Sunday, November 13, 2016

The Electoral College and National Popular Votes, Part II

In Part I, I noted that votes are not actually commensurable across state lines. When we talk about a 'national popular vote' the only thing we can be talking about without committing an error of equivocation is what the national popular vote would have been, taking state popular vote numbers as evidence. When we do this, we lose certainty and run the risk of introducing false precision.

A good example of how this false precision is misleading is seen by considering the fact that voters vote on the assumption that their vote is for how state electoral college votes will be distributed, as Jonathan Adler has noted. Suppose that John is a Republican in a blue state that is very, very blue -- it always votes overwhelmingly for the Democratic candidate. John might think a number of very different things in this context. He might think that he needs to get out and vote precisely to represent the fact that there are Republicans in this Democratic state, but he also might think that there is no point in doing so -- he will have to vote already knowing that he and other Republicans are going to be outvoted. Suppose he concludes the latter and just stays home. When we look at the numbers of actual votes in a state, do they tell us anything about John and people like him, or how they would vote if they were voting in a national popular vote system? Not at all.

In addition to this, how a candidate campaigns is obviously going to be determined by how the election itself is structured. Different structures, different campaign strategies. And every election structure leads to something being a 'swing' target, some possible focus of campaigning that has disproportionate effect in context; obviously in the Electoral College, organized by states, the target will be the primary population centers of states with split populations. And campaigning does have an effect. While it does not manufacture new numbers out of nothing, it does consolidate votes and draw some who are only leaning a certain direction.

The relevance of these points to our case is that there is one state of the Union that is so locked down by one party that it returns overwhelmingly lopsided numbers -- California, a state so Democratic that the only Republican candidates on the ballot are often the Presidential ticket, and where the normal voting reality is not whether you vote Republican or Democratic but which Democrat you prefer. Republicans very often just don't vote. What is getting called the 'national popular vote' is, besides being a fictional number made by adding incommensurable units, is not yet completely counted, and will not be for a while, but if one looks at the numbers as we have them currently, one notices that Clinton's current advantage in this (again, fictional) number is due overwhelmingly to California. If you add up all the other forty-nine states, Trump comes out definitely ahead, the result of being at least slightly ahead in a lot of different state popular votes, and only barely behind in several others; add in California, and we discover that Clinton is supported so lopsidedly in California that it counterbalances Trump's lead if you counted all the other states put together.

This is going to be a fairly general thing -- since the election is not, in fact, structured as a national popular vote but by states, you'll get a higher 'national popular vote' number by locking down one populous state than you will by being consistently, but not dominatingly, more popular over a larger population. This is why the 'national popular vote' has tended to favor Democrats -- the Republicans have not managed to lock down any state as conclusively as the Democrats have done the big-population state of California. Would a real national popular vote also favor Democrats? Perhaps, but what we usually call the 'national popular vote' number is not really evidence for this -- the advantage is so slight in close contests, relative to the larger population, that it easily falls within the large region of certainty I noted in the previous post. And in an actual national popular vote, the numbers for every state would change -- sometimes only slightly, no doubt, but in other cases more dramatically. This is intensified by the fact that campaigning would inevitably change, as well. There are too many uncertainties to be sure of anything at that point.

Prokop tries to make this campaign-uncertainty out to be a disadvantage for the Electoral College:

Second, there’s swing state privilege. Millions of votes in safe states end up being “wasted,” at least in terms of the presidential race, because it makes no difference whether Clinton wins California by 4 million votes, 400,000 votes, or 40 votes — in any scenario, she gets its 55 electors. Meanwhile, states like Florida and Ohio get the power to tip the outcome just because they happen to be closely divided politically.

This is apparently plausible, but also somewhat illusory. Whether a state is a swing state already depends on how the people in the state are voting; it is an emergent property of voting, not a pre-existing one. California's registered voters are about 44% Democratic and 27% Republican; independents lean in roughly the same proportion, although with a slightly stronger Republican lean than the population as a whole. If all Republicans voted in an election, this would be outmatched by all Democrats voting; but if all Republicans were voting, a Republican running against an unpopular Democrat (thus drawing away some Democrats and pulling a lot of independents) could do it. It would just be a matter of catching the right election at the right time. But the difficulty a consistent minority has is precisely this, to turn up in the right numbers at the right time, and where that cannot be organized by the charisma of the candidate, the only way one can get close to that is if people vote and keep voting. This encourages candidates to run, which they won't if there's no chance at all. Contrary to occasional news reports mid-race this year, there was no real chance of Texas flipping from Republican to Democrat, but the fact that Democrats did massively better in Texas this year than they usually do may well fire up Texas Democrats for the future -- and that is itself a factor in whether Texas ever becomes a swing state. It does matter whether a race is won by 4 million or 40 votes -- that's exactly what makes the difference between swing states and non-swing states. If Clinton had only won California by 40 votes, it would have become a swing state, and the Republicans would be on the warpath to capture it in the next several elections.

In reality, the entire point of holding an election is for some people to outvote some other people, and in any election there are always going to be some people who are more likely to switch their apparent vote than others. What the Electoral College does is make this unstable -- there are slow-shifters and swift-shifters, but all the states shift around the swing-state sweetspot. In a national popular vote system, campaigns are not focused everywhere; the swing-targets tend to be the same large population centers every election. (Any large group of persuadable voters will draw attention, whether in a city or not, but large groups of persuadable voters will tend in general to be in places where there are a lot of voters.) The Electoral College messes with that -- because there is no national election, only state elections, the swing-targets tend to be the largest population centers in the most populous states where the race is closest, and that varies over time according to so many factors that it is not always predictable. (As we saw in this election.)

There is more to be said on this subject of the Electoral College and the 'national popular vote', in a future post.