The Power of Plausibility Theory
A new form of decision analysis is helping executive reevaluate risk management. In a nice PDF, Tim Laseter and Matthias Hild wrote a good introduction on what might be a post-bayesian statistics applied to business decision making.
Despite the mathematical proof defending the logic of expected value, in real world, we, human, aren't much rational when it comes to making decision. We all know (do we?) current right wing ideology pretend that the market is maid of rational users. If this was true how can we explain why we are so poor at picking the rational decision.
Let's face it. If in a coin toss, I offer you $100,000 on heads but you'll pay me $50,000 on tails, few of you will rush to take the wager. Although the expected value of this bet is a positive one, i.e. ((50% x $100,000) minus (50% x $50,000)) yield effectively $25,000. At least a logical machine would bet right away. For us, mere mortals, the potential downside - loosing $50,000 - is simply too great.
That is because we do use our instinct : "how much can I lose?" "What's the likelihood of a bad resulting occuring?". Logic isn't in the radar. What I like of Plausibility Theory is that it recognizes that we people choose knowable over unknowable risk rather than a simple examination of (logical) expected value.
This explains why we do weight more risk over expected value. And sometime backing up when something new is coming. Open Source Software evangelists should take note.
Despite the mathematical proof defending the logic of expected value, in real world, we, human, aren't much rational when it comes to making decision. We all know (do we?) current right wing ideology pretend that the market is maid of rational users. If this was true how can we explain why we are so poor at picking the rational decision.
Let's face it. If in a coin toss, I offer you $100,000 on heads but you'll pay me $50,000 on tails, few of you will rush to take the wager. Although the expected value of this bet is a positive one, i.e. ((50% x $100,000) minus (50% x $50,000)) yield effectively $25,000. At least a logical machine would bet right away. For us, mere mortals, the potential downside - loosing $50,000 - is simply too great.
That is because we do use our instinct : "how much can I lose?" "What's the likelihood of a bad resulting occuring?". Logic isn't in the radar. What I like of Plausibility Theory is that it recognizes that we people choose knowable over unknowable risk rather than a simple examination of (logical) expected value.
This explains why we do weight more risk over expected value. And sometime backing up when something new is coming. Open Source Software evangelists should take note.
Martin Lessard (10:50 PM) Permalink
2 Comments:
"At least a logical machine would bet right away."
Actually, it would not. A logical machine would bet right away if you offered a series of such bets, where the expected value means something. There is no expected value with only one coin toss, since either results have the same chances of occurring.
But it is an interesting article and point of view. And I agree that we mostly go for "the devil we know" (i.e. the knowable risk). Partly because we do not have the luxury of large numbers... (isn't that why insurance companies exists, because *they* can count on law of large numbers.
You are right. It is on set of multiple draws, that we can consider betting as "safe".
I guess what we, human, have to understand, is why "instinct" is taking so much place instead of "rational" in such decision. Is it a security mechanism? As we enter in a new world of knowledge economy, old biologic strategic reflex may interfere here. At least as long as we think we should act as rational animal...
Post a Comment
<< Home