Waymo admits that some portion of their remote operators are based in the Philippines. The video is illustrative as you can see that Waymo really does not want to answer this question. I would love to know whether they got back to @markey.senate.gov with numbers, and what those were.

Waymo tells the Senate one thing, California regulators another. So do remote operators ever take control of the vehicle? When has this happened? In what circumstances will it happen again? bsky.app/profile/noah...

The claim in this article is not supported by the data. SGO and NHTSA general public data have different reporting thresholds, and the authors also seems to have used raw gen. pub. crash counts, instead of total vehicles INVOLVED in crashes. Better benchmark is around 50-100k miles b/w crashes.

Waymo is telling UK their telops can't do what they told CPUC they can do: “Waymo does not use remote driving or teleoperation where a human takes control of the vehicle....They cannot drive the car,” George Ivanov, Head of International Policy and Government Affairs zagdaily.com/featured/way...

Kalshi loves to talk about how they're a market, not a dirty little sportsbook that sets their own odds. Yet you can only bet "Yes" on NBA player scoring prop bets. There's no "No" option. So who's taking the other side? No one knows! Probably Kalshi or a sportsbook partner. Nice.

It's already been corrected online, but the @newyorker.com print edition has this insane fact. I suspect they got this from the US Sentencing Commission, but that's because sex crimes involving Native Americans and occurring on reservations go to federal court. When including state courts, it's <1%.

@aniccia.bsky.social reminded me that Waymo does teledrive in limited circumstances. This is done by the Event Response Team, who I assume are US-based since they speak directly with 1st responders, as opposed to the remote assist who use pre-recorded messages. www.cpuc.ca.gov/-/media/cpuc...

Ed's correct. Waymos get occasional assistance but I doubt they teledrive ("mechanical turk"). Partly because Waymo has consistently said this AND because there are significant technical limitations - latency would be too high and too variable. But I would love NHTSA to deep dive this.

NHTSA closed their year+ investigation of Waymo. Waymo's 2 recalls re to this investigation were for crashing into a utility pole and roadway barriers. Evidently, robots repeatedly driving on the wrong side of the road into traffic and disobeying traffic control devices is ok in USA. Scale that.

I'm late, but this is the first confirmation I've seen of Waymo using foreign-based remote operators. Prior, there was only a rumored podcast ep, immediately cease-and-desisted. www.telegraph.co.uk/business/202...

Kalshi loves this argument, but it's BS because they have an "independent" trading division who can effectively set prices, so they very much are the house.

File this under things no one asked for: yet another prediction "market" to inside trade on manipulable and/or uninvestable-for-a-reason BS events. We're gonna look back on this time like smoking in the 1950s. We know all the risks but there are dimes to squeeze.

Seeing some small errors in the Tesla safety report for no-active-safety (the 2008-2013 "baseline" vehicles) minor crashes. Here's North America. It looks like similar errors in worldwide, so probably something wrong with their counts.

Tesla now publishing crash counts and mileage. Better yet, they have re-introduced "without FSD/AP, with active safety features" which has been, as far as I could tell, simply deleted entirely since Q2 2021. Mini thread on first impressions. www.tesla.com/fsd/safety

As a former shitty hs kicker, here's what I want: designate a target like an end zone pylon. The further away the ball lands, the better the starting position. That's high drama!

Always fascinated by policies that have 99% approval yet don't happen, more so by those taken away. Politically, this is like banning direct deposit.

Thread on the Virginia Lottery Board meeting held October 30. Full video available. www.youtube.com/watch?v=TRzD...

Virginia's iLottery state-run casino app sales up 28% over Q1 last year. They still have not found the top with this thing.

Revealing Nunes quote about "control" of prediction markets. Clearly he thinks that Truth Social's markets will be more accurate, and therefore any Republican ahead in the markets but losing the election is a victim of voter fraud. Nice.

Truth Social has announced that they are entering prediction markets, powered by crypto dot com. Prediction markets are sportsbooks in disguise, as political events are few and far between. Gee, which button is the biggest? (h/t @matt-levine.bsky.social) www.reuters.com/business/med...

Calls to Virginia's gambling helpline are through the roof. I actually heard the Lottery Board try to explain this as a result of their efforts to promote the helpline LOL. Meanwhile state-run gambling has doubled, particularly the highly-addictive casino-style iLottery.

As seen in the release notes of ChatGPT Atlas. Cool cool cool

Was at the Stop Predatory Gambling national conference over the weekend. Craziest thing I learned was that Twitch competitor Kick is owned by massive gambling site Stake. It was launched after Twitch banned influencers streaming internet slots for their 12-year-old fanbases.

Total money gambled in Virginia. Those lines are different types of gambling, so you need to stack them. Talk about a hockey stick chart.

Apparently Virginia's anti-gambling school curriculum would have a section where they would teach students how to play the games. Awesome.

I always assumed gamblers felt the wins but were numb to losses. But I'm watching a VCPG presentation that says dopamine spikes during wins AND losses. One theory is that gamblers experience a loss as meaning that a win is now closer. The dopamine is about anticipation, and losses build it.

Tesla did not crash 3 times in one day. This confusion is due to NHTSA's SGO database being in Excel, where "July 2025" is encoded as "July 1, 2025." If you look through SGO ADS, every crash "occurs" on the first of the month.

Another day, another traffic signal control paper to review that assumes pedestrians don't exist.

Reviewing for TRB this year and seeing many papers with both straight and smart quotes/apostrophes. Does this indicate: 1. LLM usage 2. Co-authors copying and pasting from LaTeX 3. Other

A thousand times repost, the whole thread, forever.

8 months after the inauguration, USDOT still hiring only FAA and a few Maritime positions. Zero NHTSA or FHWA positions. One NTSB job in HR, first I've seen from them in awhile. www.usajobs.gov/job/845831600

Unsurprisingly, the Minneapolis shooter displayed multiple 30-round magazines, and a possible 50-round, in their YT video. Magazines over 10 rounds are useless in hunting and home defense, yet appear repeatedly in mass shootings. As a gun owner, I want them banned.

Honored to be speaking about state-run casino apps at Stop Predatory Gambling's national conference in October. Unlike other groups, SPG doesn't accept money from industry.

The Army Recreation Machine Program (ARMP) runs slots machines on bases, addicting vulnerable, underpaid 18-year-olds. And further confirmation that industry-funded NCPG mostly exists to provide cover at the last possible second. www.wired.com/story/us-mil...

In honor of the Transportation Research Board annual meeting paper submission deadline closing in 4 hours, here's my running list of research projects I never got around to or was unable to solve. [thread]

A buffet line flush against the wall bringing out the raging engineer in me.

Been looking at CarPlay presence in fatal crashes, and not seeing any clear trends. Since negative results are hard to publish (unless you have an APC budget), I'm sharing them here. [thread]

One thing I've wondered about is how Tesla vehicles NOT running Autopilot are crashing more in the last few years. The first quarter they crashed once every 2 million miles, now it's every 1 million. The tech isn't changing, so what's going on?

Tesla AP is not 10x safer. This is false. In the fine print, Tesla only counts their own crashes when it's an airbag or active safety restraint, whereas "US drivers" counts anything where at least one involved vehicle was towed. Many tow crashes don't deploy airbags. It's apples to oranges.

Tesla's official X account announces FSD (supervised) crash rate data, then links to a quarterly report that mentions only Autopilot. Either they are trying to combine AP and FSD data with no notice or explanation, or they are just repeating the false statement from the earnings call.

In Tesla's earnings call last night, CFO Vaibhav Taneja claimed their new safety report shows FSD is 10x safer than non-FSD. This is false. The new safety report covers AP, not FSD. The most recent FSD data covers 2023, when it was 4x safer than no active safety features at all, excluding even AEB.

I'd just like to point out how weird it is that places with hundreds or even thousands of reporters — the NYT, WaPo, CBS, NBC, AP, etc. — are all getting pummeled on every major story by the dozen computerboys at Wired. To fumble the bag this consistently is, at some point, cowardice and a choice.

I absolutely love the Traffic Records Forum conference. Small, few vendors, and full permission to nerd out. Last year I gave a talk on the differences between 3 crash databases, which would have been a dead poster at TRB but was a packed, engaged room at TRF.

Observed substance use in Connecticut injury and fatal crash data 2017-2023. From a UConn presentation at Traffic Records Forum.

On-board helpers, bad-weather suspensions, but no crashes. WIRED asked experts to grade Tesla’s Austin autonomous taxi service—and, crucially, how to know if the system is safe. www.wired.com/story/this-i...

Matt Freese got a hand on 6 of the 7 PKs last night, stopping 3. What a show.

Tesla robotaxi appears to brake for a tree's shadow. This is where lidar would come in handy. Originally posted at youtu.be/Zn8YMkLaYL0?...

Piecing together Tesla's crash stats from multiple sources is my hobby. Given reasonable assumptions, I believe AP had ~2.3 B miles and ~420 crashes in both 2023 and 2022. Tesla has said 2.16 B FSD miles in 2024, but no FSD crash rate yet as Impact Report is a month late.

Tukey xkcd.com/3104/