Tuesday, August 31, 2010

When Quality Doesn't Matter

Paul Graham, a successful programmer who now funds start-up tech firms, writes about his experience with Yahoo! in the late 1990's. He had a start-up with a search algorithm that ranked search results by user behavior, how many clicks actually created purchases as opposed to mere clicks. It was like the algorithm Google uses to sort ads today, but this was in back in 1998. It seems like a rather obvious, straightforward improvement than just treating all clicks the same.

He notes an impression from a meeting where he pitched his product to Jerry Yang:
Jerry didn't seem to care. I was confused. I was showing him technology that extracted the maximum value from search traffic, and he didn't care? I couldn't tell whether I was explaining it badly, or he was just very poker faced.

I didn't realize the answer till later, after I went to work at Yahoo. It was neither of my guesses. The reason Yahoo didn't care about a technique that extracted the full value of traffic was that advertisers were already overpaying for it. If they merely extracted the actual value, they'd have made less.

When an industry is rife with business where quality does not matter, it becomes ripe for mimics who come in and sow the seeds of a crash and possible recession. There are many stories about real-estate brokers setting up shop in the early aughts, not caring about whether homebuyers would actually pay their mortgage because it did not matter. This was a signal that rot was rampant. Basically, if quality doesn't matter, and there's free entry, there's a bubble.

When people have positions that don't do what they say they do, and make a lot of money, there are myriad bad effects. Once when I was a risk manager, I remember showing a swaps book trader a more efficient way for him to hedge his portfolio. As I had to calculate his value-at-risk I had all the data to demonstrate conclusively my superior algorithm. He found this annoying. As a market maker, his Sharpe was already well above 10, so decreasing his value-at-risk by 20% did not really matter. Like Graham's encounter, I discovered it was all marketing.

The problem with this situation is that when you really understand the game, you have to never talk about it, which is easiest to do if you really don't understand it. So, the best brokers or brokers-who-call-themselves-traders are blithely ignorant, because they don't generate 'tells' that make everyone engaging in the game uncomfortable. When they talk about trade ideas that are totally unfounded, they can't be convincing if aware of its lack of statistical evidence, or how their qualifications make everything said meaningless (this could lead to a retracement). Once you swallow the red pill, you can't go back to enjoying the Matrix.

Similarly in the corporate borg, especially in places like the new Office of Minority and Women Inclusion that is now mandated to be part of each of our 30(!) financial regulatory bodies. As true discrimination is about as rare as a Klan rally, this is all just a sop to the Indian-like ethnic group spoils system the US is becoming (are there really any bankers who hate minorities enough to forgo extra profits?). So, the Chief Diversity officer's real role is not to rid financial discrimination, but rather to spout cliches about diversity, and put a pretext on the patronage daisy-chain that led to the 2008 housing crisis. However, if you really understood this, you would go crazy, so earnest dolts plague the aristocracy because the dupes actually believe their job is about what it says it's about.

Graham notes that Yahoo! was doomed by their dominant marketing, as opposed to programmer culture. I also think that when you get paid for something other than what you say you do, it doesn't give the same satisfaction. People who are overpaid may be rich, and such richness does buy you some friends, and I'm sure many think hey, I could live with that, no reason to be a prig about principle. I think that's a bit like thinking that if you had an infinite supply of sex, beer, or ice cream that would be awesome: only for a little while. Arthur Brook's perceived earned success I think highlights the importance of honest exchanges.


Anonymous said...

"As true discrimination is about as rare as a Klan rally, this is all just a sop to the Indian-like ethnic group spoils system the US is becoming (are there really any bankers who hate minorities enough to forgo extra profits?)."

It's funny that the people who make this argument are the same people who think that racial profiling works.

Racial/religious/etc. profiling works by assuming (usually correctly)that you can sort a group based on a single variable that is highly correlated with the variable that you are interested in, instead of going through a costly process of sorting by a large number of variables to get a perfect correlation.

If you can capture 80% of the variation in loan loses by race then why would you go through the costly process of collecting information on income, wealth, etc. and not reject loans a based on race alone.

Your argument overlooks the fact that profit optimization can lead to systematic discrimination without requiring bankers to hate minorities.

Eric Falkenstein said...

Banks don't use race in their underwriting, if they did, they would be sued. This has been true for decades (I don't know pre-1980s). They use things like income statements and balance sheets. These have disparate impact, to be sure. Thus, even though a dummy variable on race often does have explanatory power , banks do not decision on this variable, they try to keep it out of databases so others can't mine it for evidence of discrimination. If you do have it, and off-the-record analysis on this variable you would severely hurt your career. If anything, there is a lot of deterioration of objective criteria to generate sufficient volume for various target groups.

I'm color blind, which makes me a racist to many. I think they are the bigots, and as India has shown, heightened sensitivity to group distributions does not lead to greater wealth or harmony.

Anonymous said...

Am I drunk or are you blaming the housing crisis on diversity programs? Do I read you correctly?

I've come to expect crazy right-wing rants on this blog, but this one is weirder than any before.

zby said...

"Banks don't use race in their underwriting, if they did, they would be sued." - OK - this is an effect of a law, this law forces banks to use some non optimal methods and thus decrease their efficiency in the name of basic justice. I hope you agree that this is a good policy.