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

Elon: FSD Beta tweets

This site may earn commission on affiliate links.
"Resupplied the midnight oil and making great strides on 10.13 but need a few more days. Should be available next week."
I would be happy if Elon would just multiply by Pi (over a factor of 3) for his estimates. He would probably still be over-estimating the timeline, but perhaps be closer. As we all know, for the past several years, he has RARELY been accurate on an estimated timeline. Frankly, I can only remember twice.

Even better, I would be even happier to hear him say it isn't coming any time soon and then pleasantly surprise me :)
 
  • Like
Reactions: aronth5
Looking at how many tweets Elon Musk has mentioned a specific version (ignoring replies to questions about a specific version), 10.13 has had the most tweets over the longest time so far:
  1. 10.1: 7 tweets (Aug 15 - Sep 24)
  2. 10.2: 6 tweets (Sep 27 - Oct 11)
  3. 10.3: 4 tweets (Oct 12 - Oct 24)
  4. 10.4: 3 tweets (Oct 27 - Nov 7)
  5. 10.5: 0 tweets
  6. 10.6: 0 tweets
  7. 10.7: 1 tweet (Dec 10)
  8. 10.8: 4 tweets (Dec 17 - Dec 23)
  9. 10.9: 2 tweets (Jan 7 - Jan 17)
  10. 10.10: 0 tweets
  11. 10.11: 0 tweets
  12. 10.12: 5 tweets (Mar 5 - May 14)
  13. 10.13: 8 tweets (May 19 - Jul 29)
Separately, I noticed this old 10.1 tweet about single stack for highway:

He did recently say Beta 11 incorporating highway "hopefully end of next month" (August), so they might accomplish it before the above tweet's anniversary.
 
If manufactures offer L4 vehicles for sale without controls and you encounter "Conditions" and L4 will not operate, what do you do?
This seems like the perfect scenario for an OnStar subscription offering. $500 a year gets you OnStar remote driving assistance where a representative will talk to you through the car’s audio system and send instructions to the car’s self-driving system to guide it out of its predicament. No already a plan subscriber? — OnStar will sign you up on the spot for an extra $200 emergency registration fee.

This is the ink for your cheap ink jet printer. It’s a new profit center for automakers.
 
  • Like
Reactions: diplomat33
Did watch a video and mentioned L4 long hall trucks. They are limited (condition) to highway driving. So that is good example of a L4 that is not geofenced. Makes me wonder if our Tesla could be dual levels. For instance after single stack could highway be rated at L3 and L2 city. Then highway move to L4 and city to L3.

Honestly I think basic production AP stack is good enough to be L3 if Elon and his AP team implemented some hand off mechanism for anything unexpected like certain weather conditions, and maybe set some minor restrictions like speed < 70mph, no lane changes, and staying out of merging lane. But there's no reason for Tesla to take on the liability when their cars are selling like hot cakes and there is no real competition to AP in the consumer space.
 
At shareholder meeting just now, Elon said that they are at about 90% success rate for Chuck's ULT. He says the goal is 100% success rate. FSD Beta 10.13 might go out as early as next week. I forget what the bet was but whoever bet that FSD Beta won't have 90% success rate might just lose if Elon's claims turn out to be accurate when 10.13 comes out.
 
At shareholder meeting just now, Elon said that they are at about 90% success rate for Chuck's ULT. He says the goal is 100% success rate. FSD Beta 10.13 might go out as early as next week. I forget what the bet was but whoever bet that FSD Beta won't have 90% success rate might just lose if Elon's claims turn out to be accurate when 10.13 comes out.
Mine was that it would fail to have a 95% success rate, but no one bet against me.

The bet was $100 to charity of the winners choosing.

Maybe now that Elon has given a goal for this ULT that someone will bet against me.

I don't know if I gave an end date, but I figure Sep 30th would be the end date so 10.13 has to be released to Chuck before then.
 
Mine was that it would fail to have a 95% success rate, but no one bet against me.

The bet was $100 to charity of the winners choosing.

Maybe now that Elon has given a goal for this ULT that someone will bet against me.

I don't know if I gave an end date, but I figure Sep 30th would be the end date so 10.13 has to be released to Chuck before then.
10.13 is not a version number. It's the release date.
 
I find Elon's passion and enthusiasm for the enormous goals surrounding him - genius. I'm from the east, where it's not uncommon to see great exaggeration and crazy goals are a kind of bend will that for some is needed to do better, try harder, and get the team hyped and on board for the impossible! This man may very well die on Mars smiling folks..

It's not a great way of communication for many, but I get it.. and I'm on board for full FSD to be here RSN (does anyone remember Pournelle's Byte column?!) =)))

 
At shareholder meeting just now, Elon said that they are at about 90% success rate for Chuck's ULT. He says the goal is 100% success rate. FSD Beta 10.13 might go out as early as next week. I forget what the bet was but whoever bet that FSD Beta won't have 90% success rate might just lose if Elon's claims turn out to be accurate when 10.13 comes out.
Is 90% one 9 or two 9’s? Need to figure out where we are in the march of nines…
 
At shareholder meeting just now, Elon said that they are at about 90% success rate for Chuck's ULT. He says the goal is 100% success rate. FSD Beta 10.13 might go out as early as next week. I forget what the bet was but whoever bet that FSD Beta won't have 90% success rate might just lose if Elon's claims turn out to be accurate when 10.13 comes out.
That was me. I’m still quite confident. My only concern with losing is if Chuck attempts the turn in light traffic. But that would basically be a victory for me as well (I’d do a double-or-nothing bet contingent on another video being done with heavier traffic, at that point).

Elon seems to have forgotten about the March of 9s entirely. “I think we’re roughly at 90% with your turn…so…we’re almost at 100…so…it’s looking good…so…I’m hopeful it might be next week.” The team is apparently working seven days a week on this!

90% is not even remotely close to 100%!

Anyway, remember the stipulations on the bet. I’ve closed the door entirely on any ability to claim failures as successes, which is likely a key to them getting to the 90% level so far. The last attempt we saw with the Tesla test team got 25% (1 of 4 attempts). However, I would guess they may have counted that as 75%. But they had two failures, because they did not go in a timely manner. What we saw also was a 7-9 second crossing of three lanes - that will need to be cut down to about 4-5 seconds maximum in order to be able to be successful per the terms (otherwise a sufficiently large gap would be missed and that’s a failure).

I do hope I am wrong though - as I said elsewhere, the changes required to do this turn 99% of the time (which while quite a poor performance, I would guess they would release at that level), if they are applied generally elsewhere, would be a huge step forward in other areas. The assertiveness required will be a big step up if it applies generally. I think planning and perception will also need big steps forward to enable the car to shoot the gaps like a human. I wish the team a lot of skill and hope they are successful at their 100% target (but that would take a very long time to test of course).

If 10.13 is not released in a couple more weeks, I’ll cancel the beer bet and call it a win. I definitely think that with sufficient time they will be able to get to the low success rate of 90% even with the strict (but realistic and reasonable) stipulations.
 
Last edited:
That was me. I’m still quite confident. My only concern with losing is if Chuck attempts the turn in light traffic. But that would basically be a victory for me as well (I’d do a double-or-nothing bet contingent on another video being done with heavier traffic, at that point).

Elon seems to have forgotten about the March of 9s entirely. “I think we’re roughly at 90% with your turn…so…we’re almost at 100…so…it’s looking good…so…I’m hopeful it might be next week.” The team is apparently working seven days a week on this!

90% is not even remotely close to 100%!

Anyway, remember the stipulations on the bet. I’ve closed the door entirely on any ability to claim failures as successes, which is likely a key to them getting to the 90% level so far. The last attempt we saw with the Tesla test team got 25% (1 of 4 attempts). However, I would guess they may have counted that as 75%. But they had two failures, because they did not go in a timely manner. What we saw also was a 7-9 second crossing of three lanes - that will need to be cut down to about 4-5 seconds maximum in order to be able to be successful per the terms (otherwise a sufficiently large gap would be missed and that’s a failure).

I do hope I am wrong though - as I said elsewhere, the changes required to do this turn 99% of the time (which while quite a poor performance, I would guess they would release at that level), if they are applied generally elsewhere, would be a huge step forward in other areas. The assertiveness required will be a big step up if it applies generally. I think planning and perception will also need big steps forward to enable the car to shoot the gaps like a human. I wish the team a lot of skill and hope they are successful at their 100% target (but that would take a very long time to test of course).

If 10.13 is not released in a couple more weeks, I’ll cancel the beer bet and call it a win. I definitely think that with sufficient time they will be able to get to the low success rate of 90% even with the strict (but realistic and reasonable) stipulations.

I would add that we don't know what conditions produced Elon's "90% success rate". I am thinking Tesla probably only tested in nice weather during the day. So it is really only 90% when the conditions were favorable, not 90% in general.

My guess is 10.13 will achieve 90-95% in favorable conditions (sunny day, low to mild traffic) and Tesla will call it good enough for wide release since the driver is expected to supervise. And I imagine Tesla is not counting "not going in a timely manner" as a failure. It allows them to boost the success rate.

I think Tesla is probably focused on 2 things: making sure the car does not attempt the turn that would result in a collision with another vehicle and making sure the car does not oversteer or understeer in the turn. The system requires driver supervision, so Tesla is probably not too worried if the system is hesitant, as long as it performs the maneuver safely.