So there has been some recent debate about redistricting or “gerrymandering.”  The crux of the debate is how much we can judge districts by their shape.

My own view, written before the debate, is that the shape of districts tells us very little about whether they are “good” districts or whether a redistricting plan for a state is a “good” plan.  Determining what a “good” district is means having a theory about how voters should be represented in Congress. And developing such a theory means making trade-offs among various criteria that are required by law (notably the Voting Rights Act) or are accepted norms of drawing districts: geographical compactness, the representation of racial or ethnic minorities, existing community boundaries, partisan competitiveness, votes-seats proportionality, and so on.

But many people don’t wrestle with all this.  And so — in my e-mail inbox at least — you get reactions only to how a district looks, and districts that have “weird” shapes are assumed to be “bad” and “gerrymandered.”

To give you a sense of why you can’t judge districts by their shape, take a look at the following four redistricting plans for Louisiana, which come via the folks at Fair Vote.

Now here’s the quiz.  Identify which of these maps is:

  1. The current plan, consisting of five Republican seats and one Democratic seat.
  2. A map that comes closest to creating actual proportionality — where the percent of seats controlled by each party matches the statewide 2012 presidential vote.
  3. A map that a nonpartisan independent commission might produce — taking into account equal population, compactness, contiguity and preserving communities of interest.
  4. A pure Republican gerrymander in which every district would be at least 58 percent Republican or greater, based on the 2012 presidential vote.

Now here’s a cute picture of two monkeys so you can’t cheat by looking down the page at the answers:

Here are the results:

  1. A is the current map.
  2. D is the map that achieves proportionality.
  3. C is drawn using nonpartisan criteria.
  4. B is the Republican gerrymander.

In other words, the plan that drew the strangest-looking districts, Plan D, was the one that actually created the least “gerrymandered” outcome in terms of the translation of votes to seats.  Moreover, two plans that drew what look like fairly normally shaped districts, B and C, were done using completely different criteria. The shape of the districts couldn’t tell you that Plan B was the purely partisan gerrymander.

For more, you can read the Fair Vote report and its ideas about how to produce better districts (hint: draw fewer). But, in the meantime, don’t judge districts by their shape.