July 30th, 2005

rock shadow


One big benefit of Google are the technical talks - in this way the Google campus feels a lot like a college campus. Thursday I saw a talk by Joseph Jacobsen from MIT Media Lab on e-paper. I remember reading about e-paper in magazines 10 or more years ago, and finding the idea absolutely fascinating, but haven't heard much about it since then outside of Neal Stephenson's nanotech dystopia and other science fiction. But here are working prototypes - printed e-ink and even transistors - that can be cut, punctured, crumpled, and powered down and still show text or images, even color, up to 50 frames a second.

The talk was mostly focused on the technology and fabrication process, but most of the questions involved applications and social implications. Can you make solar cells? (Yes.) What about digital ink pens? (Yes.) What are the environmental impacts? (E-ink is made of titanium dioxide and carbon, which is what paper is made of anyway. The printed chips were first made of cadmium selenide, and now are made of pure silicon.) When can I buy something with this in it? (Soon a bunch of things will be coming out. In Tokyo, you can buy Dynavision ebooks now; they'll be in the US the beginning of next year.) How long does it take to make a stamp for mass-production? (Stamps are made with conventional lithography, which takes about 8 hours for a typical stamp.) What is the defect rate? (It's high, but you include lots of redundancy.) Have you thought about windshields and other clear displays? (Yes - transparency is part of the chemistry. The power consumption and failure states need to be worked out, though.)

I can imagine many applications for this, some positive (cheap, bendable computer displays and even computers, for one), some negative (pervasive animated advertisements, increased power consumption, etc.). What do you think?
rock shadow


Friday I went to another Google talk, this one by the authors of Freakonomics. While the room filled to overflowing in the half hour before the presentation started, I read through the introduction of the free copy I picked up. Even the intro is filled with assertions that are very interesting but leave me thinking, "really?," even when they're something about which I've reached a similar conclusion. After all, if all the other experts have missed the causal factors that the authors have uncovered, what factors are the authors missing? Many of the anecdotes that take up a few paragraphs in the introduction could be expanded into a book, and probably would have to be for me to feel like I have enough information to evaluate the veracity of their claims. But perhaps this is my fault for wanting academic rigor in a pop-science book.

Here's one anecdote. Preceding it, the authors describe how violent crime was climbing at a rapid rate up to the mid-nineties, and then it dropped off unexpectedly but universally. They summarize some of the most popular ideas of why this happened ("These stories were not only logical; they were also encouraging, for they attributed the crime drop to specific and recent human initiatives"), and then assert that they weren't true. Then they say the crime drop started 20 years earlier, and give a quick summary of how abortion became legal.

So how did Roe v. Wade help trigger, a generation later, the greatest crime drop in recorded history?

As far as crime is concerned, it turns out that not all children are born equal. Not even close. Decades of studies have shown that a child born into an adverse family environment is far more likely than other children to become a criminal. And the millions of women most likely to have an abortion in the wake of Roe v. Wade - poor, unmarried, and teenage mothers for whom illegal abortions had been too expensive or too hard to get - were often models of adversity. They were the very women whose children, if born, would have been much more likely than average to become criminals. But because of Roe v. Wade, these children weren't being born. This powerful cause would have a drastic, distant effect: years later, just as these unborn children would have entered their criminal primes, the rate of crime began to plummet.

It wasn't gun control or a strong economy or new police strategies that finally blunted the American crime wave. It was, among other factors, the reality that the pool of potential criminals had dramatically shrunk.

So I've heard these arguments before in different forms, and I certainly agree that planned families and wanted children - of which abortion is an unfortunate but sometimes necessary part - make a huge difference in what kids do when they grow up. But I have to take on faith the authors' assertion that this cause-and-effect really exists, and that the other factors didn't have effects. I'm sure they have more data on which they based their assertions, but I want to see it and see if I reach the same conclusions.

Another reason that I'm skeptical is that even in the introduction, the authors play psychological games with statistics:

In a typical election period that includes campaigns for the presidency, the Senate, and the House of Representatives, about $1 billion is spent per year - which sounds like a lot of money, unless you care to measure it against something seemingly less important than democratic elections. It is the same amount, for instance, that Americans spend every year on chewing gum.

The comparison between campaign spending and chewing gum is supposed to make me see that campaign spending isn't so bad after all, except that the former is what an individual spends and the latter is what a large aggregate spends, and comparing them is misleading. And the use of chewing gum as the example just tries to trivialize it more.

Anyway, in the talk, the two authors introduced themselves and how they (an economist and a writer) came to work together, and then talked about a couple of anecdotes that weren't in the book. One was about a Yale economist, Keith Chen, teaching Capuchin monkeys how to use money. Over 6 months, they taught the monkeys to first associate and then trade washers for high-sugar foods that they loved, and found that they reacted rationally to price fluctuations (e.g. they bought less of their favorite food if it cost more) and irrationally to a game where they could lose their money, "just like the average day trader." They also related that researchers observed monkeys steal money from them, and claimed that monkey prostitution was observed as well. (Another of those "really?" moments there.)

In response to one question, the authors argued that the usual metrics by which schools are measuerd - class sizes, test scores, the number of students to go on to college or get good jobs, etc. - don't matter, and that the important parts are the quality of teachers and their peers. They quoted studies done on the Chicago public school lottery system, where students who won the lottery to get into a good school didn't perform better - indeed, they did slightly worse - than similar students who didn't win the lottery and had to go to a "worse" school. However, the students at the worse schools did get into trouble more.