GITREP technical problem: Attrition

We’ve reached a point in the GJOpen game where, although the number of forecasters keeps increasing, the number of forecasters who have forecast on every open and closed question in a cluster that I have defined has decreased to the point that certain clusters, such as Iran, have no covering experts.

This is a result of several factors:

  • Some of the more prolific forecasters have quit the game, in particular @einsteinjs and @cmeinel.
  • Remaining forecasters are slowing down and not covering every question.
  • The number of questions has reached the point that it is unlikely for any one forecaster to cover enough of them that the forecaster will happen to cover all of the questions in a cluster that I have defined

So I will have to rethink and reprogram my super-picker.  Here’s a thought:

  • Look at who is in an open question in a cluster.
  • Of that set, find out who accumulated the highest sum of accuracy scores among any closed question in the cluster
  • Rank them and pick the highest.

This probably a better algorithm than I’m using anyway because for Iran in particular my number of experts was dwindling from the original target of 12 down to 4 and finally down to none.

I am too tired tonight to program this, plus it’s New York election night and season-enders of Girls and Better Call Saul, so it will be a day or two before I get to it.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s