Thursday, November 13, 2008

Endowment Effect II

So I had a nice discussion about probability distributions concerning ideas. Unfortunately I failed to make a sufficient connection to answer the question.

How would a method of more accurately accounting for the truth of our beliefs counteract the endowment effect?

Normally when considering the probability of alternatives there are two keys parameters: 1.) probability of outcome's occurrence 2.) magnitude of gain or loss given the outcome's occurrence. In this way widely different "projects" become tradeable. For example, I have \$100,000. I am proposed an investment opportunity in the stock market with a 5% chance of earning a 100% and a 95% percent chance of earning 0%. The expected payoff for this investment is simply calculated [.05*(\$100,000*1.0) + .95(\$100,000*0.0) = \$5,000]. One could compare this opportunity with say an investment in alpaca farms. Let's say this opportunity is expected to yield \$20,000 during the same time horizon. Although these are wildly different projects, they are tradeable in the sense that they are alternatives.

In the market of ideas, the value of ideas don't appear to be tradeable. Ideas are limited to the topic, within the topic they are tradeable, cross-topic tradeability doesn't hold. So we have millions of topics (could be policy issues, philosophical questions, scientific inquires, etc) and within each topic are alternatives. Each alternative could theoretically have a truth probability. Hence the only value ideas hold are their truth probability. We could conceivably multiply this probability by some measure of the value of the topic in general, which would enable cross topic idea rankings. But that is not the subject of our inquiry. We are concerned with the appropriate level of confidence in our idea's truth concerning a specific topic.

Back to the endowment effect. What the endowment effect means in this context is that ideas that are "owned" are given too great of value. Hence, their truth probability is ratcheted up too high. My argument is for a method that more accurately understands truth probabilities. I do not attempt to counterbalance the bias but instead hope to improve the mechanism causing the error.

Often I find myself falling into the trap of discrete thinking towards ideas. Either and ideas is "right" or "wrong." 0% or 100% probability. When in reality nothing is so cut and dry, especially on complex issues. This flaw seems a natural shortcut for the mind and may be behind the endowment effect in ideas. (Although we certainly haven't proven there is such an endowment effect, and given its different features it may deserve a different title.) If we can learn to think probabilistically, instead of in all-or-nothing terms, we may realize greater rationality in the domain of ideas.