Preventing the chaos of super-close elections


When elections are close, they get chaotic. If the flip of a single vote, at the tie-point, can cause a massive change, like who runs a country, things can go nuts. People will do everything, from legal battles all the way up to the supreme court, to voter suppression, to voter fraud, to fake claims of voter fraud, all to move the needle a tiny bit around that tipping point.

It's a major flaw in many voting systems. After the 2000 election in Florida, I proposed a simple fix to the problem of that election. States like to hand over their electoral votes on a "winner takes all" (WTA) basis, which magnifies the chaos at the tipping point. Instead, they can still use WTA if the election is not close, but if it is close enough for a recount, say 0.5% difference between the contenders, they should instead allocate the votes one at a time, depending on where the result fits in the range from -0.5% to +0.5% for the winning candidate.

In other words, if a candidate wins 50.5% to 49.5% they get all the votes, and the loser nothing. This encourages the candidates to pay tons of attention to the state and offer it goodies, the reason states like WTA. In between, it's proportional.

In most states, this turns out to about 3,000 votes in the difference between the two candidates (which means one wins 1500 and the other loses 1500) to gain one more electoral vote. Let's say a state has 11 electoral votes, like Arizona. Leaving out the exact tie, if the margin of A to B is from +1 to +3000 votes, then A gets 6 and B gets 5. If it's from 3001 to 6000 votes, A gets 7 and B gets 4 -- all the way up to 15,001 then A gets 11 and B gets 0 from then on.

(If the number of electoral votes is even, then all the way from -1500 to +1500, including a tie, the votes are split evenly. In a tie with an odd number, drawing cards can be used but it only wins a single electoral vote, not the whole state.)

This achieves several great goals. First of all, if a state is that evenly divided, it's actually a much better reflection of the will of the people to split the votes rather than hand them all over. At the same time, the candidates are fully motivated to campaign in the state and offer it goodies because they only have to win by 0.5% to get the full prize. Of course, many people think WTA is bad -- and it is -- but states and parties usually won't give it up, though Maine and Nebraska did. We try to work with what a state would actually do.

The big gain, though, is that it takes away the chaos. With a range of 3,000 votes, most recounts, lawsuits and supreme court challenges make no sense. They can only win you a single electoral vote, and barely that. Unless the election in the rest of the country is a tie, they become pointless. Why hold a huge and expensive fight, debate each hanging chad, track down ever voting irregularity if it can't do anything for you? You might do it for show, but everybody would know that. If there were a massive problem, the kind that could affect tens of thousands of votes, you would still go after it, as you should. Recounts and lawsuits almost never change totals by over 1,000 votes.

Of course, in very rare circumstances, if the other states and the partial allocation of your state leave the electoral college in a tie, then all the chaos could still break loose. But this should be a much more rare condition. There are solutions for that, but they are dramatic and can't be enacted by single states. I believe the idea that when one party gets 50%+1 and the other gets 50%-1 that this says the first side gets to tell the other side what to do is crazy. In that situation, power should be shared -- it's what the people want when they are so evenly split. Many countries of the world have 4 or more parties, and nobody gets a majority so they have to form coalitions and share power a bit. The USA is different from most of the world and has this flaw.

While yes, in this election people were briefly worried that Biden would win only 270 EVs and Trump 268, and so we would have had a chaotic situation. In reality, Biden's margin was strong enough. Under this system, Arizona would give 9 EVs to Biden and 2 to Trump (this could change slightly as the count is not done) and Georgia would give 13 to Biden and 3 to Trump. Biden's majority would remain strong. But all the recounts and lawsuits would be pointless, and everybody not posturing could not deny they had no shot of changing the final result. Yes, Trump is posturing but it would be even more obvious.

Yes, if you're a foolish partisan, you may notice that this system, in 2000, would have given Gore the victory because he was well ahead of Bush not counting Florida, with 266 votes. He would have won at least 4 of Florida's under this rule -- the recounts would have mostly shifted it between winning 14 and 15 of them. This is the fair result if you consider that the nation favoured Gore in the college without Florida, and Florida was a tie. (A true tie from a scientist's viewpoint, where the difference was within the margin of error of the counting system and legal process.) One could not predict how such a result would go, and it would be a shame if Republicans were to reject this system because it would have hurt Bush. The next time it could easily go the other way, but more likely it doesn't benefit either side.

Thus we would probably not have this chaos problem again. It would take an incredibly extreme event to bring an election to the courts again, and certainly not the supreme court. Bush v Gore was not that court's finest hour, I think even the judges who voted for it, would agree.

A state could use my simple "3,000 votes per EV" stair-step function, or they could devise another function, such as a sigmoidal function with a wider margin. (You don't want to make the space between EVs too small since you don't want somebody to think they can win 2 by fighting, just 1.) If you make the margin wider, like 5,000 votes per EV or 10,000 votes, then it's a little more likely your state will hand out some split of the votes in a close race. Many would feel that this is actually better -- a touch closer to what the voters wanted. Of course, handing out the votes by district as Nebraska and Maine do is also closer to what the voters want, as is a fully proportional allocation. But no safe state will be likely to do that -- it's just handing some "free" votes to their opposition as they would see it. And they still want the candidates to make it rain in their state for the chance at all the marbles.


Cool idea. I can't really think of a reason not to do this.

One unintended consequence would be that it becomes even less likely that your one vote will make a difference. But that's probably not really a bad thing.

It actually increases the chance your vote will make a difference, but only of one electoral vote. It's still unlikely, but many times more likely. As to your vote flipping the whole election, that's incredibly unlikely already and remains incredibly unlikely with this system. It might be a bit less. I am not sure motivation changes much. However, there is a big change in motivation over cheating and disputing results.

It actually increases the chance your vote will make a difference, but only of one electoral vote.

By "make a difference" I meant "change the winner."

If you live in a swing state, the chance that your vote will change the election outcome goes down tremendously.

If you live outside a swing state, it stays at zero.

In WTA, if your state has 10 EV, and the election is within 10 EV, then your chance of changing the winner depends on the probability of your state being tied except for your vote, which is very small. Under the proposed system, it should be the same, in that your candidate needs some number of votes N (between 1 and 10) to win, and the chance that you will cast the vote that flips it from N-1 to N should be the same, no? (presuming a similar distribution of probabilities for each vote total around the given point.)

However, there may be many more ways to get this situation because other states are now also returning partial tallies. (This year, Georgia and Arizona would have returned partial tallies.) I have not done the math but it's not obvious that it reduces the chance that you make the difference.

Right now, your single vote has to change the outcome of the state, and your state has to change the outcome of the election. The latter is not common, but it does happen.

Under your proposed system, your vote still has to change the outcome of your state. This is slightly more likely, but still incredibly unlikely. But also the electoral college needs to have one candidate only one vote from having or losing a majority. This has never happened since the passage of the 12th Amendment.

NO variant of a single vote making the difference has happened, so that doesn't tell you much. It is highly unlikely. As far as I can tell, it is equally unlikely under both approaches.

I just explained why it is not equally unlikely.

Though I haven't done the detailed analysis. The reason I haven't done it is I don't think it's important. If people only voted out of hope/fear that their vote will be the one vote that flips it all, nobody would vote. Especially not in places like California where you are told there is zero chance of it happening, rather than epsilon in a swing state. Even if the epsilon is one in 150 million or less.

In a swing state with 10 EVs, this system makes it 10 times more likely your vote could be "the one" to change the EV count, as there are 10 thresholds now, not one. That's actually more empowering -- "your vote is needed because your state is very close and if you win us just one more EV it could make the difference."

But I continue to believe the probability is roughly equal. Take any election where a single state can decide it, like Florida 2000 with 29 EVs to hand out. Bush winning all 29 won it for him, but actually what he needed was 26 to win it, and in any situation there will be a magic number which turns out to be the number that flips the country. In the current system, Bush needed 50%+1 and that was the flip threshold where your vote could change it. In my system, Bush would have needed 50%+19501 to win, it just moves that threshold -- the probability that your vote is the deciding one is roughly the same.

Fortunately they are all extremely unlikely. Situations like Florida are super unlikely, but can still happen in the proposed system. However, what it tries to eliminate is not one voter flipping the result, but lawsuits and recounts flipping it. Maybe it doesn't even do that, though, it could just be that Florida was bad luck,

This time we had 2 states at the narrow margin but none remotely close as Florida. And that didn't stop the incumbent from trying to play games, though.

I agree with you that it's not important. I think I even said that initially.

Add new comment