Unfortunately, due RAC being a weighted moving average, there is a long tail on the distribution for all the computers that have switched projects. At the current time, the only way to know if the top 4 computers are all actively crunching and the mean is valid is if someone looks at the data by hand. I've got some ideas on how to mitigate that, but I don't have time to implement it just yet.

I'm working on saving previous calculations. Right now I'm redoing how users submit their component name. Once the component names are standardized, recycling the old results should be mostly trivial.