I was not really familiar with Chris Genovese, the blogger at Signal Plus Noise until he hosted this week’s COV. I love the tagline - All Models Are False, Some Are Useful. How true.
I looked around his site and found some interesting proposals for resolving the problem of gerrymandering. Chris does a nice job critiquing Gary Farber’s proposal so I won’t discuss that here. However, I will discuss one of Chris’s proposals.
Going even further in the “lateral” direction, and veering decidedly into crazy territory, is what we might call the “self-organizing state”. The idea is to eliminate geographical districts entirely in favor of groupings chosen interactively by the citizens. Here's a scheme to give the flavor, though I'm not seriously proposing it. Once every ten years, registered voters place themselves into groups of, say, 100–1000 people based on any criterion they like: location, family, demographics, politics, hobbies. The only limitation is that each person belong to just one. Groups can have a public statement of guiding principles, but each person joining a group would also be asked to answer pre-specified questions to characterize her positions, with non-responders allowed. Average results for each group would be public knowledge, Each person then rates any number of other groups on similarity to his own; unrated groups are assigned the highest dissimilarity rating by default. One result is that non-responders get lumped together randomly. We then average the dissimilarity ratings over each groups members, construct a minimum spanning tree, and use the tree to put together “districts”. OK, back to reality. Fun's over. Move along. [It seemed like an amusing idea at the time.]
Virtual districts? What an interesting idea. Chris did not seriously propose this, but perhaps he should have. I can think of many objections to this for State districts – for example, State representatives need to represent a physical location in their own State government or I expect we would see even larger problems when it came to local infrastructure. However, this has a lot of potential for national elections. It could dramatically improve the chances of a third party becoming viable. For example, Pennsylvania has 19 representatives in the House. Assume 10% of the registered voters in Pennsylvania were Libertarians. If they could form two districts, they would easily win 2 seats in Congress vs the zero they will win under the current situation (assuming the Libertarians are evenly spread out over Pennsylvannia).
Of course, this raises the question why do we need districts at all in national elections? Assume Pennsylvanians could vote for any candidate for Congress no matter where they lived in Pennsylvania. However, they could only vote for one person (as is the case now). The nineteen candidates who received the most votes would all win. This might be a headache for the major parties (OK Joe, you advertise in the North, otherwise you’ll get all the Republican votes and we’ll only win 1 seat), but that may be a good thing.