Congressional Districts should be defined by software - not by politicians. I do not know how to do this, in any great detail - but I do know the steps that have to be followed.
The first step is a detailed map of how each small area has voted in the past. This could be taken from voting records, that already exist. A study would have to be done on how best to define these areas - perhaps by zip code or voting district.
The next step is easy - and it is now being done by Unsupervised Machine Learning techniques in many disciplines. Similar data is lumped together into separate areas. Red people are lumped together in Red areas - and the same for Blue people. If the number of areas discovered equals the number of districts required - the job is over.
But what will probably result, is a weird pattern of some kind - including some isolated Red or Blue areas. And, as far as Software is concerned - this is where the problem becomes interesting. How to change these areas into the required number of areas?
One way of making this easier is to make districts that are not contiguous. One district could be made of a little blob here, and another little blob there. They might even be allowed to be quite far apart. This sounds complicated, but once defined, everyone would know where they are.
The only way to do this is by trial and error. Different algorithms would be devised for each situation - to see which ones work best. And to discover which ones could be applied to the largest number of situations.
There are some very sophisticated algorithms in existence now - some involving a number of steps. We can put them to work for us.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment