Chris X Edwards

I love reading The Morning Paper where @adriancolyer translates computer science papers into English. With my keen interest in automotive software I took a special note of this excellent paper.

How They Did It: An Analysis of Emission Defeat Devices in Modern Automobiles

Only after deciding to write about this fascinating work did I notice that this was from my friend, the brilliant and witty Kirill Levchenko at sysnet.ucsd.edu.

I don’t have anything techincal to add to this fine work, but reading TMP’s take on it, I have to agree that the endgame here is one of two things. First, there could indeed be an AI race to constantly game the system. Currently there are some fairly crude, but entirely effective, heuristics to determine when the system is being watched (tested). But one could imagine a more sophisticated testing regime that required the wheels be turned randomly and the speed varied. An AI classifier could be conceived of which would still endeavor to cheat, and the race would be on.

The other option, and this seems like the just approach, is to not test the car occasionally (using extrapolation to find out how much it likely pollutes). No, the correct approach is to mandate full time pollution accounting. Your car should have a "measured polution" counter that just tallies it up and you get sent a bill accordingly.

The next topic, of course, would be cheating the pollution sensors. Don’t worry, there could be montoring sensors to ensure the correct functioning of the pollution sensors. And other turtles.

The content of this article is almost superfluous thanks to its helpful title.

Americans Would Pay Almost $5,000 More for a Self-Driving Car.

Specifically, it turns out, that the mean amount that Americans would spend for a fully autonomous car is $4,900 (there is a large variance). Here is the proper source.

In 2012 a new car was owned by the original owner an average of 5.95 years. This appears to be increasing so it is a conservative number to use.

According to this article the average American commute is 26 minutes each way. At 2 trips per day, 5 days a week, for 50 weeks a year, that comes out to around 216 hours per year or 1289 hours spent commuting during the average ownership of a new car.

This means, with the huge assumption that capitalism isn’t completely absurd, that Americans value their driving labor at about $3.80/hour. That’s not joy riding; that’s commuting!

According to KBB the new car average transaction price is now $34,372. The refund for used cars is around $18,600. This means that Americans who only use their cars to commute are spending $12.24 per hour of driving just to have the car. Even if you knock that in half by the assumption that the average driver spends half of their driving time doing other, presumably delightful, kinds of additional non-commuting driving, it is still a surprisingly large number given the fact that $3.80 is apparently too much for a chauffeur. What other wretched labor besides sitting in traffic are Americans willing to do for about half of minimum wage?

If Americans love compulsory driving so much, why do Uber drivers need to be paid around $12 per hour? Shouldn’t there be a huge pool of people who would be delighted to do the job for half that?

Have I miscalculated something? If so, then I obviously don’t know what to think about this. But even if the numbers are correct, I still don’t know what to think of this.

Long before Udacity showed up to help me plan a more effective course of action, I was doing what little I could to advance the state of the art in autonomous vehicles. The very first opportunity I heard of to do that was the Simulated Car Racing competitions. Although events are officially called that (SCR), I think of it as a misnomer; the events should really have been called SDRSC, Simulated Drivers Racing Simulated Cars. These competitions were designed for teams of AI researchers at university computer science departments but were technically open to anyone who could field an entry. That was all I needed.

For two consecutive events (2013 and 2015), I entered my SnakeOil car. I put an enormous amount of work into this project because I had no other way to participate in the future of autonomous vehicles. I learned a lot and it was quite exhilarating to see my car transformed from utterly stupid, bumbling around the track, to a serious smooth driving contender, racing for victory.

At the close of the final competition (no more events have been held) I was pretty tired from the effort and was relieved to collect my $100 (Australian) winnings and be done with it. I was expecting the 2015 organizers to produce a summary video of the competition as they had in years past, but sadly that never happened.

I had built this great AI system which was going to slip into my own personal obscurity like a really excellent sand castle on an deserted island. I knew that I wanted to preserve, for myself, some memento of all of the work I put into this. Recently, I had some motivation to dig out the project and get it all working again and record some videos.

I am delighted to now be able to remember and share this work through some exciting race videos of my SnakeOil cars in action. So for your entertainment and mine, I present the SnakeOil Exhibition Races.

Enjoy!

Imagine the most complicated situation you could conceivably understand. For each and every discernible component of that situation, imagine a new situation just as complex. Now multiply all of those situations together. Still with me? Of course you’re not! It really doesn’t take much to leave a human brain stultified with incomprehension.

I Contain Multitudes by Ed Yong is a rightly popular book about microbiology. It is that multiplication operation. If you’re doing something "easy" in biology, like seeing how many rabbits get eaten by foxes, it’s amazing how this quickly explodes into some horrifically complex mess. In biology the Lotka–Volterra non-linear differential equations are easy. Relatively speaking. But what if we’re not talking about furry cute things like foxes and rabbits but rather invisible microorganisms. These are the "multitudes" of the book’s title and the complexity just goes up and up. Truly, none of us is properly equipped to even appreciate just how complex the natural world is.

The book is a best effort. It is incidentally a brutal reproach of mankind’s greatest monument to hubris and ignorance: the medical doctor. If you’ve visited one in the last 70 years, you know they are nothing more than sanitized drug dealers. The book doesn’t say that. I’m editorializing. But the book does talk a bit about how dangerous the medical profession’s foolhardy obsession with antibiotics has been. Why has it been so wrong? Because it’s complicated!

Rather than summarize this book, I’m going to let an expert do it.

Just replace "Force" with "microbiome" and you’ve got this book’s message. A message which, while complex to the point of nebulous as mentioned, also seems pretty accurate. If you think you might be interested in learning about some of the marvels of modern microbiology I do highly recommend reading this book. It’s well written and the topic is fascinating.

For me personally, it was a little oppressive. The complexity that this book introduces to many readers isn’t really new to me. San Diego is a strangely backward place, but not when it comes to biomedical research. I have had a front row seat working for research labs for a dozen or so years now. Not only am I familiar with Aedes aegypti mosquitoes from hearing about them in weekly meetings, the lab I work in does another multiplication. It is a structural biology lab. This means that in addition to all the other complexity of biological systems, we’re interested in what every atom, literally, is doing. We’re a computational biophysics lab which is where I become slightly less useless. You could spend a million lifetimes just learning about the tools of the trade.

From my field observations of molecular biologists in the wild I have already absorbed the sense of overwhelming complexity. You see, it’s not just foxes vs. rabbits, or Wolbacia vs. wood lice. What is beyond the scope of this book is that there are many more levels of complexity. How do chemokines and their receptors interact? Don’t forget the near infinite variety of other molecules that complicate the relationship! That’s still not even probing the dark magic of atomic physics itself.

The natural world is overwhelmingly complex. Every person you’ve ever heard of who claimed to significantly understand it is a liar, an idiot, or both. I knew that going in. I even knew about the microbiome. I think the author was even a little reluctant to go as far down the microbial rabbit hole as I’ve gone. I have read Thomas Gold’s The Deep Hot Biosphere: The Myth of Fossil Fuels and I accepted its possible validity right away and have yet to see serious refutations. To me surface dwelling plants and animals are the earth’s extremophiles.

So what to make of it? How is this going to improve quality of life? The book mentions Florence Nightengale recommending, 150 years ago, something that is just now apparently being rediscovered to the horror of hospitals everywhere: for best convalescence, it might be good to open a window and get some fresh air. What a concept! My advice has been steady for a long time and it all seems coherent with the well-being of my multitudes. Live a wholesome life. Balance temperance with indulgence. Avoid filth but also avoid obsessing over it. Keep your mind active with wholesome pursuits. Eat only food. Avoid unnatural small molecule chemistry. Create a lifestyle that is physically active that you can maintain your entire life. Listen to your body. And, yes, get plenty of fresh air!

If you haven’t heard of the worldwide computer security poostorm last week, you can read about some of it in this article titled, The Google-phish-that-was-also-a-worm – what happened and what to do. This is its URL.

At my university some high up executive types sent around a campus wide email which included that link. What blew my mind, given the topic, was that they used a URL shortener. Specifically, they referred to this URL.

I used to be a big fan of URL shorteners. This was because I do everything in a text console and they really clean things up nicely. But after reading about phishing using shorteners and giving it a tiny bit of thought, I realized that setting people up with URLs that they don’t really know the target for was probably a bad idea. I now consider it mildly impolite. I have also now seen too many phishing emails to count that relied on shorteners to obfuscate the danger.

In the past I used goo.gl and even had a script to convert to a Google shortened URL automatically. Now I am going to atone for using shorteners in the past. Although I consider goo.gl rotten and insecure, it probably is the most secure of any of the URL shorteners. With goo.gl you can add a .info or + to the short URL and get a full analytics page without an actual redirection to the target link.

The plus apparently works for bit.ly too. (Note that bit.ly is a Libyan domain; what could go wrong?)

For tinyurl.com (whose domain is 5 characters less tiny than mine) you can give it a hostname of preview like this.

This article has some more tips for previewing different shortening services.

All of that might help you but I don’t tend to do things the normal way and I wanted a more potent way to solve this problem. For example, I don’t want cookies set or Javascript run or ads displayed when I’m trying to figure out what some hidden URL really is. Here’s what I came up with.

Put this in your .bashrc or otherwise make it a part of your shell’s environment and then you can do things like this.

It even works for Twitter’s insidious URLs which I have always avoided like the plague.

Note that these URLs aren’t even shorter. This is especially irritating in space constrained situations. Don’t use these services as thoughtlessly as Twitter forces you to.

LinkedIn does a smarter job of it by refusing to "shorten" http://xed.ch however, they do use a more baroque two step redirection.

As a final tip, once you know the URL you’re about to visit, it can often be smart to check with reputable sources about its reputation. Here are some I use.

I’ll leave it as an exercise for the reader to have these URL checkers check each other.