Archive for October, 2009

Saturday, October 17th, 2009

Prediction Market: A Generator for Social Intelligence? (Part 1)

Prediction Market  first gained some fame when Google (who else can it be…:) announced the positive usage of the concept to allow their employees to ‘bet’ on predictions - such as certain launch dates, choice of project ideas, etc. Prediction Market has been slated to be an exciting, new technology for many years now, but until Google came out in 2007 and again in 2008 to announce the benefits, it was in realistic, imminent danger of becoming an almost-there-but-lets-RIP-it technology.

A quick primer on Prediction Market is probably helpful if you have not already wiki it: Prediction A, B, C, and D are listed in a virtual stock exchange, with a price of $1. You are informed that you can buy shares in each prediction stock, and the prediction stock will be priced at $10 if it happens, and $0 if it doesnt. (implicitly each of A,B,C and D has a 25% of occuring, and only 1 will occur). The price also changes according to market demand and supply matching, just like real world stock exchange.

When Gartner listed PM in its famous technology hype cycle chart, I had to admit my cynicism at first. For a start, how can a betting platform qualify as a technology? At closer look however, a technology empowers the organisation as well as bring new dimensions to the business process. A Prediction Market fulfills this; by incentivising a crowd to bet on predictions, an organisation can expect the following:

• Instead of merely expressing an opinion, the speculator now makes his choice based on what he believes will happen. The fundamental difference lies in what a person believes if he was or not rewarded.
• The speculator may also make his choice by ‘timing the market’. That is, he buys a prediction when it is underpriced, and sells it when it is overpriced. If he wants to benefit from groupthink, he may buy the prediction that everyone believes in irrationally, and sells it just before the final outcome of the prediction
• Participants are also naturally more interested in such platforms than standard surveys and questionaires.

It is thus no surprise with such compelling composition of data that companies such as Inkling, Crowdcast have already jumped onto the bandwagon in enabling enterprises to tap on “Wisdom of Crowd” for collective intelligence. However, there are some pre-requsites that has to exist for this predictive technology to work. 1) Sizable crowd 2) Incentives and Motivation 3) Ease of use. (Now you see why PM is easy to deploy in Google; 1) thousands of employees 2) huge chest of cash 3) super smart employees who would have no issues with simple math). In some cases, data may also be meaningless from frauds and extremely biased (see http://www.atimes.com/atimes/China_Business/JC14Cb01.html)

In my next post, I will discuss some possible configurations of Prediction Market that would address these issues. As you may have realised by now, I am a huge fanboy of PM