During a discussion on whether the electoral college is still a good idea, someone brought up the point that it’s possible to win the electoral vote but lose the popular vote, and pretty badly at that.
So I wrote a Perl script that used evolutionary computation to try to produce the most skewed electoral map possible. Here’s what it came up with:
(click to embiggen)
In this scenario (state-by-state results here), the blue candidate wins the blue states by 50% of the popular vote plus one, while the red candidate wins the red states with 100% of the popular vote (imagine Florida in 2000, times 37). Also, the map doesn’t show it, but in this scenario Nebraska split its electoral votes 4 blue and 1 red.
The net result is that the blue candidate wins the election with 270 electoral votes, but 22.1% of the popular vote.
Note that the red candidate wins nine of the top eleven states with the most electoral votes (CA, TX, NY, FL, IL, OH, MI, GA, NC) while blue only carries PA and NJ. These are the states where you can lose lots of popular votes and still win the electoral votes. I.e., the states with the highest electoral vote/popular vote ratio (that is, the number of electoral votes the state gives you, divided by the number of people who voted for you)
Likewise, the blue candidate won 12 of the 13 states (plus DC) with four electoral votes or fewere: WY, DC, VT, ND, AK, SD, DE, MT, RI, HI, NH, and ID, leaving only ME for the red candidate. These are the states with the highest electoral vote/electorate ratio (that is, the number of electoral votes the state gives you, divided by the number of voters in the state).
I’m not sure what this means. On one hand, the small and/or empty states have a disproportionate effect in that they get more electoral votes per unit of population: Wyoming has one congressional districts’ worth of people, but gets three times that many electoral votes. California has 53 congressional districts’ worth of people, but only gets 4% more electoral votes.
On the other hand, with populous states, a PR investment has a larger payoff: if California has 12 million voters, then you only need to convince 6,000,001 of them to vote for you to get all 55 electoral votes. Whereas if you got 12 million voters’ worth of smaller states (say WY, DC, VT, ND, AK, SD, DE, MT, RI, HI, NH, ME, ID) and won 6,000,001 votes, you’d only get 44 electoral votes.
At any rate, it seems that the US electoral system is due for an overhaul. The problem is how to fix it.
One nice thing about the electoral college (and perhaps its only virtue) is that it tends to compartmentalize problems. Recall the 2000 election, in which the popular vote had George Bush and Al Gore within 0.5% percentage points of each other, and there was that massive recount hoopla in Florida.
Imagine if the election were decided purely by popular vote, and there were a 1% margin of error in the vote-counting process. There would have had to be a massive nationwide recount (Florida times 50) to determine the outcome of the election. With the electoral college system, however, even a 5% error rate wouldn’t have changed how California voted, so there was no need to have a recount there. This is, I think, a feature worth preserving.
Perhaps one compromise would be to have one electoral vote per congressional district. This means that recounts would be limited to individual districts where the vote was close.
However, a number of states are gerrymandered, which has the effect of giving one party a disproportionate number of seasts in Congress, or of increasing incumbents’ margin of victory. It would be nice to introduce districting reform, so that districts encompass more “natural” areas.
Of course, a side effect of such reform would be to reduce those margins of victory that were pumped up through gerrymandering, and thus lead to an increase in the number of districts with close elections, and hence recalls.
Summary: politics is a hairy problem.