I have a powerful computer... I have encountered the rare situation where I wish I had more processing power. Performing statistical analysis to generate district maps from census data is pretty darn heavy duty. One map generation process has been griding for 20 hours so far.
On the upside, it's fun learning a new and different programming language-- the R language was certainly outside my comfort zone. Once I got the hang of that, the Better Automated ReDistricting (BARD) package wasn't too difficult to get going. But then, I've been a programmer for some time, the learning curve would be steep for Joe User.
My thanks to the R and BARD folks. You did and do great work! This stuff is very powerful and I am grateful you put it out there for the world to use.
It'd be nice to see an end user useful version of BARD like functionality but, let's face it, there's no profit in building scientifically based, politically neutral district drawing tools and making it available to the general public. The parties and their consultants will hoard their abilities to maintain their power. Unfortunately, I underestimated the learning curve and don't think I dug in soon enough to make an impact on county redistricting. Perhaps this new found knowledge can be utilized at some later point.
Update: 52 hours of processing time later the createGreedyContiguousPlan function is done running. The results are pretty bad. The districts are barely contiguous and not compact and the population totals are outside allowable ranges. The computer chose to use unpopulated river and train track blocks to link unrelated neighborhoods in to districts... It *looks* gerrymandered to me.
No comments:
Post a Comment