Scientific Sense Podcast

Sunday, February 2, 2014

Efficient designs

A recent article in the Proceedings of the National Academy of Sciences shows that a group of non-experts in aggregate produced designs for RNA molecules that are substantially better that what were created by computer algorithms. Humans have always shown significant capabilities for pattern finding that are unsystematic. However, it has also been shown that perspectives of a single individual are inferior to what could be accomplished by systematic computer searches. So, it is the group of interacting people, albeit being independently imperfect, that shows superiority over machines. The slight imperfections in each provides random excursions in the search space that could get reinforced or negated away by the crowd. In financial markets, this is called efficient prices and in other areas, wisdom of the crowds.

This idea appears to be poorly understood in many areas. In the financial markets, significant value is lost by advice and individual trading of securities, based on the premise that individuals can beat crowd wisdom. In large companies, decisions are made by a few managers without extracting the knowledge that reside in the people who make up the company, based on the premise that they know better and/or the broader information cannot be systematically gathered. In politics, representative democracy destroys value by instituting a small number of decision-makers between the population and policies, based on the premise that the representatives know better and/or the populace cannot realistically express opinions on policies.

The idea is simple. Human crowds are the most powerful logic engines. Neither machines nor a few selected people are going to have any better insights. Humans interacting with machines that allow collaboration and aggregation likely dominate in most areas.

No comments:

Post a Comment