Noah Goodall
@noahgoodall.bsky.social
Transportation researcher. Pre-prints on personal site. Views are my own, not Commonwealth's. https://profile.virginia.edu/njg2q
I'm extremely disappointed that they deceptively compared today's Autopilot to 2003-2013 Teslas for the last 5 years, claiming it showed how Autopilot is so safe. That's blatantly misleading. They paid an expert to argue this in court in Huang v. Tesla, without giving them full info. Outrageous.
November 14, 2025 at 11:21 PM
I'm extremely disappointed that they deceptively compared today's Autopilot to 2003-2013 Teslas for the last 5 years, claiming it showed how Autopilot is so safe. That's blatantly misleading. They paid an expert to argue this in court in Huang v. Tesla, without giving them full info. Outrageous.
In summary for today, this is definitely a step in the right direction. We have clearer crash thresholds, raw crash and mileage numbers for significance testing, and highway/non-highway breakdowns.
November 14, 2025 at 11:21 PM
In summary for today, this is definitely a step in the right direction. We have clearer crash thresholds, raw crash and mileage numbers for significance testing, and highway/non-highway breakdowns.
I generally ignore the comparisons to US crash data because they were extremely apples-to-oranges in the past but they are worth reexamining with these new data. These are big claims.
November 14, 2025 at 10:59 PM
I generally ignore the comparisons to US crash data because they were extremely apples-to-oranges in the past but they are worth reexamining with these new data. These are big claims.
They list 6,226,277,786 FSD miles since 2020 (it's a countup thingy so will be higher going forward), and 3,654,231,991 in prior 12-months. So 58% of FSD miles were in the last 12 months. Big growth.
November 14, 2025 at 10:59 PM
They list 6,226,277,786 FSD miles since 2020 (it's a countup thingy so will be higher going forward), and 3,654,231,991 in prior 12-months. So 58% of FSD miles were in the last 12 months. Big growth.
The safety metrics are for the prior 12 months, and will be updated quarterly, rolling 12-months. Since we have raw miles and crash counts, we can finally do some significance testing, so that's cool. The internet archive will be helpful.
November 14, 2025 at 10:59 PM
The safety metrics are for the prior 12 months, and will be updated quarterly, rolling 12-months. Since we have raw miles and crash counts, we can finally do some significance testing, so that's cool. The internet archive will be helpful.
I'm not as familiar with Mexico's crash rate, I'll try to dig in later. I'm also not sure Tesla included them in North America, but I don't see why they wouldn't.
November 14, 2025 at 10:59 PM
I'm not as familiar with Mexico's crash rate, I'll try to dig in later. I'm also not sure Tesla included them in North America, but I don't see why they wouldn't.
One curious thing: they report crash rates by either North America or Worldwide. (FSD is almost exclusively used in NA.) But they compare to general US crash rates. Canada's crash rate is roughly 40% lower than the U.S.'s. So we really should be looking at US FSD on its own.
November 14, 2025 at 10:59 PM
One curious thing: they report crash rates by either North America or Worldwide. (FSD is almost exclusively used in NA.) But they compare to general US crash rates. Canada's crash rate is roughly 40% lower than the U.S.'s. So we really should be looking at US FSD on its own.
And non-highway.
November 14, 2025 at 10:59 PM
And non-highway.
Just seeing that the data is also broken down by highway and non-highway. Not really clear how they define "highway" but their definition is below. Here's highway data.
November 14, 2025 at 10:59 PM
Just seeing that the data is also broken down by highway and non-highway. Not really clear how they define "highway" but their definition is below. Here's highway data.
To be clear, for 5 years Tesla compared Autopilot to a pre-2014 Tesla, ignoring all the 2015-2024 Teslas that were simply not running AP. This was done to juice their safety stats. It is highly misleading, and inexcusable when selling something that people put their kids in.
November 14, 2025 at 10:59 PM
To be clear, for 5 years Tesla compared Autopilot to a pre-2014 Tesla, ignoring all the 2015-2024 Teslas that were simply not running AP. This was done to juice their safety stats. It is highly misleading, and inexcusable when selling something that people put their kids in.
For the last 5 years, Tesla compared Autopilot to a Tesla without any active safety features, which always confused me because who turns off automatic emergency braking? But now we know it was pre-2014 Teslas. Tesla got away with this for years, and really needs to be held accountable.
November 14, 2025 at 10:59 PM
For the last 5 years, Tesla compared Autopilot to a Tesla without any active safety features, which always confused me because who turns off automatic emergency braking? But now we know it was pre-2014 Teslas. Tesla got away with this for years, and really needs to be held accountable.
For minor crashes, we finally have a clear delta-v definition of "Delta-V ≥ 8 km/h within 150 milliseconds." This differs slightly from (Delta-V ≥ 8 km/h within 150 milliseconds. Weirdly, it is NOT consistent with 49 C.F.R. § 563.5 which uses 250ms, but probably not significant.
November 14, 2025 at 10:59 PM
For minor crashes, we finally have a clear delta-v definition of "Delta-V ≥ 8 km/h within 150 milliseconds." This differs slightly from (Delta-V ≥ 8 km/h within 150 milliseconds. Weirdly, it is NOT consistent with 49 C.F.R. § 563.5 which uses 250ms, but probably not significant.
We finally have a clearer crash definition, with 2 thresholds. The "major" one is airbags (like we thought) plus pyrotechnics which is probably seat-belt pretensioners, maybe others things as well. I'll leave that to vehicle experts.
November 14, 2025 at 10:59 PM
We finally have a clearer crash definition, with 2 thresholds. The "major" one is airbags (like we thought) plus pyrotechnics which is probably seat-belt pretensioners, maybe others things as well. I'll leave that to vehicle experts.
Let's try again.
November 12, 2025 at 8:48 PM
Let's try again.
I gave public comment at 1:26:50. I encourage anyone to participate. You get 3 minutes, and since the Board members legally can't gamble, this is the only way their learn how predatory some of the "games" can be. Here's my one page handout.
November 3, 2025 at 5:05 PM
I gave public comment at 1:26:50. I encourage anyone to participate. You get 3 minutes, and since the Board members legally can't gamble, this is the only way their learn how predatory some of the "games" can be. Here's my one page handout.
They also previewed a new PSA to discourage parents from giving Lotto tickets to minors for Christmas. Of course, this was necessary because they have for YEARS run ads promoting gambling as Christmas gifts. The PSAs don't even say why it's wrong (it's illegal). youtu.be/TRzDPncr0rE?...
November 3, 2025 at 5:05 PM
They also previewed a new PSA to discourage parents from giving Lotto tickets to minors for Christmas. Of course, this was necessary because they have for YEARS run ads promoting gambling as Christmas gifts. The PSAs don't even say why it's wrong (it's illegal). youtu.be/TRzDPncr0rE?...
Board members asked how to handle sportsbooks that get into the predictions market. Executive Director said that if they do markets on sports, then Virginia Lottery and the sportsbook "need to have a talk about that." But also "we haven't yet had those conversations."
youtu.be/TRzDPncr0rE?...
youtu.be/TRzDPncr0rE?...
Virginia Lottery Board Meeting - October 30, 2025
YouTube video by Virginia Lottery
youtu.be
November 3, 2025 at 5:05 PM
Board members asked how to handle sportsbooks that get into the predictions market. Executive Director said that if they do markets on sports, then Virginia Lottery and the sportsbook "need to have a talk about that." But also "we haven't yet had those conversations."
youtu.be/TRzDPncr0rE?...
youtu.be/TRzDPncr0rE?...
Market share by sportsbook. Fanduel and Draftkings dominate.
November 3, 2025 at 5:05 PM
Market share by sportsbook. Fanduel and Draftkings dominate.
The Virginia Lottery likes to use inspirational quotes in their slide decks. The first, from is from their actual presentation. The second is different, related Maya Angelou quote.
November 3, 2025 at 5:05 PM
The Virginia Lottery likes to use inspirational quotes in their slide decks. The first, from is from their actual presentation. The second is different, related Maya Angelou quote.
There was one bullet about CFTC regulation of prediction markets, but a pretty healthy discussion among board members! Lottery staff seems to be "wait and see" but with even Truth Social getting into it, and I think the time for action is now. Virginia gets $80mm in sports betting tax revenue.
November 3, 2025 at 5:05 PM
There was one bullet about CFTC regulation of prediction markets, but a pretty healthy discussion among board members! Lottery staff seems to be "wait and see" but with even Truth Social getting into it, and I think the time for action is now. Virginia gets $80mm in sports betting tax revenue.
Impressively, they highlighted the increased use of gambling influencers, and even showed a few TikToks. They mentioned anti-gambling influencer accounts, but created kind of a false equivalency. (Pro-gambling accounts are industry-funded and far more prevelant.)
November 3, 2025 at 5:05 PM
Impressively, they highlighted the increased use of gambling influencers, and even showed a few TikToks. They mentioned anti-gambling influencer accounts, but created kind of a false equivalency. (Pro-gambling accounts are industry-funded and far more prevelant.)