"Did the car run out of charge and have to get rescued? No no, everyone just got so excited about pulling off the first coast-to-coast zero-intervention road trip that they didn’t want to risk driving anywhere difficult, so someone followed him out to the ass-end of nowhere with a mobile charging station."
This is a misrepresentation, and it would be interesting to hear what your source is.
"David Moss came by and we didn’t want to risk any non-FSD mileage so we brought the charger over to his car 👌 ⚡️"
I didn't mean to imply they did that for the whole trip or anything. Just that the nervousness about making sure nothing spoiled the record is evidence that it's not quite a no-brainer yet for FSD to go this long without disengagements.
"they didn’t want to risk driving anywhere difficult, so someone followed him out to the ass-end of nowhere with a mobile charging station" - So this part of the article is wrong and needs to be edited out imho. Nooone followed him. He went to @itskyleconner at their @Out_of_Spec test track. See the video on the post I linked to.
Now I'm nervous about stealth-editing and making this discussion confusing so I've now marked my new edit and pointed people to the comments here.
I'm still not sure how else to interpret "didn't want to risk any non-FSD mileage so we brought the charger over to his car".
Here's what I originally wrote, so people can judge how unfair I was initialy:
[begin original]
Did the car run out of charge and have to get rescued? No no, everyone just got so excited about pulling off the first coast-to-coast zero-intervention road trip that they didn’t want to risk driving anywhere difficult, so someone followed him out to the ass-end of nowhere with a mobile charging station.
But I’m not actually crying foul or begrudging this person aiming for idealized conditions for hitting this milestone. I have no doubt that others will follow with no deviations from a human-piloted cross-country road trip. I demand zero rooftop breakdancing.
And of course being able to navigate random patches of unmapped dirt is itself impressive.
[end original]
So I've edited it to be clearer that this only happened towards the end of the trip, that he used normal charging stations most of the time, and now an extra edit questioning whether it's fair to say he was followed by a mobile charging station even that one time.
It's possible that even if it's all technically correct now, it still gives a misleading impression about how contrived this milestone was. Of course that's kind of my whole point in the other direction, that touting this milestone would be a little misleading even with nary a thumb on the scales. Totally dismissing it would also be misleading though, and I'm aiming for the right balance in this post.
PS: I've also now edited that bit to emphasize that that was an anomaly and they mostly used normal charging stations. In the original I did already emphasize that I didn't think it was cheating and I expect fully normal cross-country trips to soon follow in any case.
Thanks for catching that spot where I was giving a slightly unfair impression of the trip!
"how crazy it is not to take advantage of such wildly superhuman sensing capability"
Tesla DOES use LiDAR data in testing to benchmark and validate the vision system's performance (e.g., detecting obstacles sooner with less jitter). See this post and video from Karpathy https://x.com/karpathy/status/1407017269083865095
Thank you, I agree this is worth clarifying about Tesla's relationship with lidar and I'm sure it improves the training to have that additional data stream to compare against. Karpathy seems pretty smart and trustworthy.
I just think most of the value is to have that data stream in deployment, to spot a ball rolling into the street that a kid might be about to chase after, or someone on a bike on a foggy night backlit by some construction lights or whatever crazy situation. The kind of thing that happens every 100k miles or something, and that it's no surprise were *not* encountered by a single driver over the course of a few months.
But I'm not expressing confidence that it's a matter of time before a self-driving Tesla kills someone or anything like that. Just that milestones like this, exciting as they are, are weaker Bayesian evidence than they seem in terms of Tesla being actually level 4, where it's safe to read a book in the driver's seat. I will believe that when I see it. I've been bouncing around between 20-60% that we'll see it in 2026.
Thank you! I love this bastion of sanity we have here where people care about what's true. It seems like the rest of the internet either thinks Tesla is an Enron-level fraud or else that Elon Musk is the love-child of Albert Einstein and Tony Stark and we'll all be driving Teslas on Mars by 2030.
"Did the car run out of charge and have to get rescued? No no, everyone just got so excited about pulling off the first coast-to-coast zero-intervention road trip that they didn’t want to risk driving anywhere difficult, so someone followed him out to the ass-end of nowhere with a mobile charging station."
This is a misrepresentation, and it would be interesting to hear what your source is.
The original tweet https://x.com/DavidMoss/status/2006793347613233291 explains that he made a visit to a test track that did not have a charger (probably should have planned that part of the trip better). All in all, he used Tesla's network to charge https://x.com/DavidMoss/status/2006590852655895018
Here's my source: https://x.com/itskyleconner/status/2006462796826099714
"David Moss came by and we didn’t want to risk any non-FSD mileage so we brought the charger over to his car 👌 ⚡️"
I didn't mean to imply they did that for the whole trip or anything. Just that the nervousness about making sure nothing spoiled the record is evidence that it's not quite a no-brainer yet for FSD to go this long without disengagements.
"they didn’t want to risk driving anywhere difficult, so someone followed him out to the ass-end of nowhere with a mobile charging station" - So this part of the article is wrong and needs to be edited out imho. Nooone followed him. He went to @itskyleconner at their @Out_of_Spec test track. See the video on the post I linked to.
Now I'm nervous about stealth-editing and making this discussion confusing so I've now marked my new edit and pointed people to the comments here.
I'm still not sure how else to interpret "didn't want to risk any non-FSD mileage so we brought the charger over to his car".
Here's what I originally wrote, so people can judge how unfair I was initialy:
[begin original]
Did the car run out of charge and have to get rescued? No no, everyone just got so excited about pulling off the first coast-to-coast zero-intervention road trip that they didn’t want to risk driving anywhere difficult, so someone followed him out to the ass-end of nowhere with a mobile charging station.
But I’m not actually crying foul or begrudging this person aiming for idealized conditions for hitting this milestone. I have no doubt that others will follow with no deviations from a human-piloted cross-country road trip. I demand zero rooftop breakdancing.
And of course being able to navigate random patches of unmapped dirt is itself impressive.
[end original]
So I've edited it to be clearer that this only happened towards the end of the trip, that he used normal charging stations most of the time, and now an extra edit questioning whether it's fair to say he was followed by a mobile charging station even that one time.
It's possible that even if it's all technically correct now, it still gives a misleading impression about how contrived this milestone was. Of course that's kind of my whole point in the other direction, that touting this milestone would be a little misleading even with nary a thumb on the scales. Totally dismissing it would also be misleading though, and I'm aiming for the right balance in this post.
PS: I've also now edited that bit to emphasize that that was an anomaly and they mostly used normal charging stations. In the original I did already emphasize that I didn't think it was cheating and I expect fully normal cross-country trips to soon follow in any case.
Thanks for catching that spot where I was giving a slightly unfair impression of the trip!
"how crazy it is not to take advantage of such wildly superhuman sensing capability"
Tesla DOES use LiDAR data in testing to benchmark and validate the vision system's performance (e.g., detecting obstacles sooner with less jitter). See this post and video from Karpathy https://x.com/karpathy/status/1407017269083865095
You can find many references to "validation units" on Twitter, as Tesla does testing in multiple cities for Robotaxi https://x.com/SawyerMerritt/status/2003523665363894291
Thank you, I agree this is worth clarifying about Tesla's relationship with lidar and I'm sure it improves the training to have that additional data stream to compare against. Karpathy seems pretty smart and trustworthy.
I just think most of the value is to have that data stream in deployment, to spot a ball rolling into the street that a kid might be about to chase after, or someone on a bike on a foggy night backlit by some construction lights or whatever crazy situation. The kind of thing that happens every 100k miles or something, and that it's no surprise were *not* encountered by a single driver over the course of a few months.
But I'm not expressing confidence that it's a matter of time before a self-driving Tesla kills someone or anything like that. Just that milestones like this, exciting as they are, are weaker Bayesian evidence than they seem in terms of Tesla being actually level 4, where it's safe to read a book in the driver's seat. I will believe that when I see it. I've been bouncing around between 20-60% that we'll see it in 2026.
Thank you! I love this bastion of sanity we have here where people care about what's true. It seems like the rest of the internet either thinks Tesla is an Enron-level fraud or else that Elon Musk is the love-child of Albert Einstein and Tony Stark and we'll all be driving Teslas on Mars by 2030.