The current redistricting solver is probably pretty close to being a workable distributed client that people could download and run and help with the computation on their home computers. It has a reasonably small ram footprint, and runs at low priority well, I do this much of the time on my own computers. It just needs a wrapper to manage downloading a state of data and uploading results.
There is another option that I've considered some times which is to expand the graphical display version that I have for MacOS X and make it into a full editor that could be used to hand-tweak district maps. It would have the measurement ability built-in to get compactness measures and demographics. It might be interesting to see what people come up with that the automatic solver wouldn't. I'm sure it's sometimes more satisfying to get in and work on it yourself instead of just letting the solver work, and I think there could be better following of actual community boundaries if someone who knows the area draws the map by hand. I don't know how much that will deviate from 'optimal' compactness, so I guess we'll just have to try it and see.
The blog poll for the next few months is this, do you want a nice quiet background solver or tweaking and tinkering by hand?
(With a little luck, there will be time to make both.)