confirmation bias

I recently finished reading The Wisdom of Crowds and am now just starting The Black Swan. One of the ideas discussed in both books that I find interesting is that widely disseminated information can make a crowd less wise because it decreases the variety of thought within that crowd due to a clustering of opinions resulting from that information.  If you start with an already homogeneous crowd, conformity is stronger, unchallenged preconceptions become more likely, and significant confirmation bias results.

It is interesting to see these processes work in operational weather forecasting. The inherent uncertainty associated with weather forecasting often leads to the bandwagon effect, possibly due to the desire to herd in response to the threat of uncertainty. Like anybody else, forecasters often have preconceptions about what will occur, resulting in confirmation bias. Finally, there is a significant threat of anchoring, or placing excessive emphasis on a single piece of information.  Computer forecast models (also known as Numerical Weather Prediction - NWP) produce new forecasts at the rapid rate of every 6 hours or less. Observational data (e.g., radar, satellite, surface weather conditions) update at an even faster rate.  Very often the latest piece of information is considered the "best" ( recency effect ) because the atmosphere, being non-linear, has much less predictability with time.  Therefore, datasets that are sampled or presented closer to a given future time are automatically considered more useful.

It will be interesting to plow through this book as time allows. Weather certainly presents its share of events that have great impact and have a high degree of improbability, uncertainty, and unpredictability.  It is remarkable how many of these events later get presented as case studies that overstate the ability to have predicted the event in advance ( Hindsight bias ).