A couple of years ago, there was a big to-do about Republicans using computer software to very aggressively create the best congressional gerrymanders possible—that is, to stuff most Democrats into just a few congressional seats and thereby dilute their representation. The art of the gerrymander had become an exact science. (Not that the traditionally practiced art was all that ineffective: After Texan Democrats pulled "the shrewdest gerrymander of the 90s," Republicans took 59% of the 2000 congressional vote but only won 43% of the seats. Which begins to explain why Texan Tom DeLay was so hellbent on shafting Lone Star Dems in the next decade.)
With most people predicting that the Democrats will win back the House today (and the smart money giving the GOP 9-1 odds), I wonder if this exquisite gerrymandering will bite the GOP in the ass. Consider this simplified example:
Say the country is split exactly 50-50 between purple and yellow. The purple team gets to draw the maps, so it makes 1/3 of the districts have 10% purple supporters, 1/3 of the districts have 55% purple, and 1/3 of the districts have 85% purple. So while they receive just 50% of the votes, they get a commanding 67% of the districts. (Time to whip up some amendments!)
But then imagine that purple starts a war that goes sour and they have a few scandals. Support for purple drops 6% across the board, so in the three districts they now have 4%, 49%, and 79% support. Although they get 44% of the votes, they grab only 33% of the districts.
And that's a risk of gerrymandering: when the tide goes out, it can go way, way out.
(I'm talking, by the way, only about risks as defined by one's self-interest, not risks to the polity or any high-minded ideals like that. Because really, it's politics.)