Interesting that things have changed so quickly: seems like only a few years ago, the idea of somebody voting early via an absentee ballot, unless you were military or otherwise stationed overseas, was strange and unusual. Now, in some places, a majority of votes or all votes are cast early, sometimes by mail and sometimes in person like we now can do in DC — or at least try to.
Apparently, this trend of more and more early voting surprised the BOEE.
What will happen if the electricity is still out after Monday, all through Election Day, from this storm they are predicting? If the storm is as bad as they foretell, much of the northeast section of this entire country could be without power for weeks — just like after this summer’s famous Derecho, and also after last summer’s hurricane…. (Isabel?)
It could really very strongly alter the direction of the election. I hope that the next few weeks won’t make the Florida 2000 vote look marvelously fair.
Mathematicians and computer programmers do not generally make any guarantees about electronic voting, either. It can make it much, much easier to steal entire elections without being detected. Compared to truckloads of doctored ballots that have to be physically carried around and burned or erased and changed, electronic fraud is in many ways much easier to carry out. Electrons can easily change direction and change a bunch of 1’s into a bunch of 0’s, or vice versa, and no-one would notice at all. It happens all the time, all over the world, not just here. The fact that the trend in some areas is to have NO ALTERNATIVE to voting electronically is worrisome. There should be a way that things can be checked independently, because no single method can ever be completely trusted. Any testing or measurement method involves error of one sort or another.
Let me try to give an example. Suppose an imaginary state has exactly a million voters.
You agree? Exactly a million voters in this state?
No. That’s utterly impossible. There is no way on earth or anywhere else to know if you have exactly a million voters, or a few dozen more, or a few dozen less.
Why? Very simple: people DIE, at unpredictable moments. They move to other states. They move in from other places. They may forget to go to the voting booth. They may decide not to go vote after all. They may get sick. Who knows? That exact million number is a myth. It’s plus or minus a certain number, and I have no idea by how much. If you inserted an Radio-Frequency Identification Device into every single person it still wouldn’t solve the problem completely, because some of the RFIDs would fail completely every so often, or give out erroneous information, or be cut off by some accidental or purposeful action, or would not detect if someone was now dead, or sick, or moved away, or …
Perhaps this RFID is 99% accurate, which would be pretty darned good. But that still means that there is 1% error of some sort: that’s 10,000 people about whom there is no good data one way or the other — and you don’t know which ones. Maybe they voted, maybe they didn’t. You don’t know, because your system has failed, even though it was working as well as anybody could possibly make it.
Comes election day. Ballot boxes are opened, PLUS, the absentee mail-in ballots are opened, PLUS, the computer memory banks that hold the data in microscopic, intangible form, are accessed under hopefully very strict controls. Plus a number of voters marked their written ballots in such a way that you can’t easily understand what they meant — or at least, two observers with different biases might be very much inclined to read differently.
Let’s say that you end up with candidate A earning 495,012 votes, and candidate B getting 497,770 votes. The difference between them is only 1,758 votes. And there are 8,249 votes that are still being disputed. But that’s way less than any reasonable margin of error, so they do a recount. Suppose your next count, using slightly different methods, gets the result that candidate A was the victor this time, by 369 votes, but this time, there are 14, 299 challenged ballots. Do it again, and you get a different result. If it was Florida in 2000, the Supreme Court might step in and make the decision.
Perhaps they should have just had a run-off election, with the two top-grossing candidates only? Or else just have a coin toss? The latter is what I think John Allen Paulos would recommend.
So, as much as I enjoy bad weather (though I also like nice weather, too!), I hope that all of the forecasters I know and hear are wrong, and that all the models are, somehow, wrong.