Monday, October 24, 2005

Irrational Games

Thomas Schelling and Robert Aumann were this year's Nobel Prize winners for economics. They shared the prize for their contributions to game theory, with Aumann helping to create the theoretical foundation of the field, and Schelling using the theories to help interaction between nations.

Other than the Peace Prize, it seems that the economics prize always generates the most controversy. Typically, this is driven by a conflict in the political ideologies that support or attack the winner's economic theories. This year, the conflict was no less, but it seemed less political than usual, and more academic in nature. Several articles came out that attacked game theory as a field itself, claiming that it was too idealistic and based on too many assumptions shown to be false in real life. One such article was written by Michael Mandel in Business Week. In it, he asserts:

Instead, the real progress in economics these days is coming not from game theory, which has been around for 60 or more years, but from the much newer fields of behavioral and experimental economics. Behavioral and experimental economics don't start with the assumption of rationality used by game theory. Rather, as the name suggests, the focus is on looking at how individuals and organizations actually make decisions in practice, including systematic biases, misperceptions, and just general all-around bloody-mindedness.


The article basically states that game theory is intricately intertwined with the assumption of fully rational behavior, and that fields such as behavioral and experimental economics are simply incompatible.

With all due respect to the author's PhD (I'm just an MBA student, and not even specializing in economics), but this isn't how I've learned game theory. I have always understood game theory to be more of a framework to determine strategy, based on the analysis from all the players standpoints. It doesn't matter whether the players make rational, utility maximizing choices, or not. What does matter is that you know what drives their choices.

If you can predict other parties' action, no matter the motivation, the framework of game theory can still apply. These choices can be based on truly rational self-interest, on altruistic social conscience, or on the ego of the CEO. In determining these decisions, behavioral and experimental economics are indeed valuable, as they often predict individual behavior better than the standard assumption of purely rational value maximization.

Therefore, the important thing to learn through game theory is not the fundamental assumptions of what drives behavior. Rather, it is more important to learn how to build upon the assumptions to find the correct response to a given situation - in order to maximize whatever drives YOU.

Tuesday, October 11, 2005

What is Google?

A few months ago, there were several rumors that Google would create a payment system that would compete with PayPal. Several peopled took this as a sign that Google had strayed away from their core competency, and that they had no relevant experience in being a financial company. However, as you look at the growth of new services whose sole revenue is from Google's AdSense, you start to wonder if Google really is new to the financial business. Paul Kedrosky's worry that Web 2.0 is just a google affiliates program is like wondering if Web 1.0 was just a program for Visa and Mastercard.

Once again, the beginnings of this idea came from a conversation I had with Charlie. The value of his attention is important to him, and he was wondering what the best way to get compensated for his interactions on the web.

I think the biggest obstacle to this goal is with payment systems. Every day, each of us performs a series of activities on the web that could potentially create value. The trouble is that typically each discrete activity produces a very small amount of that value. I may generate a lot of value in a day - through my browsing history, or my tagging, or through referrals to new services - but it's broken into too many pieces. Even worse, very rarely do more than a few of the pieces have value to the same party. In aggregate, the total value created and the total value enjoyed are large, but the value of each transaction is so low.

It's a little difficult to explain clearly what I'm trying to say, so here's an example. Let's say 1000 people in an ecosystem each perform 100 activities every day. Each of these transactions create 5 cents of value to one of 100 other parties (websites, advertisers, etc). So each of these users creates $5 of value a day (100 x .05). Each of the "payers" could conceivably receive $50 (1000 x .05) worth of value each day. So as a system, there is $50K of value created everyday. But any two parties would only ever have a transaction worth 5 cents - so the costs of sharing that value are too large, and they end up giving up that value. Either one person gets all the value (like when I read a blog and learn something, but don't give anything back to the author), or worse, the transaction doesn't even occur (if I didn't refer a good service, because I got nothing out of the referral).

There 2 options to monetize your created value in this environment

  • Limit your own interactions such that all 100 of your transactions are with 1 person, so that you can build up $5 of value with them, where it's actually worth splitting it.


  • Figure out a way to keep a big central clearing house, lowering the costs of splitting the value via economies of scale, to the point where it's worth doing, and then charging for it (google or paypal model)


Google has done the second with it's AdSense program. People who have been talking for years about developing a reasonable micropayment scheme for paying for content on the web seem to have missed the fact that Google has done just this. Because of AdSense, I can now monetize my attention, and pay the producers of content with that attention.

This is just the first step though - my interest in a subject is just one transaction that can be monetized, and arguably the one with the least value. Just yesterday, I referred people to 2 different online services - Writely, and NumSum (future post about trying to move entirely online is forthcoming). In both cases, I've provided value to these sites, but got nothing in return. If each of those referrals was worth < $1, it's not worth it to the sites to set up a reward program, or to me to sign up for it. However, what about a "referral central", that would allow smaller sites to subscribe to a joint referral program. They pay for 1000s of referrals at a time, and I get credit for all my referrals across multiple services.

Moving even further up in value, is pay for sales. Amazon and others have affiliate programs, but this is essentially option one - limiting your referrals to one store, to build enough credit there. What about a meta-affiliate network, where I could refer people to a variety of specialty stores, and get credit for every sale. This gets closer to the holy grail of advertisers of "pay-for-sale", and allows me to refer others via blog/email/IM to the best place, not just where I have an affiliate set up.

These are exactly the types of things that CAN'T happen in the off-line world. Why haven't we made them happen on-line yet?