One of the bigger stories this week is definitely that of George Hotz’s Comma.ai announcing that it is shelving their flagship product after a letter from the US National Highway and Traffic Safety Administration.
The comma one is cancelled. comma.ai will be exploring other products and markets. Hello from Shenzhen, China. -GH 3/3
— comma ai (@comma_ai) October 28, 2016
You can read the full letter here. The interesting part – and the one most pertinent to this abrupt turn in the company’s strategy – starts on page 8, under the “REQUESTS” section, where some very reasonable requests for information are made, including a detailed description of the types of roads, geographies, speeds, and traffic and weather conditions under which the product, Comma One, can safely operate. Critically (and reasonably), the NHTSA has also asked Comma.ai for “a description of any testing or analysis to determine safe operating conditions for a vehicle equipped with comma one”.
In my humble opinion, there’s no way that Comma.ai, with less than 13 months in operation during which the product was very much still in the oven, has done this analysis. From all the sources I could find, there appears to be no one on the team with the capacity or experience to systematically plan and execute the kind of testing framework, nor do I believe the company has had the time to pull this kind of testing off, even assuming a single product iteration (read: no bugs).
Many people on my Twitter feed seem to believe that the company’s decision was triggered by the letter it received yesterday, on October 28th. But judging the company’s latest blog post, it is very clear from the title and the content that the company was bracing for impact. This is probably because, on the same day, NHTSA published new Federal guidelines for the testing and deploying autonomous vehicles. This is why the post tries to argue that Comma One doesn’t turn cars autonomous, but rather does a better job than existing features built into the cars.
Regardless, ten days don’t make much difference if the company hadn’t even considered the regulatory compliance aspect of their product during its inception and development, which, as outlandish as it seems, is what I think happened. The biggest proof is that just last month, Hotz stated that the product will be ready to ship by the end of the year.
So how could this be? If I had to guess, there were a few contributing factors:
- George Hotz is a hacker, and a damn good one, but lacks the wholistic view needed to successfully operate in a heavily regulated environment.
- Comma.ai’s investors (A16Z, who put in ~$3M at a $20M valuation) did not practice enough oversight or control to ensure that the company addresses the regulatory aspect of the operation.
- Many startups still have trouble with friend or foe recognition when it comes to the regulator. When it comes to public safety, a startup just can not ignore the regulator.