Lots of people have trouble with numbers. A good friend of mine is Chair of the Government Department at a major New York university. He is an extremely bright guy but he has trouble understanding numbers, especially probabilities. But I suspect that lots of bright Kossacks have similar issues. If you are not one of them, ignore this diary.
Nate Silver's Fivethirtyeight.com site is fantastic. (I've been following Nate's baseball predictions for years.) For many people, part of the problem with Nate Silver's approach is that it applies probability percentages to a field (polling) where percentages have long been used for voting totals. A 59%-41% forecast in an election race used to mean that one candidate was likely to get 59% of the vote. Those who been following elections for any length of time have internalized the fact that such a candidate would typically be a 99%+ favorite to win, probably a landslide.
However, when Nate says (as he currently does) that Romney is a 59%-41% favorite in Florida, that means that Florida is very, very close! And, indeed, his overall electoral vote forecast giving Obama a 77.4% chance of winning, while a lot better than it was a few weeks ago, is still too close for comfort. It means that Obama will lose 22.6% of the time, and in something as important as this election, that is quite terrifying to me. Those diarists who keep emphasizing GOTV are dead on the money.
Below the squiggle, I am going to try to provide a realistic sense of what Nate Silver's numbers at Fivethirtyeight.com are really telling us, to avoid overconfidence that could lead to disappointment.
I am a games player: chess, backgammon, poker. To survive in backgammon and poker, one has to have a good sense of probabilities and what they mean. So let's take a pair of dice and see exactly what percentages like those favoring one candidate or another mean.
There are 36 possible results from rolling a pair of dice: 30 where the numbers on the dice are different (6-5, 3-1, etc.) and six where the numbers are identical (double 1s, double 2s, etc.). Thus the chances of rolling any single number (1, 4, or whatever) on either die is NOT 1/6 (or 16.7%) but 11/36 (1-2, 2-1, 1-3, etc. plus 1-1) or 30.6%. If Nate gives a candidate a 70% chance of winning, that means such a candidate will only win if they can avoid rolling a 1 on the next roll. I'd hate to tell you the amount of money I have lost at backgammon in exactly that situation -- admittedly, less than half of what I have won, but it is still scary when there is a lot on the line. Try it -- take a pair of dice and roll them a few times and try to avoid any single number coming up. That number will show up far too often for comfort.
Right now, in the last Senate race, Nate has Heller ahead of Shelley Berkley 72%-28%. This is approximately 10/36 -- the chances of rolling a number on EITHER die but NOT on BOTH dice. Try that and see how often it happens: Berkley is definitely still in that race!
Nate's presidential race forecast is currently 77.4%-22.6%. This ain't bad -- it is almost identical to a situation of having eight losing dice rolls out of 36 (8/36 equals 22.2%). So now try rolling the dice and avoiding a 1 UNLESS the other die is either a 1 or a 2: you lose if you roll 1-3, 3-1, 1-4, 4-1, 1-5, 5-1, 1-6 and 6-1. Again, you will win a lot more often than you lose -- but if the stakes are high, you'll lose much too often for your palms not to sweat!
So what is a comfortable lead? Nate currently has my Senate candidate Elizabeth Warren up 95%-5%, and I'm pretty darned comfortable there (basically, Elizabeth wins unless she rolls 1-2 or 2-1). McCaskill in Missouri (88%-12%) has to avoid 1-2 and 2-1 as well as 1-3 and 3-1 -- pretty good, but I've lost a lot of money in situations like that, too.
What it all means is: Nate is telling us that things are looking good for Obama and most of the Dem senatorial candidates. But nothing is in the bag! This election is close, and if we take anything for granted, we are asking for a huge disappointment.
Now: GOTV!