diving into the terminology weeds, you are.
I have extreme trepidation about reigniting this debate, but I just want to make one short comment: I say I'm driving when there are no automatons assisting, and it's driving when I have engaged TACC or FSDS.

I say I'm legally responsible (with Level 2) no matter who/what is driving.

That's my story and I'm sticking to it.