-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
-
Tesla's Supercharger Team was recently laid off. We discuss what this means for the company on today's TMC Podcast streaming live at 1PM PDT. You can watch on X or on YouTube where you can participate in the live chat.
Elon: FSD Beta tweets
- Thread starter Terminator857
- Start date
Sigh. The symbol that ex-Twitter is using for their logo is a defined Unicode Character called Mathematical Double-Struck Capital X “𝕏” (U+1D54F) since 2001.
The new logo is nearly identical to the lowercase "x" in this font for example:
Special Alphabets P04: Free Font Download
Special Alphabets P04 font free download. A great choice for highlighting your design elements.
It think it will be impossible to trademark that logo for X Corp.
Look. It's a bit more nuanced than that.
There have very definitely been times where, if I didn't DO something (namely, disengage and drive myself) Something Bad would have happened. Now, what's following is not just with the current 11.4.4 release, but these have happened:
- On a 3-lane major surface road (lights and all) FSD-b attempted to do an auto lane change to the left into an occupied lane. No surprise, as the car started to move, I took over.
- Controlled intersection, with lights. At this location, once with the current release, several times with the previous releases, when first in line, stopped, with a red light in my direction, the car tried to go and run the light. There's been some discussion over at FSD-b forum about this (I'm not the only one this has happened to), but, as near as anybody can tell, we think that the stop lights' are LED, they flash at a frequency higher than a human can perceive, but the car can "see" it; and the car's sampling period of the image is close to the flashing rate, or some multiple/submultiple of that rate, so, to the car, it looks like a flashing red. So it tries to go and, yeah, there was crossing traffic at all the times. At the last time this occurred, a couple of weeks ago, my SO was in the car, so had a witness.
- Car has tried and, once in a fit of experimentation, I let it try, to run four-way stop signs without stopping. Yep, it'll do that. Might be the lack of a white stripe, might be something else.
- On a two-lane road, with wide lanes, on a particular unprotected left, the car literally swerves to fairly far to the right, blocking following traffic that would like to pass on the right (perfectly legit), so badly that, at one point, a confused follower passed me on the left while the car was trying to go left through the confused person. Mind you, FSD-b probably would have stopped and not actually run into the guy, but, at this point, I wasn't having it.
After a bit, one tends to become aware of the places where FSD-b is probably going to have issues, but, if one is sane, one remains alert and ready to take over at any time. It doesn't take a human much time to realize the collection of hardware and software driving the car is more like a student driver than anything else. Having helped trained two kids (who are now out of the house and are pretty good drivers themselves), I can report that when a student driver is in the driver's seat, the instructor/mentor, if they're the least bit sane, is on high alert, watching, and ready to yell. It's pretty much like that with FSD-b; in fact, FSD-b comes with a no-kidding, serious as a heart attack you-must-click-through-this warning that, "FSD-Beta will do the wrong thing at the worst time."
Do I like driving around in FSD-b? Sure, I like new techie-style stuff. On long trips (mostly superhighways) it is, to my mind, somewhat relaxing: it drives, and, this is not a joke, I supervise. Alertly. It takes care of the minutia of keeping the car in lane, keeping up or slowing down with traffic (which means my brain isn't doing that), and I look for errors and maniac drivers.
So, I can believe that people driving FSD-b tend to have fewer accidents: If they're human and exposed to the foibles of the package, they will be paying attention, and that, at least, will drive the accident rate down.
But is the current FSD-b good enough on its own so the driver doesn't have to pay attention? No way.
Now, going back to the, "It's Junk!" brigade: Dunno what's going on with these folks. On the one hand, some members of the brigade report lots more interventions in worse situations that I, and numerous other people are seeing. On the other hand, it's possible that these people somehow have the idea that the Beta isn't a Beta, it's some kind of finished product. Which it very definitely is not. Better than it has been: but perfect, or near perfect, no way.
There does seem to be some car-to-car variations going on; there's speculation that there's some machine learning going on in the individual cars (lots of arguments about that one); there's been some notable comments from some drivers that recalibrating cameras, or, in one case, discovering some EMI between a GPS antenna and the top center camera (!) that was fixed with a new camera, or other hardware fixes improved the FSD-b considerably; so it's all an interesting collection of potential causes. Like, for example, that suspected sampling rate issue on the red lights, which have been seen by multiple users.
But all the above is why the FSD-b forums exist, so we can all get in there, debate what's going on, and maybe even send some pertinent info back to Tesla for improvements.
Running around screaming, "Junk!" doesn't seem to be moving the whole process forward.
DanCar
Active Member
Knightshade
Well-Known Member
I don't think ANYBODY disagrees working generalized L5 is worth insane amounts of money.
There they disagree- somewhat widely- is how close Tesla is to delivering such a thing.
There they disagree- somewhat widely- is how close Tesla is to delivering such a thing.
There’s literally abundant video evidence of this. Thousands of it. You Tesla fans are something else.
Secondly your statement is completely nonesense and is also confirmed by studies. The worse something is, the more attentive you are because you know it will fail. Even Elon has said this.
You Tesla fans will say anything it’s unbelievable
The global ride-hailing and taxi market size was valued at USD 199.14 billion in 2021 and is expected to expand at a compound annual growth rate (CAGR) of 11.7% from 2022 to 2028.
Autonomy might be worth more as a whole sector, but not that. Also Tesla is not alone in this market nor first. I’d argue they are last, if they’re even in the race.
I honestly have a hard time deciding if it’s all a marketing ploy or if they actually believe vision only will get there in five years.
Knightshade
Well-Known Member
Why would generalized L5 be limited to that tiny market?
Every person on earth who owns a personal vehicle and dislikes driving full time (and can afford it) would also be a customer.
So would every vehicle involved in ground cargo transport. And all on-road public transit.
Again I'm not suggesting Tesla is remotely near delivering this- but the size of the market for whomever does is VASTLY more than you suggest.
You can make up any number, but I wanted to put that one into context. I'm not saying it's limited to that. The global taxi market is still around 200 billion. Again, for context.
There will not be one winner globally in autonomy. Mode and geo will separate the different players - trucking in china will have a few players, robotaxis in the US will have 3-5 perhaps and so on., and I'll be shocked if Tesla is a major player in any market in five years and surprised if they are in ten.
Secondly, I believe there will likely be no privately owned robotaxis in my lifetime. Perhaps L4 in limited ODD:s such as sleep on the highway. If all the stars align, perhaps in 10-15 years.
Thirdly, no one will ever rent out their private car for ride share and clean out vomit on a Sunday morning.
Last edited:
willow_hiller
Well-Known Member
Your mental gymnastics could qualify you for a spot on the US Olympic team.
Tesla has managed to make a driver-assistance package that performs so poorly (will literally kill you if you don't constantly disengage, in your words), that it actually makes drivers safer on average.
Incredible. We need to write NHTSA immediately to inform them to mandate bad driver-assistance on all vehicles to substantially increase safety.
willow_hiller
Well-Known Member
And my point is that, statistically speaking, humans often fail to pay attention. So if FSD Beta was actually constantly placing itself in dangerous situations, we would have seen many deaths and injuries, just based on the small fraction of FSD Beta testers that were inattentive at the time.
It's not possible that every FSD Beta tester has been perfectly attentive for every mile driven. So the only logical conclusion is that many of the "close call" disengagement videos weren't actually as close as they appeared, and FSD Beta would have responded to the situations in a way that eliminated deaths or injuries if it wasn't disengaged.
Quite frankly, the system is too bad right now for people to space out. It's decent on the highway during daytime.
i agree. See point 1. There has been impact accidents but not likely to a large extent that some worried. The Gerber incident would have been a crash. Some might argue it would have been, but no one can convince me otherwise. The bison is dead and so is the dog. But we dont count dogs since Waymo hit one too
Last edited:
Curious how many miles You drive annually on FSDb or are your facts opinions internet assumption based? Not that you shouldn’t have an opinion but when someone doesn’t Own or Use a product I find their input rather speculative. I know My personal experience is Far from what you and a few others banter but at least I can validate my experience from use.
Silicon Desert
Well-Known Member
I'm not sure what that person said to get your response as I ignored that person long long ago, but based on what I have seen in the past, I agree. I have a couple people on ignore who regularly post in this thread.
I own FSD with my Model X. Tesla sold it to me, but won't deliver the beta to my car even after four years. Go figure
I have only driven a few hundred miles on FSDb personally (while visiting friends this spring), but I do know quite a lot about the field in general since it's a long standing hobby of mine.I primarily look at the data on the tracker and the forum reports here along with some Youtubers (CYBRLFT and DirtyTesla to name a few, because they report KPI:s) to form my opinions. I find that approach more reliable than looking at anecdotal evidence from individual shareholders.
Given your vast experience with FSDb and the seemingly slow rate of progress the last 12 months in terms of reliability, do you think your current Tesla will drive your kids across town to practice while you watch along on the phone this decade? Do you think you'll be able to sleep in the car while it drives you on the highway this decade? I am hopeful some cars will be able to, but I do not think those include Teslas on the road today.
Last edited:
Based on thousands of videos of FSD Beta versions 8, 9, 10 and pre 10.69.
This was literally the experience of early FSD Beta versions
I mean are you being serious right now?
But yeah all those disengagements is FSD Beta being 5x better than humans.
It was even wayyyyy more than the rate I'm saying. The accident rate of version 8,9,10 (pre-10.69) without a human intervening was mind numbing. Anyone who is saying otherwise weren't watching the videos. And I watched almost ALL of them back then.
But according to Tesla it was still 5x better than humans even back then. Its nonesense.
Mike1080i
Member
I thought that X logo looked familiar!
Similar threads
