Thursday, April 28, 2011

Spam warfare

Instead of just filtering spam into a junk mail folder, why not bounce it back to the sender with a message that it has been flagged as spam?

If it's spam, this is at worst neutral and at best a minor hassle for the spammer. More importantly though, if it's not spam, the sender is informed that the message didn't get there. Like many people, I never go through my spam folder the alternative is for no one to know that the email was lost in transmission.

Why doesn't this exist? (or does it?) All the odd technical issues I can imagine (spam pong, anyone?) seem easily fixable, but what do I know about these things? Am I missing something?

Tuesday, April 26, 2011

Silent moo

Tyler Cowen links to a new book on the economics of animal welfare.

To me, the most obvious and fundamental tradeoff has always been the question of how many cows should live versus how good their lives will be. You will have to take a stand before you tell people what sort of meat they should (or should not) eat.

We are allowed to care how extant animals are treated. But of course we're also allowed to care about future animals that may or may not exist. The problem is, if we decide to treat living animals better, it's going to cost more, which will raise the price of meat, which all else equal will reduce the amount of meat people buy, which means fewer cows will ultimately be raised. Great for the cows that do exist, not so great for the ones that never get to exist as a result. (Of course, it's pretty bad if their lives are actually worse than death).

For the most part, I see people who are outraged about animal treatment basically failing to notice this important tradeoff. Silent moo? That which is heard, and that which is not heard? That which is herd, and that which is not herd? The language gods are smiling on me, so I must be right, right? Indeed, silent moo is not an anagram of economics, which means that in this case, economics is not all mixed up.

Friday, April 22, 2011

Charitable arguing

I was planning to write this post, but Greg Mankiw did it for me.

Here is my tangential advice. Arguments are usually not fully specified and you have your choice on how to interpret them. Taking a moment to hunt for an interpretation that makes an argument good -- before you denounce it as a bad argument -- is a nice heuristic that forestalls the tempting leap from "There exists an interpretation that makes this a bad argument, but it may not be what he had in mind," to "This is a bad argument!" And as someone who has spent a significant fraction of his life engaged in serious (written) debates on all manner of topics, I have to say: I used to do it the first way, now I do it the second way, and there is no question about which attitude is superior. If you look for good interpretations, you end up spending most of the time asking questions about what people really meant by things. Besides clarifying the meaning though -- and almost more importantly -- it also gives them a chance to refine or update their claims without the strong implication of having been wrong. People happen to reach agreement much more quickly when you make their true initial position easy to abandon by interpreting it much more loosely than they actually had in mind. Which is sort of obvious, in retrospect.

This means being as charitable as possible to the arguments of others, and asking lots of questions. It pays off.

Thursday, April 21, 2011


Here's some nice cross-pollination, especially if you're interested in political economy:

Economics tries to describe (a) what an optimizing agent looks like, and (b) what happens when we add up a bunch of optimizing agents. Of course we're not the only ones interested in some facet of this question, and it's always nice to encounter different perspectives.

As a side note, where Lemire points out that "computer" originally referred to humans, I am reminded that "manufacture" comes from the Latin for "make" and "by hand." How times have changed.

Wednesday, April 20, 2011

E.O. Wilson's War

I'm a fan of E. O. Wilson, and I find myself -- quite by accident -- in an evolutionary game theory class right now. So this article caught my eye.

What Wilson is trying to do, late in his influential career, is nothing less than overturn a central plank of established evolutionary theory: the origins of altruism. His position is provoking ferocious criticism from other scientists. Last month, the leading scientific journal Nature published five strongly worded letters saying, more or less, that Wilson has misunderstood the theory of evolution and generally doesn’t know what he’s talking about. One of these carried the signatures of an eye-popping 137 scientists, including two of Wilson’s colleagues at Harvard.
For those of you with access to Nature, the actual paper is here.

They are basically arguing about the extent to which altruistic behavior is driven by individual selection versus group selection. The mainstream stance (as I understand it) is that individual selection is the primary driver of most everything, while Wilson is now saying it doesn't seem to add up. Which has a lot of people angry:
“It’s almost universally regarded as a disgrace that Nature published it,” Dawkins said. “Most people feel the reason they published it was the eminence of Wilson and Nowak, not the quality of the paper.”
As an outsider, I can't speak to the quality of the paper, and I'm not here to take sides on the actual issue. But I do want to take a stand on the meta-issue of whether it's good or bad that Nature published this paper. The question is whether this paper helps or hurts the rate of progress of science (in expectation).

First of all, let's be clear that even if our attempts at science have us converging asymptotically to true understanding of the universe, that doesn't say we're getting there nearly as fast as we could be. The current approach has a lot going for it, but it's hardly optimal and sacred and immune to criticism; to the contrary it seems pretty biased towards the status quo and hence slowness. The basic problem is that entrenched academics are both quite attached to the status quo (since they built it) and the ones who set the rewards for research on the various topics a researcher could choose to pursue. Academic rewards are heavily influenced by what sorts of topics are currently fashionable/approved of by mainstream academics. In the long run we're all dead, but the long run is a long time from now, which is to say: science marches on, but slowly.

By contrast, what Wilson has demonstrated is a willingness to change his mind about an issue he has already spent many decades thinking about. Pretty darn rare, wouldn't you say? His many detractors are basically saying he has changed his mind too easily, but note that even if this is true in isolation, it could easily be the optimal response to a field that is too resistant to change on average. In theory it could be optimal to release this paper even if Wilson himself suspected it was wrong.

Here, let me wave my hands a little. We could think of science as a machine that (a) generates a lot of theories, (b) evaluates those theories, and (c) ultimately keeps the good ones and discards the bad ones. And when we get to keep the good and throw away the bad, we want high variance. So I find it somewhat refreshing when entrenched academics shake things up, because they usually have the most power but least willingness to do so.

On the other hand, here's another sort of argument, in favor of the current scientific consensus.


What happens now? People will talk about this. Wilson will be right, or he will be wrong. If he's right, science jumps forward. But even if he's wrong, science comes away with more confidence in its conclusions than it had before, by virtue of successfully countering a new sort of negative argument (which can be a lot more productive than continuing to find new positive arguments). Whatever the case, you have to admire his willingness to be the target of so much outrage when he could just rest on his (substantial) laurels.
“All new ideas go through three phases,” Wilson said, with some happy mischief in his voice. “They’re first ridiculed or ignored. Then they meet outrage. Then they are said to have been obvious all along.”
Of course, that's not quite true. Most new ideas are first ridiculed or ignored, and then forgotten completely. After all, most new ideas are bad! But if all ideas are first met with ridicule, then what can we outsiders infer about the quality of this idea from the fact that it's ridiculed?

Q-tip: MR

Saturday, April 16, 2011

Blast from the Past

This is incredible. My building is getting its windows fixed up, and this morning the repairman pulled out a piece of newspaper that had been stuffed as filler into a gap underneath the window frame. An artifact from the past, sealed in there for who knows how many decades...

Wait, scratch that. Actually we know exactly how long, because it's a newspaper:

The Chicago Sun-Times, Sunday, July 23, 1950.

In fact, pretty much all the questions I would think to ask upon discovering a relic -- where and when did it come from, what sort of environment would have produced it and how did it come to be here -- are answered in this case. For one, there may be no better object than a newspaper for giving a sense of the time it hails from. But even beyond that, it almost feels like this particular page of newsprint was carefully selected for the journey from 1950 to 2011. On the one hand, it's just one page, and not even the front page at that:

And indeed, at first glance I thought it was just a random and unimportant page from the middle of the paper. But on closer inspection this page seems to contain a representative smattering of news items that would normally only be captured over many pages. It's also full of advertisements that give a sense for the trends and technology and prices of the time.

For starters there's the Headline Quiz: Test your knowledge of the news of last week! (complete with Answer Key).

Always good to provide a convenient boxed summary of what's been happening lately for anyone who's just tuning in from, say, The Future.

By the way it's possible the average score on this quiz is not monotonic in the year people take it. I'm not sure if this one was a softball or if it's just easy from the perspective of 2011:

Added: Note the answer to #5 is General Eisenhower. In 1950 he was not yet president.

But not all news can be dressed up as fun and games. Right next to the quiz is a sobering reminder of the era:

That's some heavy stuff. It sure is good to be on this side of the Cold War.

But to get away from that, we need only flip the page over. Advertisements! What were they buying in 1950?

A portable phonograph for only $29.95?!? Carry it like a lightweight overnight bag! Music anytime, anywhere with this handsome Silvertone portable!

How far we've come. Our definition of "portable" has been refined considerably. And size aside, note also that an iPhonograph would not be "handsome" would be iPretty. (Not sure I consider this an advancement).

By the way, if you're wondering how much $30 was in 1950, take a clue from the fact that the buyer could opt to pay in less burdensome installments of $5/month.

Besides phonographs, there's also photographs. TREMENDOUS SAVINGS!

Mail and phone orders promptly filled! Based on this and a couple other ads, it seems to be pretty normal for certain businesses to stay open for one or two days a week.

Finally, back to the fun and games. I wonder if anyone ever found this treasure?

I think we can safely infer that in 1950, pirates were "in"...


I know you can just go to XYZ Archives and pull up records of old newspapers and all. But this was just pulled out of my wall.

Thursday, April 14, 2011

Smoking or nonsmoking?

I have found myself in (okay, perhaps instigated) a lot of conversations about information lately. And there's one tension that keeps coming up, and which never seems to be entirely clear, even though when I sit down and think about it, I feel like I ought to be able to make it clear without too many words. This is an attempt at that.

Information is a double-edged sword. In one sense it is necessarily good for you, but in another sense it need not be. Which sense is right, depends specifically on the question you're trying to answer. If you're trying to find the optimal solution to a problem, then becoming more informed can only help you find it. But finding it not the same thing as attaining it. In some contexts it's important to recognize that information itself is an end goal -- a consumable thing that people may or may not want -- and it is the unfortunate nature of the beast that sometimes you cannot know what information would be good for you unless you know the information in the first place. Which is to say, sometimes finding the optimal solution comes at the expense of attaining it; what if in the process of finding the optimal solution, we get stuck with a bunch of information we don't want? Information doesn't satisfy free disposal!

Because finding and attaining are both worthwhile goals which I go back and forth pretty freely between, it's not hard to see why confusion might arise. I will quite happily agree with you that the pursuit of Truth need not be synonymous with trying to attain your goals in life...but at the same time I will uncompromisingly assert that Truth is uniquely best suited to the task of finding optimal solutions to any problem, including even the question: "What is the best way to attain your goals in life?"

I mean, suppose you are wondering whether cigarettes are right for you. The answer certainly isn't obvious; there are costs and benefits to smoking, both of which depend quite a lot on how your body will react, which unfortunately you won't know unless you actually start smoking. But if you happen to have a sympathetic clone who starts smoking, then your clone can simply report back to you whether it was a good or bad idea in retrospect. Indeed, you may end up in a situation where you don't smoke but your clone does; in the end you both just wanted to know the optimal action, and it's unfortunate that to find out, your clone also had to irrevocably learn the taste of nicotine. Your clone's body now knows something yours doesn't, and it happens that this additional knowledge prevents the clone from attaining the optimum to the original problem. (Even if the clone manages to quit smoking, (s)he will forever carry around a memory that you don't have, making cigarettes annoyingly tempting when encountered here and there, instead of neutral).

More generally, if someone knew everything, and they cared about you, and you knew (somehow) that you could trust them, then they could tell you exactly what was best for you to hear. You would get all the benefits of them finding the optimal solution for you, without being burdened by the information that was necessary in order to find it, information that might prevent you from attaining the goal.

In any case, in the absence of either an experimenting clone or an all-knowing benevolent and trustworthy soul, we must typically ask the much harder question of what information we really should seek out, in order to get as close as we can to attaining our goals, given the extremely limited information we begin with.

And so I say: If we are in a serious discussion about what is optimal, I will invoke truth and logic and rationalism to find the answer, and in this matter you cannot win by departing from that path. But that is not at all the same thing as saying that truth and logic and rationalism are what is optimal.

Is that clear?

Wednesday, April 6, 2011

Foolish optimism or optimal foolishness?

Publicly stating what you're planning to do is a good way of pressuring yourself to do that. It raises the cost of not doing so, and thus acts as a partial commitment device.

I don't feel like sacrificing myself for science today, but I can definitely think of examples where I wanted myself to do X, so I said I would very publicly. And in some cases I failed to do what I said I would do. And then people thought I was deluded or naive about my own preferences.

For example, you may have seen a study that goes something like this: Students have a paper due. They are asked when they are 50%, 75%, 95%, 100% sure they will have the first draft done by. And of course, some people end up overshooting even the dates they were "100% sure" they wouldn't overshoot. Fools. We conclude that people are terribly naive about their procrastination.

No. Because what if the act of saying you're sure has an effect on when you actually finish? If such an attitude actually helps to motivate you to finish earlier (in expectation), then it can be quite sensible to say you're sure. The alternative, after all, is defeatism...and what's the use of that?

Indeed, it wouldn't be difficult to construct a model where the only people who publicly announce their intentions are the ones who might actually renege on them. Or to take it further, let's say that people can announce their intentions with different levels of conviction. Set it up right, and we can get a scenario where people who announce their intentions with more conviction are -- even after doing so -- less likely to make good on the promise!

(For this, imagine that people want themselves to do X, but (when the time comes) there is some underlying likelihood of not doing X, which an individual can mitigate by announcing with conviction that she will, in fact, do X. We we would just need to rig it up so that, as that underlying likelihood of [not X] grows, it is optimal for people to increase their conviction but not by so much that the probability of [not X] actually falls or stays the same.)

In this model, when you terribly "mispredict" your behavior, it doesn't mean you're terribly naive; to the contrary, it means you know you have a serious problem! We would actually expect "hypocrites" to show up quite often in this world.

Hey, maybe Eliot Spitzer was such a fierce opponent of prostitution in order to increase the expected embarrassment that would befall him, should he succumb to certain temptations...

In Therefore judge not, I said that when people's pronouncements impact the behavior of others, we shouldn't assume these pronouncements actually reflect their beliefs. Here I'm saying that the same is true when people's pronouncements impact their own behavior, too.

Sunday, April 3, 2011

Surely it's melted by now

I like google as much as the next person, but they have clearly gone overboard with their custom logos ("doodles"). Today's offering? It is apparently the 119th Anniversary of the First Documented Ice Cream Sundae.



I liked it better when the doodles were more occasional, and therefore more remarkable, more worthy of notice. Today the value has been diluted to almost-nothingness. Ice cream sundaes? 119th anniversary? Really?

When are the exam?

Of course it's impossible to design an exam that encourages exactly the right type of learning. One big problem is that we want people to understand things over a wide interval of time, but when we tell them they will be measured in time t, they can get away with just understanding in time t. But even if we can only give 1 exam, we can make progress by randomizing when it actually occurs, effectively spreading it over the interval.

[Note: This post is not about the unexpected hanging paradox, although you may want to read about it!]

The problem is that cramming for an exam does not seem to promote retention. Knowledge gained through cramming follows a sharp peak that drops off steeply on both sides. If that's bad, then why don't we see more exams that are given at some random time? Instead of "the exam will be given on day t," why not "the exam will be given randomly on one of these 3 dates"? If retention really is better accomplished by not cramming, then in fact spreading the possibility of testing over a few weeks may actually make it worth learning the material more deeply, rather than just cramming repeatedly.

There are surely logistical issues to consider. But if you have a problem with random exams in theory, why? In considering this, keep in mind that the twin goals of testing are to incentivize good learning and to measure that learning. Here are the important questions: Are we motivating better learning? Are we moving toward a less biased estimator of learning? Are we moving toward a less noisy estimator of learning? (If it comes at the cost of significantly higher variance, it may not be worth it).

Why are random ("pop") quizzes rare rather than standard, and why do they so frequently only cover yesterday's material?