Elon: FSD Beta tweets

[diplomat33]

https://twitter.com/x/status/1429903213726093315

Kinda surprised that Elon would admit that V9.2 is not great. Usually, he is hyping FSD up.

The part about "highway+city" FSD requiring "massive NN retraining" is also interesting. It implies that Tesla has a lot of work left to do. Elon seems to be trying to lower expectations with this tweet.

 

[Microterf]

https://twitter.com/x/status/1429903213726093315

Kinda surprised that Elon would admit that V9.2 is not great. Usually, he is hyping FSD up.

The part about "highway+city" FSD requiring "massive NN retraining" is also interesting. It implies that Tesla has a lot of work left to do. Elon seems to be trying to lower expectations with this tweet.

He's doing that because we want our button.

 

[Terminator857]

Reminds me of what Elon once said: it doesn't matter if the Tesla news is good or bad, sales still go up with either.

 

[diplomat33]

Reminds me of what Elon once said: it doesn't matter if the Tesla news is good or bad, sales still go up with either.

True. CNBC already has an article about Elon's tweet. Even when Elon criticises FSD Beta, he gets free publicity. Maybe that was the point?

Elon Musk says Tesla's latest beta self-driving software is ‘not great’

Musk's critical tweet on Monday came just days after he touted Tesla's prowess with autonomous systems and components for them at an event called Tesla AI Day.

www.cnbc.com

 

[diplomat33]

Interesting FSD beta observation:

https://twitter.com/x/status/1429956587783811083

https://twitter.com/x/status/1429957189519626241

 

[Level 1]

Kinda surprised that Elon would admit that V9.2 is not great. Usually, he is hyping FSD up.

I think in this case it's a reasonable thing to say, given some of the driving behavior regressions the YouTube beta testers found.

 

[DarkandStormy]

https://twitter.com/x/status/1429903213726093315

Kinda surprised that Elon would admit that V9.2 is not great. Usually, he is hyping FSD up.

The part about "highway+city" FSD requiring "massive NN retraining" is also interesting. It implies that Tesla has a lot of work left to do. Elon seems to be trying to lower expectations with this tweet.

Taking money for 5+ years for a product/service that doesn't "work great"....hmm, I'm sure there's a term for that.

 

[RDoc]

Well, obviously the Tesla's next step is to increase the price to $15K.

 

[diplomat33]

Elon just says that V9.3 is "much improved".

https://twitter.com/x/status/1430038942909272071

I'm guessing this means V9.3 is coming this Friday?

 

[Mardak]

Presumably he was talking specifically about 9.2 not being great relative to either 9.1 or 9.3. There were some "last minute issues" discovered before 9.2 release to early access:

https://twitter.com/x/status/1426824562600300544

Perhaps those were regressions from the several "new" things added in that version:

https://twitter.com/x/status/1426934048136499209

So maybe a quick fix workaround was added to release a day late but resulted in "not great" at least for Elon Musk's routes. Looks like he/internal-QA gets the next version less than a week before the rest of early access?

 

[DarkandStormy]

Elon just says that V9.3 is "much improved".

https://twitter.com/x/status/1430038942909272071

I'm guessing this means V9.3 is coming this Friday?

I'm sure he upgraded from 9.2 to 9.3 in the ... *checks notes* ... 9 hours from his previous tweet saying it's not great.

 

[Dan D.]

I'm sure he upgraded from 9.2 to 9.3 in the ... *checks notes* ... 9 hours from his previous tweet saying it's not great.

Are we expecting to see the difference from 9.2 to 9.3 that meets the overall criteria of "not great" -> "much improved"? Possibly Elon ran into a Road Closed sign on 9.2?

 

[gearchruncher]

I'm guessing Elon's experience with any "FSD" version goes this way:

1) He gets it before anyone else. He drives it on limited routes in the LA or Brownsville areas that engineers know he will drive. It works pretty good in those 30 minutes (he's a busy guy, let's not pretend he ever drives it more than that). He tweets that it's awesome.

2) It goes to "FSD" beta testers. They use it in actual, hard traffic and city streets, not just highway entrances. They post YouTube videos. It's not great. Elon sees this and adjusts his opinion and resets his expectations for the next version (but usually doesn't tweet that)

3) Repeat #1, where now his drive is better than the YouTube videos of the previous version, so it must be amazing, because he forgot he already said the previous version was amazing before he learned that it was not.

 

[aronth5]

I'm guessing Elon's experience with any "FSD" version goes this way:

1) He gets it before anyone else. He drives it on limited routes in the LA or Brownsville areas that engineers know he will drive. It works pretty good in those 30 minutes (he's a busy guy, let's not pretend he ever drives it more than that). He tweets that it's awesome.

2) It goes to "FSD" beta testers. They use it in actual, hard traffic and city streets, not just highway entrances. They post YouTube videos. It's not great. Elon sees this and adjusts his opinion and resets his expectations for the next version (but usually doesn't tweet that)

3) Repeat #1, where now his drive is better than the YouTube videos of the previous version, so it must be amazing, because he forgot he already said the previous version was amazing before he learned that it was not.

At least this time he didn't use adjectives like "amazing", "blow you mind" or "super". Just "much" improved. So maybe Elon is getting the hang of this. I'm taking that as progress and we'll get to see how much improved soon enough.

 

[ZeApelido]

I'm guessing Elon's experience with any "FSD" version goes this way:

1) He gets it before anyone else. He drives it on limited routes in the LA or Brownsville areas that engineers know he will drive. It works pretty good in those 30 minutes (he's a busy guy, let's not pretend he ever drives it more than that). He tweets that it's awesome.

2) It goes to "FSD" beta testers. They use it in actual, hard traffic and city streets, not just highway entrances. They post YouTube videos. It's not great. Elon sees this and adjusts his opinion and resets his expectations for the next version (but usually doesn't tweet that)

3) Repeat #1, where now his drive is better than the YouTube videos of the previous version, so it must be amazing, because he forgot he already said the previous version was amazing before he learned that it was not.

Having creating product algorithms that the CEO love to evaluate solely on how it operates for them, I would totally bet this is the case.

 

[diplomat33]

I'm guessing Elon's experience with any "FSD" version goes this way:

1) He gets it before anyone else. He drives it on limited routes in the LA or Brownsville areas that engineers know he will drive. It works pretty good in those 30 minutes (he's a busy guy, let's not pretend he ever drives it more than that). He tweets that it's awesome.

2) It goes to "FSD" beta testers. They use it in actual, hard traffic and city streets, not just highway entrances. They post YouTube videos. It's not great. Elon sees this and adjusts his opinion and resets his expectations for the next version (but usually doesn't tweet that)

3) Repeat #1, where now his drive is better than the YouTube videos of the previous version, so it must be amazing, because he forgot he already said the previous version was amazing before he learned that it was not.

Based on past experience and all the times that Elon has talked about how it works great for him, I think this is totally the case. In fact, I think Elon judges how close Tesla is to "solving FSD" based on how it works in his 30 mn drive. It would explain a lot.

 

[ItsNotAboutTheMoney]

Based on past experience and all the times that Elon has talked about how it works great for him, I think this is totally the case. In fact, I think Elon judges how close Tesla is to "solving FSD" based on how it works in his 30 mn drive. It would explain a lot.

Or, maybe, just maybe, he's lying to try to get people to buy it.

 

[diplomat33]

Elon clarifies his "not great" tweet, saying that it is already "great by most standards" but it's not achieved 1000% safe than average human yet:

https://twitter.com/x/status/1430266644971069446

But wait, he's said before that it needs to be 100-200% safer than human. Now it needs to be 1000% safer than average human? Is Elon moving the goal posts?

 

[dramsey]

He hasn't so much moved the goal posts as forgotten or ignored them.

After almost 5 years of blown FSD schedules from Elon, it's silly to expect consistency from him. Remember when FSD v9 was "mind blowing"? Now it's "Not that great."

It's pretty obvious (to me at least) that we'll never get anything like the FSD features we were promised-- at least not with the current generation hardware.

 

[S4WRXTTCS]

He hasn't so much moved the goal posts as forgotten or ignored them.

After almost 5 years of blown FSD schedules from Elon, it's silly to expect consistency from him. Remember when FSD v9 was "mind blowing"? Now it's "Not that great."

It's pretty obvious (to me at least) that we'll never get anything like the FSD features we were promised-- at least not with the current generation hardware.

At least NoA is still Super human in its released form.