Antonio E. Porreca - Do waterfalls play chess? and other stories

From the better late than never files, here’s something I should have blogged about back when it came out.

Scott Aaronson wrote paper called Why philosophers should care about computational complexity and posted about it on his blog.

To make up for the delay, I’m offering a bonus link to Antonia Perreca’s commentary on the same.

As for me, I’m still waiting to see some make a good definition of “computation.”

Directed Edge - Google Spam Heresy: The AdSense Paradox

Being something of a search weenie, as my eyelids were feeling heavy today I found myself mulling over the problem, “How would one detect Google spamming?”

The answer turns out to be surprisingly easy. Who has an incentive to spam Google? People living from advertising. Who owns the largest online display ad network? Google.

…

So, here’s the heresy: the spamminess of a web site is inversely proportional to its ad click-through.

Think about it — in a typical internet search, a navigation path terminating at that page is the best result. If they click on an ad, it probably means you missed serving up the right page in the first place. As a corollary, the pages best optimized to pull you in via a search term and send you back out via a related ad are among the worst results.

I’ve said for years that Google exists on the razor’s edge of a paradox: if their results are good, then web surfers will ignore the ads on the side and just click on the top search result. If their search results are good, then the top companies in each field will already be the top search results and won’t need to advertise. That means that the ones who do advertise will be the crappy companies in the field, so surfers will be further trained to ignore the ads. The result is that Google’s natural settling point is with results that are crappy enough to get people to look at the ads but not so crappy that users switch to Bing or something.

I don’t see how Google gets around this paradox. I kind of wonder if search isn’t one of those areas of natural monopoly that ought to be tightly regulated or dispersed into open source or something. Mickey Kaus recently predicted that Google will get itself sucked into some political controversy sooner or later, and I tend to agree that it’s inevitable. What result should you get when you search for “health care bill cost”? “Obama birth place”? That the answers to these queries are generated algorithmically doesn’t mean they’re neutral! There’s an old hacker koan:

In the days when the Sussman was a novice, Minsky once came to him as he sat hacking at the PDP-6.

“What are you doing?”, asked Minsky.

“I am training a randomly wired neural net to play Tic-tac-toe”, Sussman replied.

“Why is the net wired randomly?”, asked Minsky.

“I do not want it to have any preconceptions of how to play”, Sussman said.

Minsky then shut his eyes.

“Why do you close your eyes?” Sussman asked his teacher.

“So that the room will be empty.”

At that moment, Sussman was enlightened.

The algorithm always has a bias.

Kevin Carey - Decoding the Value of Computer Science

Here’s a Chronicle of Higher Education story illustrating two of my obsession points:

There, at the Indiana state-budget office, my role was to calculate how proposals for setting local school property-tax rates and distributing funds would play out in the state’s 293 school districts. I did this by teaching myself the statistical program SAS. The syntax itself was easy, since the underlying logic wasn’t far from Pascal. But the only way to simulate the state’s byzantine school-financing law was to understand every inch of it, every historical curiosity and long-embedded political compromise, to the last dollar and cent. To write code about a thing, you have to know the thing itself, absolutely.

Over time I became mildly obsessed with the care and feeding of my SAS code. I wrote and rewrote it with the aim of creating the simplest and most elegant procedure possible, one that would do its job with a minimum of space and time. It wasn’t that someone was bothering me about computer resources. It just seemed like the right thing to do.

Eventually I reached the limit of how clean my code could be. Yet I was unsatisfied. Parts still seemed vestigial and wrong. I realized the problem wasn’t my modest programming skills but the law itself, which had grown incrementally over the decades, each budget bringing new tweaks and procedures that were layered atop the last.

A good Computer Scientist, like anyone else who is good at something, is motivated not just by a desire to get things done, but a desire to get it done the right way. Of course, just because this desire exists, that doesn’t mean it can be realized. Because of this, moral suasion is the best way to lead the good. Those who are bad will eventually just copy the good, because they are too foolish to figure out how to do something creative.

So I sat down, mostly as an intellectual exercise, to rewrite the formula from first principles. The result yielded a satisfyingly direct SAS procedure. Almost as an afterthought, I showed it to a friend who worked for the state legislature. To my surprise, she reacted with enthusiasm, and a few months later the new financing formula became law. Good public policy and good code, it turned out, go hand in hand. The law has reaccumulated some extraneous procedures in the decade since. But my basic ideas are still there, directing billions of dollars to schoolchildren using language recorded in, as state laws are officially named, the Indiana Code.

The other point this illustrates is that systems will evolve in the direction of the tools with which they are maintained. The Indiana Code was created to bring the law into line with the ideal program, even though there already existed a less-than-ideal program in line with law. Because this effect exists so powerfully, we must be careful about how we arrange our systems.

Via llimllib