Comma.ai, the brash startup attempting to make a self-driving system entirely from a neural network has announced it will cancel the "comma one" add-on box it has planned to sell to owners of certain Honda vehicles. The box stuck on the rear-view mirror and used the car's own bus commands to provide an autopilot similar to those offered by car makers, with lane-keeping and adaptive cruise control.

Of particular importance is the letter from NHTSA to comma.ai which I suggest you read. This letter creates several big issues:

1. There are many elements of this letter which would also apply to Tesla and other automakers which have built supervised autopilot functions.
2. Of particular interest is the paragraph which says: "it is insufficient to assert, as you do, that the product does not remove any of the driver's responsibilities" and "there is a high likelihood that some drivers will use your product in a manner that exceeds its intended purpose." That must be very scary for Tesla.
3. I noted before that the new NHTSA regulations appear to forbid the use of "black box" neural network approaches to the car's path planning and decision making. I wondered if this made illegal the approach being done by Comma, NVIDIA and many other labs and players. This may suggest that.
4. We now have a taste of the new regulatory regime, and it seems that had it existed before, systems like Tesla's autopilot, Mercedes Traffic Jam Assist, and Cruise's original aftermarket autopilot would never have been able to get off the ground.
5. George Hotz of comma declares "Would much rather spend my life building amazing tech than dealing with regulators and lawyers. It isn't worth it. The comma one is cancelled. comma.ai will be exploring other products and markets. Hello from Shenzhen, China."

To be clear, comma is a tiny company taking a radical approach, so it is not a given that what NHTSA has applied to them would have been or will be unanswerable by the big guys. Because Tesla's autopilot is not a pure machine learning system, they can answer many of the questions in the NHTSA letter that comma can't. They can do much more extensive testing that a tiny startup can't. But even so a letter like this sends a huge chill through the industry.

It should also be noted that in Comma's photos the box replaced the rear-view mirror, and NHTSA had reason to ask about that.

George's declaration that he's in Shenzen gives us the first sign of the new regulatory regime pushing innovation away from the United States and California. I will presume the regulators will say, "We only want to scare away dangerous innovation" but the hard truth is that is a very difficult thing to judge. All innovation in this space is going to be a bit dangerous. It's all there trying to take the car -- the 2nd most dangerous legal consumer product -- and make it safer, but it starts from a place of danger. We are not going to get to safety without taking risks along the way.

I sometimes ask, "Why do we let 16 year olds drive?" They are clearly a major danger to themselves and others. Driver testing is grossly inadequate. They are not adults so they don't have the legal rights of adults. We let them drive because they are going to start out dangerous and then get better. It is the only practical way for them to get better, and we all went through it. Today's early companies are teenagers. They are going to take risks. But this is the fastest and only practical way to let them get better and save millions.

"...some drivers will use your product in a manner that exceeds its intended purpose"

This sentence, though in the cover letter and not the actual legal demand, looks at the question asked so much after the Tesla fatal crash. The question which caused Consumer Reports to ask Tesla to turn off the feature. The question which caused MobilEye, they say, to sever their relationship with Tesla.

The paradox of the autopilot is this: The better it gets, the more likely it is to make drivers over-depend on it. The more likely they will get complacent and look away from the road. And thus, the more likely you will see a horrible crash like the Tesla fatality. How do you deal with a system which adds more danger the better you make it? Customers don't want annoying countermeasures. This may be another reason that "Level 2," as I wrote yeterday is not really a meaningful thing.

NHTSA has put a line in the sand. It is no longer going to be enough to say that drivers are told to still pay attention.

Black box

Comma is not the only company trying to build a system with pure neural networks doing the actual steering decisions (known as "path planning".) NVIDIA's teams have been actively working on this, as have several others. They plan to make commentary to NHTSA about these element of the regulations, which should not be forbidding this approach until we know it to be dangerous.

I frequently say that there is no "internet of things." That's a marketing phrase for now. You can't go buy a "thing" and plug it into the "internet of things." IoT is still interesting because underneath the name is a real revolution from the way that computing, sensing and communications are getting cheaper, smaller and using less power. New communications protocols are also doing interesting things.

We learned a lesson on Friday though, about why using the word "internet" is its own mistake. The internet -- one of the world's greatest inventions -- was created as a network of networks where anything could talk to anything, and it was useful for this to happen. Later, for various reasons, we moved to putting most devices behind NATs and firewalls to diminish this vision, but the core idea remains.

Attackers on Friday made use of growing collection of low cost IoT devices with low security to mount a DDOS attack on DYN's domain name servers, shutting off name lookup for some big sites. While not the only source of the attack, a lot of attention has come to certain Chinese brands of IP based security cameras and baby monitors. To make them easy to use, they are designed with very poor security, and as a result they can be hijacked and put into botnets to do DDOS -- recruiting a million vulnerable computers to all overload some internet site or service at once.

Most applications for small embedded systems -- the old and less catchy name of the "internet of things" -- aren't at all in line with the internet concept. They have no need or desire to be able to talk to the whole world the way your phone, laptop or web server do. They only need to talk to other local devices, and sometimes to cloud servers from their vendor. We are going to see billions of these devices connected to our networks in the coming years, perhaps hundreds of billions. They are going to be designed by thousands of vendors. They are going to be cheap and not that well made. They are not going to be secure, and little we can do will change that. Even efforts to make punishments for vendors of insecure devices won't change that.

So here's an alternative; a long term plan for our routers and gateways to take the internet out of IoT.

Our routers should understand that two different classes of devices will connect to them. The regular devices, like phones and laptops, should connect to the internet as we expect today. There should also be a way to know that the connecting devices does not want regular internet access, and not to give it. One way to do that is for the devices to know about this, and to convey how much access they need when they first connect. One proposal for this is my friend Eliot Lear's MUD proposal. Unfortunately, we can't count on devices to do this. We must limit stupid devices and old devices too.

When people vote, what do they think it will accomplish? How does this affect how they vote, and how should it?

My apologies for more of this in a season when our social media are overwhelmed with politics, but in a lot of the postings I see about voting plans, I see different implicit views on just what the purpose of voting is. The main focus will be on the vote for US President.

The vast majority of people will vote in non-contested states. The logic is different in the "swing" states where all the campaign attention is.

The social networks have access (or more to the point can give their users access) to an unprecedented trove of information on political views and activities. Could this make a radical difference in affecting who actually shows up to vote, and thus decide the outcome of elections?

I've written before about how the biggest factor in US elections is the power of GOTV - Get Out the Vote. US Electoral turnout is so low -- about 60% in Presidential elections and 40% in off-year -- that the winner is determined by which side is able to convince more of their weak supporters to actually show up and vote. All those political ads you see are not going to make a Democrat vote Republican or vice versa, they are going to scare a weak supporter to actually show up. It's much cheaper, in terms of votes per dollar (or volunteer hour) to bring in these weak supporters than it is to swing a swing voter.

The US voter turnout numbers are among the worst in the wealthy world. Much of this is blamed on the fact the US, unlike most other countries, has voter registration; effectively 2 step voting. Voter registration was originally implemented in the USA as a form of vote suppression, and it's stuck with the country ever since. In almost all other countries, some agency is responsible for preparing a list of citizens and giving it to each polling place. There are people working to change that, but for now it's the reality. Registration is about 75%, Presidential voting about 60%. (Turnout of registered voters is around 80%)

Scary negative ads are one thing, but one of the most powerful GOTV forces is social pressure. Republicans used this well under Karl Rove, working to make social groups like churches create peer pressure to vote. But let's look at the sort of data sites like Facebook have or could have access to:

• They can calculate a reasonably accurate estimate of your political leaning with modern AI tools and access to your status updates (where people talk politics) and your friend network, along with the usual geographic and demographic data
• They can measure the strength of your political convictions through your updates
• They can bring in the voter registration databases (which are public in most states, with political use allowed on the data. Commercial use is forbidden in a portion of states but this would not be commercial.)
• In many cases, the voter registration data also reveals if you voted in prior elections
• Your status updates and geographical check-ins and postings will reveal voting activity. Some sites (like Google) that have mobile apps with location sensing can detect visits to polling places. Of course, for the social site to aggregate and use this data for its own purposes would be a gross violation of many important privacy principles. But social networks don't actually do (too many) things; instead they provide tools for their users to do things. As such, while Facebook should not attempt to detect and use political data about its users, it could give tools to its users that let them select subsets of their friends, based only on information that those friends overtly shared. On Facebook, you can enter the query, "My friends who like Donald Trump" and it will show you that list. They could also let you ask "My Friends who match me politically" if they wanted to provide that capability.

Now imagine more complex queries aimed specifically at GOTV, such as: "My friends who match me politically but are not scored as likely to vote" or "My friends who match me politically and are not registered to vote." Possibly adding "Sorted by the closeness of our connection" which is something they already score.