Thursday, July 31, 2014

The Signal & The Noise - Nate Silver

I am currently reading a book written by American statistician, baseball modeller, political pundit, Nate Silver.


Besides creating some very successful baseball models, Silver came to national prominence in the US for correctly and very accurately predicting the 2008 and 2012 Presidential Elections. 

For those that are unaware, in the lead up to both elections the right wing media were adamant that the polls predicting Obama's victories were wrong and that the race was neck and neck. It wasn't and Silver called them on it. In the end he was virtually 100% correct. 

In the book he discusses how human behaviour interacts with maths. Do 10 polls on a subject and if nine show one thing and the other something else, people (and media) will seize on the outlier and assign it more significance. 

So where does this intersect with Wargaming? Well one of the points Silver makes is that people are generally overconfident in their predictions. I reckon this is a trait evident in wargamers. Check out forums, podcasts and dare I say it blogs and you'll find an environment that is strongly confident in its views. 

The tournament reports on podcasts contain little in the way of objective analysis of success and failure - instead assigning results to general all round "awesomeness" or unbelievably bad luck. Generally good luck or opponent skill gets dismissed pretty quickly. 

Over 15 years of internet interaction with Wargaming I believe that the number of truly objective commentators I have come across could be counted on one hand - in Fantasy I'll give hat tips to Charles Black and the Polish gamer Furion. 

A classic example of this overconfidence in ability is when a community goes to an international gaming event for the first time e.g ETC. Generally they will overstate where they will finish because they are basing their prediction on irrelevant info - performance in their own jurisdiction - while ignoring important information - the skill and experience of other teams. You see a reality check hit pretty fast. The next step is important. Do they dismiss the result as an outlier (a lot do) or do they use it as fair reflection of that particular environment. Similarly if generally you finish 10th to 20th and one year you finish 3rd or 4th, is this a freak result? 

Anyway the book is an excellent read that combines stats with psych. I love this shit. 

No comments:

Post a Comment