JOHN ROBB

Information Cascades

Information cascades are important. They explain many of the big problems we currently face. Here's how an information cascade works:

1) An event occurs or a problem surfaces.

2) A person who is perceived to have good data/insight into the event or problem makes a decision.

3) Other people, observing the first person's decision, opt to avoid original analysis/discovery and copy the earlier decision.

4) The more people that copy the earlier decisions, the less likely any new discovery or analysis is done.

If the earliest decision was correct, then everything works out. If it isn't, the error is compounded until it becomes a major problem when it collapses. In today's world, with its copious communications systems, information cascades occur with increasing rapidity. They can spiral out of control in hours/days. This also means that on the flip side, an information cascade can collapse suddenly, when the original decision(s) is(are) proven wrong.

An information cascade is a good explanation for why the financial industry became so hopelessly lost. They based their decisions on economic and financial theories that claimed to have predictive power in EXTREMELY complex domains (since this methodology came from careful study in academia, it was assumed to be correct). The information cascade was the rampant application of these theoretical models throughout the financial industry over decades. However, as Nassim Taleb (although the writing is often tough to decipher) points out in this brief, the true predictive capacity of these models was nil/nul/void, since they are extremely vulnerable to black swans -- the collapse of the massive hedge fund, Long Term Capital Management proved this point nearly a decade ago. So, what happened when the black swan arrived that demonstrated that these models were wrong? The system imploded as confidence in all of the previous decisions made with those models are called into question (the information cascade collapsed).

In warfare, information cascades can generate disasters too. For example, the government's recent claims of success in Iraq (although better models of the conflict indicate that this is a temporary lull, far from "victory") can lead to similar adventures in the future (as we are seeing that now in Afghanistan) as the press and population echos the assessment. Finally, information cascades also can work as a means of disruption. As a guerrilla, information cascades can turn an attack on nominally stable social systems into a systempunkt (the critical point within a system, that if attacked, amplifies damage throughout the entire network). Fore example, think 9/11 or the Golden Mosque.

Posted by John Robb on Thursday, 18 September 2008 at 09:44 AM | Permalink | Comments (4) | TrackBack (0)