Read This!

If you had followed the previous two blogs on assessing the impact of new information on prior estimates, here is more context on the Bayes Theorem.

Great reading:

Sharon Bertsch McGrayne: The Theory That  Would Not Die: How Bayes’ rule cracked the Enigma code,hunted down russian submarines & emerged triumphant from two centuries of controversy.

Absolutely brilliant. Opened my eyes to the fact that even the most scientific or mathematical of disciplines can be fought over with religious zeal.

What prompted this series of blogs was an excellent companion to McGrayne’s book treating the subject of forecasting in general – and showing how indispensable Bayes’ theorem is, highly delightful with examples from sports betting, poker and scouting for sports talent, then going on to treat major policy issues. This is where I picked up the breast cancer example again, and I realised that this is a much more attention grabbing example than any of the more mundane management examples I had previously used in my workshop.

The signal and the Noise by Nate Silver is a must for every one in the fields of information technology and also the dismal sciences that spawned off economics. He shows convincingly how the explosion in data and information contributes to more noise and more failures in judgement. More about that in a coming blog.

SIlver defines the difference between the so called frequentist approach in statistics with the Bayes’ approach very well as two different approaches to the value we put on human judgement. The first departs from the position that our judgement would be perfect, if only we had exact information – and Bayes departs from the position that human judgement is by its nature flawed. We can only be less and less wrong as we get newer and relevant information. The frequentist approach assumes an objective reality that we can access without contaminating with our perception, Bayes that we need to incoporate our human fallibility in our assessments…

For those of you who participated so enthusiastically in the conversation, you will absolutely enjoy these two books and then go and ask why your statistics lecturers did not acquaint you with Bayes!