So I found this paper by Gelman, King, and Boscodarin (1998), where they simulate the electoral college using models fit to previous US elections, and find that the probability of a decisive vote came out between 1/(3 million) and 1/(100 million) for voters in most states in most elections, with most states lying very close to 1/(10 million).
So… what you’re telling me here is that, in America, different people’s votes count for different amounts. If I have Tom, whose vote has a one-in-a-hundred-million chance of affecting the election; and John, who lives in a different state and whose vote has a one-in-three-million chance of affecting the election, then won’t the average politician completely ignore Tom’s state in order to concentrate more on influencing John’s vote?
This seems backwards.
It also implies that Tom’s state is going to end up with any messy industries that have to go somewhere, but that no voter wants in their back yard.
then won’t the average politician completely ignore Tom’s state in order to concentrate more on influencing John’s vote?
Yes, that does happen to some extent.
There are a number of factors influencing these decisions, but the main counterpoint to the idea of popular vote for President is to consider states to be relevant entities. The Federal government was originally conceived similarly to the EU government—an association of states. And some would say that France should have just as much say as Germany in how the EU is run, even if Germany has twice the population. Thus, the German voters in that scenario would each have half as much say as the French.
The way it actually pans out, the US House of Representatives roughly corresponds to equal representation by population, while the Senate has 2 senators per state regardless of population, and then the Electoral College has one Elector for each Representative and Senator, making it a compromise between the two approaches.
The way it actually pans out, the US House of Representatives roughly corresponds to equal representation by population, while the Senate has 2 senators per state regardless of population, and then the Electoral College has one Elector for each Representative and Senator, making it a compromise between the two approaches.
Okay, if the states are considered separate entities, then that doesn’t look too unfair. (Though I do think that that premise is no longer valid). Given the technological limits on long-distance communication that were present at the time that America was founded, it’s quite a sensible system—for that time. With modern technology, and considering what America has become in the meantime (more like one country than a collection of individual states), I don’t think that the system still remains as valid today.
Ideally, the Electors should then spend their votes in the same proportion as the voters: if 53% of voters vote for candidate A, then 53% of Electors (or as close as rounding errors will allow) should vote for candidate A. Having done a bit of research, though, I find that it does not appear to work that way—it seems that a state tends to spend all its electors on the candidate that got most of the vote, even if it’s only 53% of that vote. Fixing that would seem to me, naively, to result in a choice that more closely represents the choice of the American people in aggregate.
it seems that a state tends to spend all its electors on the candidate that got most of the vote
In principle, electors are able to vote for whoever they want. In practice, their behavior is governed by state law, which for about half of the states basically requires the electors to vote for whichever candidate won the majority of the vote.
Maine and Nebraska split their electors similarly to your suggestion.
One quirk of the electoral system means that states have strong incentives not to split their electors: a state that allows a split electoral slate will likely be swinging only a fraction of the electoral votes that a state with a unified slate would. The number of electoral votes a candidate is expected to get per unit campaign effort, therefore, is much lower in such a state. Since there’s an element of competition in the electoral system (states want attention to their citizens’ unique needs, and if possible promises to address them), it’s in their interests to avoid such a situation.
...so the actions of the Electors seem to be a 50-way Prisoner’s Dilemma, with a few strange little quirks.
Consider. The President, and the overall U.S. government, has a certain amount of attention (and budget) to give to each of the 50 states. In general, being politicians, we can assume that they will give the greatest consideration to those states which are most likely to affect the next election.
If all the states cooperate by splitting their Electors, then the President’s attention and budget will need to be split reasonably evenly; a few more votes in any single state could give him an additional Elector, and decide the next election.
However, any state can choose to defect, by putting all their Electors behind the majority candidate. Now the President will spend more budget on, and pay more attention to, that state, because if he loses votes there he could lose a lot of Electors. So it’s to the individual benefit of each state to defect. Since the President is paying more attention to the defecting states, however, the cooperating states get less attention.
But if they all defect in such a way, or even if a majority of them do, then the end result is that the voting within each state is only considered between the two leading parties; third- or lower candidates have no chance of getting Electors, and people may vote for one party entirely to keep the other party out. This feeds into the two-party swindle as well, which is bad for everyone. Another effect of everyone defecting is that, suddenly, the “swing states”, where the voting could go either way, become dramatically more important; for those states, the “everyone-defects” case is actually preferable (in terms of attention given) to the “everyone-cooperates” case, so they have even more cause to defect. (Note that the swing states still suffer from the two-party swindle).
Hmmmm. I don’t think America is going to be able to escape the two-party swindle until it gets rid of the electoral college—and I don’t think that’s going to happen, because the people who can get rid of it are in the position of gaining the greatest advantage from not getting rid of it.
That would solve a lot of the problem, yes. If it gets the 50%-of-the-Electors that it needs, of course.
...
Come to think of it, that looks like a simple change that seems likely to have far-reaching beneficial effects on the future, at least in the medium term. Is there anything I (or anyone else reading the replies to this comment) could do to make it more likely that it would be more widely adopted?
And does the answer to the above question change when I add the data point that I am not American?
Well, everything I said in the grandparent would still apply if there were more than two major political parties in the US, or if electoral votes were tabulated in a different way (a ranked-preference voting system, say). There seems to be a loose consensus that first-past-the-post election schemes tend to lead to two-party systems, but I’d expect the presence of an electoral college to have a minor if any effect.
(The last time this was significant relative to the two-party system was probably the election of 1992, when Ross Perot swung just under 20% of the popular vote as an independent but carried no states.)
Thinking about it, I think that the main effect of the electoral college is to amplify the two-party-ism of the first-past-the-post voting system (by making it that a person with less than 50% of the votes can beat a person with more than 50% of the votes).
I also think that, from the outside, removing the electoral college looks like an easier fix than changing the first-past-the-post nature of the vote.
Yes, first-past-the-post is a strange system for implementing a national preference. But it functions to preserve the power of “swing” states—and states generally. And preserving the power of the states is the purpose behind a lot of the structure of the US Constitution—otherwise, two Senators per state regardless of state population makes very little sense.
If I have Tom, whose vote has a one-in-a-hundred-million chance of affecting the election; and John, who lives in a different state and whose vote has a one-in-three-million chance of affecting the election, then won’t the average politician completely ignore Tom’s state in order to concentrate more on influencing John’s vote?
Diminishing returns come into play. There’s only so much effort it’s worth putting into the most important state before the next most important becomes marginally equal.
It also implies that Tom’s state is going to end up with any messy industries that have to go somewhere, but that no voter wants in their back yard.
Diminishing returns come into play. There’s only so much effort it’s worth putting into the most important state before the next most important becomes marginally equal.
...true. One should probably expect thirty-three times as much effort going to John’s state as to Tom’s state.
That does look armchair-likely. Does it happen?
I don’t know, I live on the wrong side of the Atlantic to have personal experience.
It also implies that Tom’s state is going to end up with any messy industries that have to go somewhere, but that no voter wants in their back yard.
Congressional elections work much differently than presidential elections. My vague impression is that nuts-and-bolts level concerns like what industries go in what state are handled by Congress, not the president.
Congressional elections work much differently than presidential elections. My vague impression is that nuts-and-bolts level concerns like what industries go in what state are handled by Congress, not the president.
No, actually, the federal government doesn’t do all that much to determine what industries go into which states. Mostly its private decision making. Businesses are generally free to locate their operations or management wherever they want to. Of course, some businesses, like gambling, are illegal in some localities or they can get more favorable treatment in certain locations rather than others, but typically the federal government doesn’t make those decisions.
While that’s true in most cases, federal contracts can make a pretty big difference in some industries, particularly those connected with infrastructure or defense. (All those highway projects with signs saying “Funded by the American Recovery and Reinvestment Act” form a recent and salient example.) This is a semi-adversarial process, since allocation of contracts is zero-sum and Congresscritters have incentives to funnel them to their own states or districts; “should” isn’t the only concern and may not even be a primary one.
On the other hand, these incentives are tied more to regions than to parties, so choice of candidate isn’t likely to make a big difference in how they’re executed.
So… what you’re telling me here is that, in America, different people’s votes count for different amounts. If I have Tom, whose vote has a one-in-a-hundred-million chance of affecting the election; and John, who lives in a different state and whose vote has a one-in-three-million chance of affecting the election, then won’t the average politician completely ignore Tom’s state in order to concentrate more on influencing John’s vote?
This seems backwards.
It also implies that Tom’s state is going to end up with any messy industries that have to go somewhere, but that no voter wants in their back yard.
Yes, that does happen to some extent.
There are a number of factors influencing these decisions, but the main counterpoint to the idea of popular vote for President is to consider states to be relevant entities. The Federal government was originally conceived similarly to the EU government—an association of states. And some would say that France should have just as much say as Germany in how the EU is run, even if Germany has twice the population. Thus, the German voters in that scenario would each have half as much say as the French.
The way it actually pans out, the US House of Representatives roughly corresponds to equal representation by population, while the Senate has 2 senators per state regardless of population, and then the Electoral College has one Elector for each Representative and Senator, making it a compromise between the two approaches.
Okay, if the states are considered separate entities, then that doesn’t look too unfair. (Though I do think that that premise is no longer valid). Given the technological limits on long-distance communication that were present at the time that America was founded, it’s quite a sensible system—for that time. With modern technology, and considering what America has become in the meantime (more like one country than a collection of individual states), I don’t think that the system still remains as valid today.
Ideally, the Electors should then spend their votes in the same proportion as the voters: if 53% of voters vote for candidate A, then 53% of Electors (or as close as rounding errors will allow) should vote for candidate A. Having done a bit of research, though, I find that it does not appear to work that way—it seems that a state tends to spend all its electors on the candidate that got most of the vote, even if it’s only 53% of that vote. Fixing that would seem to me, naively, to result in a choice that more closely represents the choice of the American people in aggregate.
In principle, electors are able to vote for whoever they want. In practice, their behavior is governed by state law, which for about half of the states basically requires the electors to vote for whichever candidate won the majority of the vote.
Maine and Nebraska split their electors similarly to your suggestion.
One quirk of the electoral system means that states have strong incentives not to split their electors: a state that allows a split electoral slate will likely be swinging only a fraction of the electoral votes that a state with a unified slate would. The number of electoral votes a candidate is expected to get per unit campaign effort, therefore, is much lower in such a state. Since there’s an element of competition in the electoral system (states want attention to their citizens’ unique needs, and if possible promises to address them), it’s in their interests to avoid such a situation.
...so the actions of the Electors seem to be a 50-way Prisoner’s Dilemma, with a few strange little quirks.
Consider. The President, and the overall U.S. government, has a certain amount of attention (and budget) to give to each of the 50 states. In general, being politicians, we can assume that they will give the greatest consideration to those states which are most likely to affect the next election.
If all the states cooperate by splitting their Electors, then the President’s attention and budget will need to be split reasonably evenly; a few more votes in any single state could give him an additional Elector, and decide the next election.
However, any state can choose to defect, by putting all their Electors behind the majority candidate. Now the President will spend more budget on, and pay more attention to, that state, because if he loses votes there he could lose a lot of Electors. So it’s to the individual benefit of each state to defect. Since the President is paying more attention to the defecting states, however, the cooperating states get less attention.
But if they all defect in such a way, or even if a majority of them do, then the end result is that the voting within each state is only considered between the two leading parties; third- or lower candidates have no chance of getting Electors, and people may vote for one party entirely to keep the other party out. This feeds into the two-party swindle as well, which is bad for everyone. Another effect of everyone defecting is that, suddenly, the “swing states”, where the voting could go either way, become dramatically more important; for those states, the “everyone-defects” case is actually preferable (in terms of attention given) to the “everyone-cooperates” case, so they have even more cause to defect. (Note that the swing states still suffer from the two-party swindle).
Hmmmm. I don’t think America is going to be able to escape the two-party swindle until it gets rid of the electoral college—and I don’t think that’s going to happen, because the people who can get rid of it are in the position of gaining the greatest advantage from not getting rid of it.
Indeed, it seems we have a 50-way Prisoner’s Dilemma. May I interest you in a conditional public precommitment solution?
That would solve a lot of the problem, yes. If it gets the 50%-of-the-Electors that it needs, of course.
...
Come to think of it, that looks like a simple change that seems likely to have far-reaching beneficial effects on the future, at least in the medium term. Is there anything I (or anyone else reading the replies to this comment) could do to make it more likely that it would be more widely adopted?
And does the answer to the above question change when I add the data point that I am not American?
Well, everything I said in the grandparent would still apply if there were more than two major political parties in the US, or if electoral votes were tabulated in a different way (a ranked-preference voting system, say). There seems to be a loose consensus that first-past-the-post election schemes tend to lead to two-party systems, but I’d expect the presence of an electoral college to have a minor if any effect.
(The last time this was significant relative to the two-party system was probably the election of 1992, when Ross Perot swung just under 20% of the popular vote as an independent but carried no states.)
Thinking about it, I think that the main effect of the electoral college is to amplify the two-party-ism of the first-past-the-post voting system (by making it that a person with less than 50% of the votes can beat a person with more than 50% of the votes).
I also think that, from the outside, removing the electoral college looks like an easier fix than changing the first-past-the-post nature of the vote.
Yes, first-past-the-post is a strange system for implementing a national preference. But it functions to preserve the power of “swing” states—and states generally. And preserving the power of the states is the purpose behind a lot of the structure of the US Constitution—otherwise, two Senators per state regardless of state population makes very little sense.
Diminishing returns come into play. There’s only so much effort it’s worth putting into the most important state before the next most important becomes marginally equal.
That does look armchair-likely. Does it happen?
...true. One should probably expect thirty-three times as much effort going to John’s state as to Tom’s state.
I don’t know, I live on the wrong side of the Atlantic to have personal experience.
Yep, you read right, and yeah, it’s nuts.
Congressional elections work much differently than presidential elections. My vague impression is that nuts-and-bolts level concerns like what industries go in what state are handled by Congress, not the president.
No, actually, the federal government doesn’t do all that much to determine what industries go into which states. Mostly its private decision making. Businesses are generally free to locate their operations or management wherever they want to. Of course, some businesses, like gambling, are illegal in some localities or they can get more favorable treatment in certain locations rather than others, but typically the federal government doesn’t make those decisions.
Yep, I was being facetious.
While that’s true in most cases, federal contracts can make a pretty big difference in some industries, particularly those connected with infrastructure or defense. (All those highway projects with signs saying “Funded by the American Recovery and Reinvestment Act” form a recent and salient example.) This is a semi-adversarial process, since allocation of contracts is zero-sum and Congresscritters have incentives to funnel them to their own states or districts; “should” isn’t the only concern and may not even be a primary one.
On the other hand, these incentives are tied more to regions than to parties, so choice of candidate isn’t likely to make a big difference in how they’re executed.