The Problem With Claiming Issues With Vote Counting
As I mentioned in my last post, a recount is unlikely to change anything. And, I felt it would be fun to illustrate this is a bit more depth.
And to do that, I decided a good first thing to point out is what a recount does and does not do.
Firstly, and I think this is crucial; a recount does NOT tell you how many counting errors were made. The only way this could be uncovered would be to do something like watch a video of the counting process and catch the errors as they are happening. Or, if the votes were audited in some way for comparison.
All a simple recount tells you is (presumably) the amount by which the prior count was off. Of course, if the first count was fallible, then the second count likely is as well. But, we can ignore that for the sake of this.
To illustrate this. If you had 1 million ballots, and the first count said that there was a tie and a recount said that candidate A won by 100 hundred vote. How many ballots were counted incorrectly? The tempting answer is 50 (or 100 if you forgot how math worked). But, the true answer could be as many as all but 50. Or, as many as 999,950 to as few as 50. Now, statistically, without meddling, the extremes are unlikely. But, you start to get a sense of the scope of the problem.
Basically, the problem is this. The closer an election is, the more ways the ballots could be counted incorrectly which would yield the same, or similar results. If a ballot has 2 options and I get only 1 vote out of say, 4, there are 4 ways the vote could have been counted to arrive at this number.
1 of which is the actual way, and 3 incorrect ways out of a total of 2^4 or 16 possible outcomes as each voter has 2 choices. However, if I got 2 votes out of 4 in this scenario, things change. There is still only 1 TRUE result; the way it actually happened. But, the number of ways the vote could be counted to arrive at the same number are higher. The number of incorrect circumstances which yield the same total are higher. See the combinations below.
Highlighted in yellow are the combinations where there are 2 votes each. Notice, there are 6 possible ways to arrive here, more than the 4 for the other outcome. And this pattern grows as the number of votes grows. Numbers closer to the middle are more likely than numbers toward the edges.
Given that mistakes are likely random and as likely to go to one candidate or the other, the closer the two candidates are to a tie, the less likely it is that a recount will substantially shift the results. In fact, the problem actually grows exponentially. Which makes every 1 vote changed exponentially less likely to have happened than the last.
This is why a recount in a state of million in the last election resulted in just 102 votes changing hands. Yes, there are millions of votes. And yes, there were probably a LOT of counting errors. And while the candidates SEEMED close together that we would like to THINK it likely that a recount could change something, the fact is, even that 102 votes changing was quite a huge number in this sense, while being nowhere near enough to have flipped the result.
Comments
Post a Comment