Monday, June 21, 2010

Perverse Incentives

From an institutional standpoint, developing a fair incentive structure is extremely important. Many of the excesses on Wall Street are great examples of incentive schemes gone awry.

I read Roger Lowenstein's "The End of Wall Street", a couple of months ago. It presents the circumstances that led to the Enrons and MCI Worldcoms of the last big crisis on Wall Street. Besides telling a great story, it fleshes out a laundry list of perverse incentives that were structured to encourage the very irresponsible behavior, that eventually undermined the system.

Auditors, encouraged by the consulting business that writing lax reports would bring to their parent companies (think Arthur Anderson, but most big accounting firms were complicit in some form) allowed, and even crafted the bogus accounting smoke and mirrors, that was their fiduciary duty to reveal. Board members, received cushy salaries, for doing everything like schmoozing, playing golf together, etc., everything that is, but their job - keeping an eye on the management on behalf of the shareholders. The list goes on and on: stock analysts, politicians, stock-options, CEO compensation plans, etc.

Very few incentives in the system, encouraged the right behavior.

The reason I bring this up, is because academia is beset by many of the same afflictions.

Incentive schemes are just as perverted inside our campuses, as on Wall Street.

Funding has become the primary metric of scientific worth.

I know this is an over-generalization, but sadly it is also extremely close to mantra that guides most research universities in the US.

The university gets to keep 1/3 of the funds as overhead, which keeps the administration happy, especially in these times of declining governmental support. Funding lets the researcher hire more grad students and post-docs, which keeps the departments happy, since they can boast about the great number of degrees awarded.

Since this is the "path to happiness", everybody wants to raise a lot of money.

Thanks to the modern word processor, it is easy to flood the systems at NSF, NIH with proposals, with the hope that one of them hits the jackpot. This directly leads to a broken review system, with low acceptance rates, which unfairly, but understandably, favors well-connected big-shots and academic rock-stars with name recognition and a fan following.

Because of low acceptance rates, the amount of time wasted on writing proposals is a huge time sink, which draws resources away from places where they could be deployed more fruitfully: for example, to advance science, instead of fundraising.

The scientist suffers. Science suffers.

We probably overproduce PhDs, again due to perverse incentives. The reputation of a university is often built on the shoulders of its graduate program. Many of these PhDs now spend a lot of time looking for their first real job.

Economics instructs me to diagnose this evidence as an excess of supply over demand.

The grad student suffers.

There are many other manifestations of this perverse incentive scheme, including sky-rocketing journal subscription fees, the cancer of cheap, for-profit journals which exist only to pad resume's of scientists too busy writing proposals, the near complete break-down of peer review in many hot fields, which publish garbage, and random SEM images with reckless abandon, the toll that all this takes on teaching core classes, diluting the learning experience for students.

Most of these ills exist and flourish, because somewhere the incentive structure was compromised.

That good science still happens is a miracle. It is despite, not because of, the system.

Wednesday, June 16, 2010


1. Steven Stogatz's popular columns on the elements of mathematics in NYT, are now all available at one convenient location.

2. A bunch of game theory lessons (videos) now available as a playlist on YouTube here. I was first introduced to game theory via Richard Dawkins popular books. I found, and continue to find, the link between basic math, logic, psychology, and economics endlessly fascinating.

3. The computing power of the first supercomputers and many mobile phones are now comparable. The funny thing is that I didn't even raise an eyebrow when I read the blog, despite my natural tendency for skepticism.

Sunday, June 6, 2010

John Hussman Commentary

I find John Hussman's weekly commentaries on financial markets very informative, mostly because they are very different from noise that important looking "experts" with (faulty?) crystal balls spew on CNBC. Sometime back, I linked to this page, which clearly demonstrates that forecasters, as a group, merely extrapolate the past into the future.

This is a great disservice, because most of the numbers that these experts divine, pretty useless from a practical standpoint. In addition, the mask of conviction with which these experts conduct themselves instills a false sense of confidence in people who like to listen to such talking heads.

I like John Hussman's balanced commentaries because of the how heavily he uses concepts of probability, and Bayesian inference in his analysis. I wish they taught this stuff more widely. For example, in his most recent commentary linked above, he comments about the Gulf of Mexico oil-spill:
With regard to oil spills, however low one might have believed P(we'll have an oil spill) to be, prior to the recent accident, the "prior" probability estimate should change given that we've now observed one of the worst oil spills in history. Even if the oil industry previously argued that the probability of an oil spill was one in a million, it's hard to hold onto that assessment after the oil spill occurs, unless your faith in the soundness of the technology is entirely unmoved in the face of new information.
(John Maynard Keynes would have paraphrased it as: "When the facts change, I change my mind. What do you do sir?")

Hussman goes on to do a back-of-the-envelope calculation which suggests that since the number of deep sea oil rigs has increased dramatically, the chances of seeing catastrophic oil spills are actually quite significant.