Saturday, March 11, 2006

FutureWatch Predict

FutureWatch : Prediction ranking by Google

Search engines like Google cannot examine and rank the future , but they can certainly rank past predictions about the future .

Benefits : the continual feedback will lead not only to the identification and ranking of predictors ,  but hopefully make existing ones take more care (eg think-tanks) . The creation of verifiable marketplace for predictors will lead to much better prediction systems , something a high-tech society desperately needs .

Existing systems:
This ranking is already done to some extent by money ( eg Warren Buffet in investments)  . This is also the only category where the accuracy of predictions is routinely measured  .

Political elections also can be seen as to a measure of the accuracy of a politician’s predictions .

There are a large number of other prediction systems : think-tanks , futurologists , science-fiction writers , scenario-planners , politicians , climatologists , astrologers , mediums , tarot , numerology , etc , etc .

The problem in a fast-changing , high-tech society is that irreversable  changes might take place before this evaluation process has caught up .

Argument for a new system:
It is clear that some individuals and organizations are better at prediction than others . The reason  why might be unclear , but some consistently outperform others .
The problem is to find them . This is the same as ranking them for accuracy above chance . A formidable undertaking .

A large statistical base with built-in feedback is required for any method .
Until now , only money was available as a reliable ranking mechanism .

A Google-type voting ranking system seems another viable method .

Google does not have to decide whether a predictor is better : superior predictors will draw more links , and poorer ones less .

Categorization :  some suggestions
By time and by type.

Scientific : Climate , population , etc
Etc , etc

Initial set-up problems:
This could be costly (if not impossible) .
Ask for help from your users .
Set up the categories .
Ask the predictors and the users to evaluate each other . This is viable only if a statistically large amount of data can be assembled .

Commercially interested predictors and users have an obvious interest in blowing their own horn and shooting down the competition .

But there is also a large class of vitally interested users in predictions : civil services , farmers , etc .

To repeat , this exercise would only be possible on a large statistical base .

But it will be very lucrative .

Andre Willers

No comments: