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

Crazy conspiracy theory tinfoil hat time in regards to AP...

This site may earn commission on affiliate links.

cdub

OG 2011 Leaf / 2023 Model Y LR
Oct 7, 2014
1,072
1,831
Glendale, CA
I had this crazy shower thought this morning...

What if we are seeing a lot of reports of AP performing poorer than expected because of something nefarious in the corporate sabotage realm?

What if before the former director of AP left Tesla (the one who allegedly stole their secrets and whom Tesla is suing), a "poison pill" was inserted deep in the AP code to make AP perform poorer than usual?

Then... he could say... look how poor AP is with Tesla is now that I'VE left... everyone should invest in MY company instead.

Just a crazy thought.

Is it crazy? YES. I've been reading way too many spy novels recently.
Is it likely? NOT REALLY. It's probably just growing pains with the new hardware and software.
Should Tesla look in their code for it? DEFINITELY.

NOTE: I am NOT saying this guy did this. I don't think it's likely at all... but it's certainly something that Telsa should at least look into the possibility of.
 
  • Funny
  • Like
Reactions: diesel and Cowby
1, That argument would backfire. "It happened right after my watch" in the software realm would trigger internal examination and a scathing blog post that reveals otherwise.
2. Criminal behavior. Not worth the gamble for an intellectual.

I agree. I've been reading too many spy novels. That's why I prefaced it with a crazy tinfoil hat. :)
 
but it's certainly something that Tesla should at least look into the possibility of.
Nah, this has nothing to do with Sterling. Although, you might be able to sort of build off AP 1.0 data, it's much more difficult than that because you're incorporating more data sources. It simply needs more training data with the current hardware. This process should go fairly quickly compared to how long AP 1.0 took. It's not going to be overnight, but it's clear that with repetition and hitting corner cases that it'll do just fine.

Consider the FSD demo, it used the same hardware except that it was trained over and over on the same stretch of road until it was nearly perfect. Imagine having a much larger data set. Instead of it needing to be the same path over and over, it'll be able to generalize and be usable practically anywhere. We simply need to give it time for more miles to be driven by AP 2.0 equipped cars.
 
I suppose it would be possible for someone to apply solar film in front of the camera(s) decrease it's performance.

I think bulk of AP2 issues are due to staffing reshuffle which probably delayed AP2 by a quarter or so.

We will probably learn more as the things unfold as we did with Model X.
 
You don't think it has anything to do with an entirely new hardware platform and additional sensors?

This. And possibly there were fatal flaws discovered that can't be surmounted hence the deafening silence and the trickle release pace?

Who knows...

But if tesla Inc. would only talk a little more forthrightly to its owners, none of us would have to be wasting time speculating like this.
 
You don't think it has anything to do with an entirely new hardware platform and additional sensors?

Yes, of course, AP2 needs more training data- after all it can make fully autonomous trip when given time to learn.

Every time when management changes it slows down development. Given Tesla's tight road-map it probably affected development even more so( from what I read it was not just the top guy but also three developers that left )

I am very optimistic about this however because I think Chris is better suited to carry this project forward.
 
Since this is a tinfoil thread, imagine Chris's reaction when he sees what's left of the code base once MobileEye's system code is extracted:

##
10 PRINT "Hello World!"
20 GOTO 10
##

"That's all that's left?"
 
1, That argument would backfire. "It happened right after my watch" in the software realm would trigger internal examination and a scathing blog post that reveals otherwise.
2. Criminal behavior. Not worth the gamble for an intellectual.

Made me ponder all the 'intellectuals' at VW that thought emissions 'cheats' in software could be hidden and kept quiet...