Research Fellow @joinFAI; opinions mine not theirs | Working to make US transportation faster, cheaper, more reliable, & less safe | prev: Econ & CS @UChicago

Inside the Beltway, USA
Joined January 2022
some important descriptive statistics which more people in transit should be aware of
5
13
2,905
... huh?
There is an (unpatchable?) supply-chain zero-day in PyTorch autograd that allows a regularization term to use the model's weights as RAM and SGD as its CPU clock to run any computation during training.
1
3
272
Not really very enthused about speed limits that you could reasonably violate on foot
We met up with cycling advocate and lawyer Steve Vaccaro in Central Park to hear about why he's helping with a legal battle to block NYC's proposed 15 mph speed limit along the Central Park Drives.
1
12
458
Lucas Pombo (aka Lib Development) retweeted
When I talk about Hoboken's land use being a disaster, this is what I have in mind. Yes, it's better than the suburbs, but this is not nearly dense enough that close to PATH. jerseydigs.com/425-newark-st…
10
3
43
4,946
It seems that the MTA is going to try to speedrun the funding process for their proposed 125th street subway; very exciting
The MTA is proposing a new 125 Street Subway, which would extend west from the future 125 St/Lexington Av Q station. The project would connect existing subway lines—including the Second Avenue Subway—and bus routes along the corridor. Learn more about this project at a public scoping meeting; we're hosting two virtual sessions on October 22, and one in-person session on October 27. Register for a session and get all the details: mta.info/project/125-street-…
4
36
1,080
Honestly this may be a situation where it's better to just not apply for federal funding and front the entire cost using state + local contributions
1
6
169
The headline of Tesla's release today is that they're probably about as safe as Waymo.
Replying to @Lib_Development
To close this off, here's the same data using the 4 Waymo categories, compared to Waymo and humans:
2
4
379
If we use Waymo's definitions of any-injury-crashes, Tesla is probably at two over ~3 million miles, which is almost exactly the same rate as Waymo. In both incidents, Tesla was clearly not at fault.
1
1
92
Assuming similar performance to Waymo, though, you'd expect it to still be zero. At this point, Robotaxi has not driven far enough to know the at-fault crash rate. But then again, that itself is enough to know it's clearly good enough.
1
96
To close this off, here's the same data using the 4 Waymo categories, compared to Waymo and humans:
1
462
The report says they went to the hospital "for an unknown reason", and there would be no reason to make this reason known to Tesla. But... 4 car accident, someone goes to the hospital, not an injury? That's, uh, surprising to say the least.
1
1
97
In the second crash, the description says "unknown" but this probably is because information didn't make it back to Tesla. The description is that the robotaxi was stationary when hit by two other vehicles in a collision of at least 4 cars. And someone went to the hospital.
1
1
101
Waymo: Any SGO reported crash with the field “Highest Injury Severity Alleged” is “Minor”, “Moderate”, or “Serious”, or “Fatality”). “Unknown” reported severity where the SGO narrative mentions injuries of unknown severity are also included. waymo.com/safety/impact/#met…
1
2
96
lmfao. I kind of can't help but laugh at this one. It's a reportable incident, and it caused an injury, but it isn't a crash. Under Waymo definitions (see below) it would count, but I kind of forgive Tesla for not counting it.
1
3
141
The first injury is described as follows: "Summary: The ADS was proceeding straight through a work zone on a street and drove over a pothole at approximately 29MPH. The ADS had two riders present in the vehicle. One customer alleged they sought medical care."
1
4
149
Oh, very clever. Tesla is using a really strange definition of "injured"; in practice, Tesla's SGO data lists two events with injuries (thread)
Tesla now has a robotaxi safety page to match Waymo: tesla.com/robotaxi/safety Of course, no Robotaxi has been in a serious crash yet, so it makes for an excellent looking dashboard. But Tesla has driven just over 1% of Waymo's 271 million driverless miles.
2
11
1,801
Tesla now has a robotaxi safety page to match Waymo: tesla.com/robotaxi/safety Of course, no Robotaxi has been in a serious crash yet, so it makes for an excellent looking dashboard. But Tesla has driven just over 1% of Waymo's 271 million driverless miles.
3
2
15
2,445
More big news for rail transit today. Congratulations to Miami, although there's definitely still a long way to go on infill station construction.
🚨Breaking news🚨 The Northeast Corridor is fully funded!
1
15
779
No wonder I feel like I have to get out of DC lmao; this definitely describes my building
1
8
645
The thing that has caused the most trouble for me is chairs. Robot vacuums are really not very compatible with many small obstructions in close proximity (ie, chair legs). We need robot-friendly chairs!
The whole home needs to be rethought from the ground up with robotics in mind. Why are we treating robots like toasters? They are friends, and can be so much more than a fancy single chore appliance.
2
5
407
Congratulations to Los Angeles for breaking ground on a new light rail line. That doesn’t happen anywhere in America very often
The capital of American car culture is now building more transit than any other city in America. Los Angeles just broke ground on a nearly $4 BILLION light-rail line through the San Fernando Valley: • 6.7 miles • 11 new stations • Connections to Amtrak, Metrolink and the G Line • Access to roughly 150,000 jobs across the region Rail service is returning to Van Nuys Boulevard for the first time in more than 70 years. Los Angeles isn’t abandoning the car overnight but it is building a future where driving is no longer the only option.
5
1
58
1,556