FSD video completely fake?

[mongo]

Ok, let's say I believe you. Let's say there is this magical mythical unicorn code base ( and not @verygreen ape oh, how I miss those reports). But let's go with it, this mythical code base exists !! IT WORKS !!! It's does EVERYTHING!! Billions of miles of driving and it's knocking it out of the park!!

Ok, now with that assumption, let's go back to our fearless leader Elon who's on record of saying "I believe in releasing software as soon as I can prove it's safer then a human driver ala AP1"

So, with those two statements, why on this earth is my car still diving for ditches on well marked major highways? Why does my car swerve so far right on left curves that I have to disengage because I'm riding the lane line. Why is my car not leveraging the "unicorn" codebase and NN's, heck not even the codebase, just the mythical NN's would be just fine. Why does my car still want to ram into the backside of a stopped car at a light if it was previously untracked?

So, your telling me, Tesla has already solved ALLL of this, they just don't want us to stop playing Russian roulette, because it's fun to watch the disengagements I guess?

I think you are arguing that FSD is done. That is not my point. What I am advocating for is that FSD testing and development can occur (even in CA) without being reportable as autonomous driving in CA.

In the non logical based proof by analogy discussion: Did anyone know/ report about the 2020 Roadster or the Semi center driving position before the reveal?

 

[mongo]

Well, in that case, share with us how many rooted Teslas you tinkered with

Are we arguing what 'shadow' means? I'm talking about Tesla testing with their cars, not running full FSD in parallel with consumer vehicles.

Unless the vehicle is an internal development vehicle, there is zero reason to expect it has FSD development code on it. If @verygreen had access to one of those, I'd love to see their report. The only software general population vehicles would have (opinion) is data collection to increase the image database and other interesting edge case items.

Non logic analogy: you can reset the odometer on mule test vehicles. Same button combo does nothing on production vehicles.

 

[croman]

@mongo

@verygreen rooted his APE. (i.e. went to space in the analogy I was using (see my earlier post to which you've responded). If we are to have a conversaton, its helpful not to forget what the hell I've stated.

So either you're a flat earther who believes something after someone verifies it isn't true (i.e. goes to space and sees Earth is a sphere, not flat) or you'll acknowledge the evidence.

Its not about who knows more about auto parts. Verygreen is the real deal.

 

[croman]

Are we arguing what 'shadow' means? I'm talking about Tesla testing with their cars, not running full FSD in parallel with consumer vehicles.

Unless the vehicle is an internal development vehicle, there is zero reason to expect it has FSD development code on it. If @verygreen had access to one of those, I'd love to see their report. The only software general population vehicles would have (opinion) is data collection to increase the image database and other interesting edge case items.

Non logic analogy: you can reset the odometer on mule test vehicles. Same button combo does nothing on production vehicles.

Tesla is saying all cars are in shadow mode all the time. First of all shadow mode can't really help with FSD development. That's been debunked and I'd rather not waste my time going through that again.

Also Tesla doesn't maintain an independent code base for anything. All code is uploaded in snippets. Verygreen noted that as features were being developed and deployed. So your speculation is also wrong based on verygreen's observations and analysis.

 

[mongo]

@mongo

@verygreen rooted his APE. (i.e. went to space in the analogy I was using (see my earlier post to which you've responded). If we are to have a conversaton, its helpful not to forget what the hell I've stated.

So either you're a flat earther who believes something after someone verifies it isn't true (i.e. goes to space and sees Earth is a sphere, not flat) or you'll acknowledge the evidence.

Its not about who knows more about auto parts. Verygreen is the real deal.

AH, sorry I did not tie in your reference to a previous analogy.
I see in review, I was also in err regarding the shadow mode post. I was not using the commonly accepted definition of public vehicles running FSD, I was thinking of the internal vehicles in a FSD+ nag state. My mistake. I understand your argument against that and agree.

 

[croman]

AH, sorry I did not tie in your reference to a previous analogy.
I see in review, I was also in err regarding the shadow mode post. I was not using the commonly accepted definition of public vehicles running FSD, I was thinking of the internal vehicles in a FSD+ nag state. My mistake. I understand your argument against that and agree.

As far as a test fleet with test code, if it does exist, it has not been used in public roads in CA during the 2017 reporting period (which, does not include the alleged FSD test drive)).

That is troubling.

We know Tesla has a beta testing code group (Elon is the alpha in all, including testing apparently) but this is insufficient for real FSD development. The main argument against Cruise and Waymo by Tesla was that they had a larger fleet to gather and refine FSD but we're not seeing it. Tesla is not leveraging the fleet.

@verygreen noted some initial steps but it has been haphazard and disorganized and mostly useless. Hopefully that will change quickly but now we have no inside information and Tesla's statements are disheartening.

 

[daktari]

My point @mongo was that we don't have any evidence the FSD video was real either. All we saw was a car driving with a guy in the front seat, not touching the controls.
I have thought that was the real FSD until recently.

 

[mongo]

My point @mongo was that we don't have any evidence the FSD video was real either. All we saw was a car driving with a guy in the front seat, not touching the controls.
I have thought that was the real FSD until recently.

Yeah, I am operating from the hopeful side.
My theory is that video was made from an evolutionary dead end. The level of regression that can occur due to required changes to achieve final performance is one of the reason I feel Tesla has not released anything cool, it could just end up getting broken later.

 

[d21mike]

Ok, let's say I believe you. Let's say there is this magical mythical unicorn code base ( and not @verygreen ape oh, how I miss those reports). But let's go with it, this mythical code base exists !! IT WORKS !!! It's does EVERYTHING!! Billions of miles of driving and it's knocking it out of the park!!

Ok, now with that assumption, let's go back to our fearless leader Elon who's on record of saying "I believe in releasing software as soon as I can prove it's safer then a human driver ala AP1"

So, with those two statements, why on this earth is my car still diving for ditches on well marked major highways? Why does my car swerve so far right on left curves that I have to disengage because I'm riding the lane line. Why is my car not leveraging the "unicorn" codebase and NN's, heck not even the codebase, just the mythical NN's would be just fine. Why does my car still want to ram into the backside of a stopped car at a light if it was previously untracked?

So, your telling me, Tesla has already solved ALLL of this, they just don't want us to stop playing Russian roulette, because it's fun to watch the disengagements I guess?

You posted earlier that you "believe in evidence". Is your point in this post that if your personal car has these problems it is "evidence" that all or a high percentage of cars have the exact same problems? I personally feel your post is extremely misleading in this discussion on the status of EAP/FSP. The only item in your list that I think can be proven with "evidence" is the last one which is actually documented in the manual provided by Tesla. Maybe you should take your car into a SC to have it checked out. I have seem many many youtube videos which do not show your experience plus my personal experience does not show it. One off experience is not proof of a huge problem. Of course my personal experience is just 1 person so like your personal experience is not "evidence" that there is not a problem.

 

[BigD0g]

You posted earlier that you "believe in evidence". Is your point in this post that if your personal car has these problems it is "evidence" that all or a high percentage of cars have the exact same problems? I personally feel your post is extremely misleading in this discussion on the status of EAP/FSP. The only item in your list that I think can be proven with "evidence" is the last one which is actually documented in the manual provided by Tesla. Maybe you should take your car into a SC to have it checked out. I have seem many many youtube videos which do not show your experience plus my personal experience does not show it. One off experience is not proof of a huge problem. Of course my personal experience is just 1 person so like your personal experience is not "evidence" that there is not a problem.

Ok, I'll stop commenting, my apologies for misleading the group, apologies for bringing my first hand accounts to these forums. I'll leave you all with your fairy tales and unicorns. They really do exist.... I have proof!

 

[d21mike]

Ok, I'll stop commenting, my apologies for misleading the group, apologies for bringing my first hand accounts to these forums. I'll leave you all with your fairy tales and unicorns. They really do exist.... I have proof!

View attachment 277668

Funny.... My (and everyone else on this forum = "all") personal experience is "fairy tales and unicorns" and your personal experience is the only scientific evidence of how bad EAP is.

 

[BigD0g]

Funny.... My (and everyone else on this forum = "all") personal experience is "fairy tales and unicorns" and your personal experience is the only scientific evidence of how bad EAP is.

I just want to correct something, what EAP do we have? We have AP, we do not have any feature differences from AP that I am aware of, rain sensing wipers is not an EAP feature, that was a standard feature when I bought my car, it just took a year+ to deliver it. So, I can't possibly comment on EAP features sucking, because we haven't been delivered any yet.

We can't do off ramp to off ramp, we can't do automatic highway transitions, I sure as heck can't have the car self seek it's park spot....

 

[mongo]

Ok, I'll stop commenting, my apologies for misleading the group, apologies for bringing my first hand accounts to these forums. I'll leave you all with your fairy tales and unicorns. They really do exist.... I have proof!

View attachment 277668

Don't go.
I need someone to tell me I'm off base!

 

[scaesare]

I realize I'm not a rocket scientist, but accomplish or test a cross country trip, wouldn't it have to start in CA and therefor require some form of autonomous miles in CA?

Do Oregon and Washington not count?

 

[d21mike]

I just want to correct something, what EAP do we have? We have AP, we do not have any feature differences from AP that I am aware of, rain sensing wipers is not an EAP feature, that was a standard feature when I bought my car, it just took a year+ to deliver it. So, I can't possibly comment on EAP features sucking, because we haven't been delivered any yet.

We can't do off ramp to off ramp, we can't do automatic highway transitions, I sure as heck can't have the car self seek it's park spot....

I can not argue with this, I was commenting about you posting about "evidence" than posting on your personal experience in a way that would give the impression it is a major problem which for the most part could be very limited. Many do not have most of these problems or there are very rare. The one is red is well know and documented in the Tesla manual. All of us would like to see that resolved.

So, with those two statements, why on this earth is my car still diving for ditches on well marked major highways? Why does my car swerve so far right on left curves that I have to disengage because I'm riding the lane line. Why is my car not leveraging the "unicorn" codebase and NN's, heck not even the codebase, just the mythical NN's would be just fine. Why does my car still want to ram into the backside of a stopped car at a light if it was previously untracked?

 

[scaesare]

Tesla is much further behind than GM/Waymo that I could have ever imagined. To think just a few years ago they were the clear cut leaders in the industry.

I mean christ, some company named "Zoox" logged more AV mile testing in CA than Tesla. Lol

Yes. That's right.

Clearly their requirements to report activity based on testing locations inside vs. outside of CA, and/or the specific legal requirements as defined under CA law when compared against "Zoox" is entirely indicative of their progress.

Thanks for helping us see the light.

 

[verygreen]

My theory is that video was made from an evolutionary dead end.

You mean they used stock NVidia SDK code to produce the video, and then started to redo it from scratch on their own? That's one of the theories that sort of makes sense.

 

[mongo]

You mean they used stock NVidia SDK code to produce the video, and then started to redo it from scratch on their own? That's one of the theories that sort of makes sense.

Maybe from stock, that would be most expedient. Pick a path that avoids objects on the right and lanes on both sides. Sort of the level of a masking tape following robot. Functional, but not necessarily expandable. Guessing Karpathy had a lot to tear up at the low level/ front end (based on tweets around Sept 16th last year).

 

[J1mbo]

For those not on Twitter...

 

[croman]

So I guess its a question of reliability? I'll believe it when I see it.