Jim McPherson Profile picture
Perspectives on law & policy related to new mobility technology, including autonomy, micromobility & web3. Californian.
Jan 22, 2019 5 tweets 1 min read
There's a sense in which we have left AV 1.0: the 2013-2018 experiment marked by Elaine Herzberg's death, 60+ companies getting CA AV testing permits, AV Coalition for Safer Streets failing to secure nat'l AV legislation, disappointed press who expected L4 by Dec. 1/ 2019 starts AV 2.0 w/ a push to get out of trough of disillusionment, new industry consolidation & partnerships, pullback of some investors & re-commitment by others, PAVE advocacy/education group approaching safety promise from new direction w/o using the 94% statistic, ... 2/
Dec 1, 2018 4 tweets 1 min read
“The only human that the AV is absolutely sure exists in the one inside the car. Everything else is a confidence figure based on an imperfect feature matching classifier (or some other variant of computer vision algorithm).”

From a new voice on AVTwitter. As I look at AV confidence scores (for pedestrians, cars, etc), I’m struck by the fact that I, as a human, have only one stereo camera—my eyes—yet I have *100% confidence in everything I see, and a contextual understanding of what it is, what it’s doing, & what it might do next.
Nov 27, 2018 21 tweets 5 min read
Some people may have missed this story. It is the single-most important story on AV safety I have seen in more than two years of following the topic closely. It was originally published behind BI's pay wall, but it apparently viewable by anyone now. 1/ "Much has been written about the death of Herzberg, most of it focused on the failings of the driver. But until now not much has been revealed about why engineers and senior leaders turned off the car's ability to stop itself."
Sep 16, 2018 6 tweets 2 min read
Could not disagree with this more; the Trayvon Martin hypothetical, or the conclusion of the piece. "By contrast, a robot that would not be racist, fearful, or hot-tempered would be appealing." Joh doesn't think algorithms can be biased, or use racially-discriminatory data sets? Autonomous weapons systems are abhorrent, unethical, against national & international law. No question, no debate. You leave a kill decision up to software program and you have taken all agency out of it. You don't analyze algorithm for reasonableness. It's a spring gun.
May 29, 2018 33 tweets 6 min read
NTSB's preliminary report on Uber-Tempe crash suggests the self-driving system was effectively impaired--just like a DUI. Someone at Uber could go to jail or prison for vehicular homicide. 1/ ntsb.gov/investigations… Vehicular homicide involves the death of a person other than the driver as a result of either criminally negligent or murderous operation of a motor vehicle. It focuses on what the defendant should have known & the risks associated with what defendant did know. 2/
Apr 21, 2018 13 tweets 4 min read
Everything old is new again. Amelia Earhart
Feb 6, 2018 35 tweets 15 min read
A lighthearted title about a very real dispute over urban geometry — & a larger dispute between technologists & urbanists, generally, about the future of shared, autonomous, electric mobility. (Alt. title: Why e-bikes, & not AVs, are the real transportation disruption.) 1/ How disruptive will they be? Answer: Not as much as tech thought leaders claim.

Transportation disruption, historically, comes from changes in the source of motive power (& safety advances, & changes in law).
Dec 4, 2017 27 tweets 6 min read
I read this important RAND Corp report in detail and have some observations. (Kalra, Groves: The Enemy of Good: Estimating the Cost of Waiting for Nearly Perfect Automated Vehicles. Rand 2017) rand.org/pubs/research_… 1/ Some observations are related to its assumptions, but most are related to the way it has been mis-cited—which the authors themselves caution against. 2/