Noah Goodall

@noahgoodall

Transportation researcher. Pre-prints on personal site. Views are my own, not Commonwealth's.

71 Following1,172 Followers390 Posts

Bluesky

Is this your profile?Claim it and noahgoodall.dirty.social is yours.Claim profile

Noah Goodall@noahgoodall· 15d

Increasingly convinced that some prediction markets are triple cheeseburgers which exist only to let regulators feel better about allowing the double cheeseburger.

From "Dave's Way" by Dave Thomas founder of Wendy's, p. 118, can be borrowed here: https://archive.org/details/daveswaynewappro00thom

Noah Goodall@noahgoodall· 15d

It's full of weird little nuggets: you can't bet on someone playing a slot machine, but you CAN bet on someone playing poker because that has a skill element. They also have the broadest argument for "economic value" in history, to include single game NBA player point totals.

"In short, to take just one example, the price discovery value of event contracts on how many points a basketball player will score in a game depends on more than whether the event contracts can be used to hedge the purchase price of a ticket to the game. In some circumstances, the prices of those contracts could also, along with other information including the prices of many other event contracts, be factored into models used for commercial forecasting or audience-demand analysis, which are economic questions."

From: https://www.regulations.gov/document/CFTC-2026-1189-0001

Noah Goodall@noahgoodall· 19d

Explanation from Waymo.

"The study’s definition of “serious injury” for TLC crashes only includes injuries that required life-saving medical support. It then compares that TLC data to Waymo’s data, which uses a much more expansive definition of serious injury (‘KABCO’ which is used by NHTSA, NYS DOT and others, and includes things like broken bones) without attempting to control for this difference, which is how at arrived at this misleading conclusion."

Noah Goodall@noahgoodall· 19d

The safety analysis in the Waymo / taxi / Uber-Lyft comparison report has been removed. Appears to be related to the serious injury / death crashes, which was the only data that made Waymo look bad. static1.squarespace.com/static/5e713...

"As of July 23, 2026, we have temporarily removed our safety analysis
from this report, as well as relevant references to the analysis
throughout the report (denoted in-text by a reference to this section).
We were made aware of data discrepancies between City agency
data for New York City for-hire vehicle crashes with serious injuries
or deaths, as well as City agency data definitions related to such
crashes. We are working to resolve these discrepancies and ensure
our findings properly represent the data and the driving records of
New York City for-hire vehicle drivers and Waymo vehicles. Other
than this section, the remainder of our report remains unchanged."

Noah Goodall@noahgoodall· 19d

Strike my earlier comment about taxi/TNC drivers looking good. The injury crash rate is not good for them.

Noah Goodall@noahgoodall· 20d

Back to the Waymo vs. NYC taxi vs. Uber/Lyft (TNCs) report: this is the money chart. Waymo's injury crash rate looking very good when there is adequate data. The report's framing is way off and misleading in my opinion.

Figure 2 from here: https://static1.squarespace.com/static/5e71380706dc865d40a6f93c/t/6a5a8f5f6ff6a5473092ee00/1784319839886/How+Would+For-Hire+AVs+Impact+NYC_SSTRP_Open+Plans.pdf

Chart recreates data from Table 2.

Noah Goodall@noahgoodall· 20d

Fatal and serious injury crashes are rare -- Waymo has only seen 3 in 220 million miles. That's why we use the crash / injury crash / airbag surrogates.

Noah Goodall@noahgoodall· 20d

Glad to see TNC data compared to robotaxis. This is the swap, and the correct benchmark right now. TNC drivers look relatively good. But to be clear, the report found Waymo had lower injury crash rates, and NON-SIGNIFICANT differences in fatal / serious injuries. Authors (and media) are misleading.

Quote reads: " Initial data shows that Waymo vehicles are
involved in fewer injury-causing crashes overall,
but have a significantly higher rate of crashes
with a serious injury or fatality than the existing
New York City for-hire fleet. "

But the data shows non-significant differences, i.e. not statistically significant for fatal and serious injuries. There's just not enough data to make that determination yet.

Figure 1 from the report here: https://static1.squarespace.com/static/5e71380706dc865d40a6f93c/t/6a5a8f5f6ff6a5473092ee00/1784319839886/How+Would+For-Hire+AVs+Impact+NYC_SSTRP_Open+Plans.pdf?utm_source=substack&utm_medium=email

Massive error bars, these findings are not statistically significant despite authors' comments and media coverage.

Noah Goodall@noahgoodall· Jun 8

Safer cars are getting undermined by device usage. NHTSA collects this data by standing on the side of the road and watching drivers stopped at intersections. We really need to reboot a large-scale naturalistic driving study. Phones are VERY different since the the last one ended ten years!

https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/813788

Figure 5.

NHTSA has to estimate age from visual appearance, hence the massive bins.

Noah Goodall@noahgoodall· May 31

The original Freedom 250 lineup was basically New York's 95.5 WPLJ playlist in 1989-1991.

Noah Goodall@noahgoodall· May 26

This is the most adorable take. You know those hedge funds, always scared of alternative data like "scraping."

Noah Goodall@noahgoodall· May 26

Bob Dancer did professional video poker for years, and described it as a grind, closer to extreme couponing while playing the worlds most boring video game for 12-hour stretches. He lived in terror of a cold streak, despite having multiple other income streams.

Noah Goodall@noahgoodall· May 26

Media loves to portray the professional gambler as a savant, when it's more like living with your parents, flying to a Dallas Homewood Suites to knock on doors for a senate primary and STILL not betting until odds are 91%, all while carrying enormous risk. www.nytimes.com/2026/05/26/m...

Noah Goodall@noahgoodall· May 20

If you have to put this disclaimer in a prediction market, maybe don't have the market.

Damaging or disrupting animal enterprises is a federal crime under the The Animal Enterprise Terrorism Act (AETA) of 2006 (18 U.S.C. § 43). Penalties include fines in excess of $1,000,000 and life imprisonment. Polymarket will cooperate fully with law enforcement in any investigation of interference. Polymarket reserves the right to void any market if there is credible evidence of coordinated manipulation or external interference with the gorilla or its enclosure.

From: https://polymarket.com/event/will-little-joe-escape-again

Noah Goodall@noahgoodall· May 15

This is interesting only to me, but one of the Tesla crashes had a dog running into the road, and the Tesla positioned itself to the left side of the lane to try and avoid striking it. Which combines two examples I used to give for why risk management in an AV is so much harder than "just brake"!

Both are from https://osf.io/preprints/osf/bzkas_v1

Noah Goodall@noahgoodall· May 15

I'm curious how often they do remote driving like this, and also what's the per-mile crash rate? Most of what we know of remote ops is from Senator Markey's report, and Tesla's response skipped question 1(f) about latency, so I'd like to know that, too!

Noah Goodall@noahgoodall· May 15

This is interesting, mainly because Tesla had said they only take control of a vehicle (throttle/brake/steering) as a "last resort" and in both instances there was a Tesla employee in the car, albeit in the front passenger seat per Tesla's policy.

Noah Goodall@noahgoodall· May 15

Tesla reports ADS crashes to NHTSA via the Standing General Order. For the Austin robotaxis, they just today unredacted the crash narratives. Most surprising are two instances where the in-vehicle safety monitor asked for a virtual driver, who promptly crashed.

Noah Goodall@noahgoodall· May 15

Some of these have at least plausible deniability, but Boeing? How is in their business interests to promote road trips?

Noah Goodall@noahgoodall· May 15

Nothing will ever top politicians' meme coins, but as a bribe, Duffy's road trip is pretty blatant. slate.com/life/2026/05...

Noah Goodall@noahgoodall· May 15

Jokes on them: NHTSA takes 3 years to process complex requests.

Noah Goodall@noahgoodall· May 15

Reminder that CFTC couldn't catch a politician betting on his own market using his real name and bank account. They don't need AI, they need a spreadsheet with an IF/THEN statement.

Noah Goodall@noahgoodall· May 15

"most prestigious" ... "dimentia"

Those 2,000 requests are just the tip of the iceberg, Howell told ProPublica in an interview. Howell, the executive director of the Oversight Project, estimated that his group had submitted more than 50,000 information requests over the past two years. He described the project as “the most prestigious international investigative operation in the world.”

Among 744 requests that Aamot, Jankowski and Howell submitted to the Department of the Interior over the past year are 161 that seek civil servants’ emails and texts as well as Slack and Microsoft Teams messages that contained terms including “climate change”; “DEI,” or diversity, equity and inclusion; and “GOTV,” an acronym for get out the vote. Many of these FOIAs request the messages of individual employees by name.

NHTSA 2024 FOIA logs: "Jankowski,
Roman
The Daily Signal Records from David Winter, Director, Office of Highway Policy
Information for all sent communications [via the Custodian's
outbox; not received communications] mentioning 'Biden'
AND '25th' OR 'parkinsons' OR 'dimentia' from January 20,
2021 to the present."

Noah Goodall@noahgoodall· May 12

Their FOIA phone number goes to a busy signal, has for weeks. Prior to that, it went to a voicemail box that was always full.

Noah Goodall@noahgoodall· May 12

Reminder that federal agencies are not "overwhelmed" with FOIA requests. NHTSA gets 2 per work day. 2! Their budget is $1.6B. I've had one with NHTSA since January for a single document whose existence was all over the news. They stall and ghost and hope you forget. www.reuters.com/business/aut...

USDOT FOIA statistics by agency, available at https://www.foia.gov/data.html

Noah Goodall@noahgoodall· Apr 28

I co-authored Stop Predatory Gambling's response to CFTC on prediction market regulations. The full response is here: comments.cftc.gov/Handlers/Pdf... My favorite sections.

The Commission argues that prediction markets allow everyday Americans to hedge
legitimate financial risks. In narrow circumstances, that may be true. But roughly 90% of
prediction market volume flows through sports contracts. Whatever legitimate hedging use
exists at the margins does not justify a regulatory regime built around it. If the hedging
rationale is real, the Commission should be able to name who benefits and how.
When Commission Chairman Michael Selig was asked how a regular person might use
prediction markets for risk management, he said:
“This is a real opportunity for everyday Americans to be able to manage certain risk. I
do view this as something that democratizes access to our derivatives market. Say
somebody has a condition. They want to be able to manage the potential risk of having
a future treatment or medical costs. They can enter into a prediction market and place
a bet on the likelihood of something passing a drug trial.”16
The Chairman is arguing, on the record, that people should bet on drug trials as a means to
pay for treatment. This is shocking, dystopian, and inhumane. Worst of all, it is not even
feasible. Drugs that fail their trials become losing bets, leaving a sick person poorer at the
moment they can least afford it, unable to fund the next treatment. That this is the
Commission’s best argument for everyday hedging demonstrates it is far removed from the
reality of ordinary Americans.

The stock market comparison fails for a simple reason: the stock market is not zero-sum. [continuing through] It is a casino where the house has simply rebranded itself as
another player at the table.

Noah Goodall@noahgoodall· Apr 23

*former* federal workers. Like a steel plant closing, laying off 30% of the workforce has political implications.

https://www.nytimes.com/2026/04/22/us/elections/virginia-redistricting-voter-data-election.html

Noah Goodall@noahgoodall· Apr 23

"Bettor legally required to tell us his name and address, which matched the candidate. Then we Googled him to confirm." This is literally how all 5 of their cases went down. Kalshi is so lazy.

From Kalshi PR today: " After running its investigation, the surveillance team used internal information obtained from the trader and open source intelligence to confirm that the identity of the trader was the same as the candidate."

Noah Goodall@noahgoodall· Apr 23

Kalshi bragging about catching a candidate betting $50 on himself to win a primary. But they have know-your-customer rules! The algo is literally IF bettor_name = candidate_name THEN flag. They seem unable to catch staff, family, friends who are surely trading. "Over a dozen" not even 20, please.

https://news.kalshi.com/p/kalshi-political-insider-trading-enforcement-update

https://news.kalshi.com/p/kalshi-trading-violation-enforcement-cases

Noah Goodall@noahgoodall· Apr 15

This is a pretty comprehensive answer as to why ped deaths are increasing IMO. The smartphone / vehicle-screen issue is probably the easiest to address, and we REALLY need more data on this. The big NHTSA naturalistic driving study ended in 2016; phones have gotten much more distracting since then!