Kos plans on voting for Arnold Schwarzenegger's "redistricting reform" ballot initiative, while Nancy Pelosi's making it her "top priority" to defeat the measure. On the face of things, I think Pelosi's right and Kos wrong. Yes, redistricting reform makes sense if it can stop legislators from gerrymandering themselves into permanently safe seats, but there are good ways and bad ways to enact reform, and Schwarzenegger's proposal seems like one of the bad ways, judging by the initiative's proposed guidelines for drawing up districts:

Judges must maximize the number of whole counties in each district, and minimize the number of multi-district counties.

Judges must maximize the number of whole cities in each district, and minimize the number of multi-district cities.

Districts must be as compact as practicable. To the extent practicable, a contiguous area of population shall not be bypassed to incorporate an area of population more distant.

This looks dubious. Under the second guideline there, the judges drawing the boundaries could end up packing the majority of urban voters into a few concentrated, ultra-Democratic districts. (The first guideline might, equally, pack Republicans into conservative "counties," but I can't tell without data, and am guessing this would be a smaller effect.) Schwarzenegger's plan wouldn't necessarily lead to more competitive districts either, as is widely hoped. Since "[j]udges must maximize the number of whole cities in each district," you'd have a handful of ultra-safe single-city seats that would vote overwhelmingly Democratic. If you wanted more electoral competition, then you'd try to create a bunch of districts that, say, combined parts of "blue" urban areas with parts of "red" suburbs. Schwarzenegger's plan does the exact opposite.

Now his plan would give representatives more "natural" regions to represent (i.e., it makes sense to represent a whole city rather than parts of two different regions), but that's a different goal from either a) ensuring competitiveness or b) making sure that voters have anything like proportional representation in Congress, and should be sold as such. Plus it looks for all the world like a naked, calculated power grab, rather than a solid reform that just happens to hurt the Democrats. (I'd happily support the latter; not so much the former.)

Moreover, the initiatives's requirement that districts must be "as compact as practicable" doesn't necessarily make sense. Pundits love to bemoan the fact that many congressional districts are long and squiggly and funny-looking, but sometimes long and squiggly districts are more appropriate than a compact, block-like district. Geography is funny, and people arrange themselves in all sorts of funny ways. There's no reason why districts shouldn't represent this fact. Again, it depends what you want. Let's say you had a state with two small Democratic enclaves on either end—together comprising a fourth of the population—and a large Republican middle section. If you divvied the state up into four blockish districts, then you'd likely get four Republican representatives—two of them semi-competitive—whereas if you created a funny-looking district that encompassed the two pockets with a long corridor in between, then you'd get three Republican districts and one Democratic one, as the state's population might warrant. But those wouldn't be competitive!

Indeed, Iowa has sensible, block-like districts, drawn by computer, but it also has only one Democratic district out of five in the House, despite the fact that 49 percent of Iowans voted for Kerry in 2004. Ultimately, if you wanted to make the representation "fairer" for Democrats in Iowa, you might have to draw some funny looking districts that connected disparate "blue" parts of the state. Indeed, some political scientists believe that a focus on compactness will always hurt the party that relies heavily on the urban vote. But then elections might become less competitive! So it depends on what your goals are. In the abstract, if you think that a state with X percent of its population voting for a given party should have X percent seats in the House hailing from that party, then compactness won't always help.

Basically, it's not at all easy to figure out what the best way to do redistricting reform is, because reformers aren't always clear on what exactly they hope to achieve. Do they want more competitive districts, or do they want the representation to approximate the popular vote in a state? (These aren't always compatible goals.) Or do they want something else? At any rate, there are smart and not-so-smart ways of achieving each of these goals, but there's no reason to support "reform" in the abstract, especially if the goals are unclear, the plan seems poorly designed, and it looks strongly like a partisan power grab. And that's exactly what Schwarzenegger's plan looks like.

MORE: As several commenters point out, the best way to have competitive and proportional elections would be to turn California into a single district and elect at-large candidates (via party lists, or the single-transferable vote, or what have you). I agree completely, although this seems difficult to pull off in practice, since U.S. voters seem to enjoy having "local" representatives.

EVEN MORE: See Mark Kleiman for some numbers (down at the bottom).