Mary’s prayer: The odds

At FiveThirtyEight Harry Enten provides the odds that Republican Rep. Bill Cassidy will defeat incumbent Democratic Senator Mary Landrieu in the runoff election for her Senate seat today: 99.8 percent, based on “the FiveThirtyEight model.” I know this is science, but where does the 0.2 percent come from, and how can we verify it?

Perhaps we will learn another day. Today we can savor FiveThirtyEight’s exploration of the possibility that Landrieu will suffer “a top 10 historic defeat.” The odds are fair (my words, not theirs): “According to data compiled by Geoffrey Skelley of the University of Virginia’s Center for Politics, Landrieu only needs to lose by a little more than 20 percentage points to earn all-time infamy.”

My prayer is that Landrieu makes history today by joining the list of top 10 losers. She richly deserves it, and there is a local Minnesota angle here as well. To make the top 10 of incumbent senatorial losers, she would have to lose by more than 20.3 percent. Were she to do so, she would displace Minnesota Republican Joseph Ball, who lost by that margin in 1948. FiveThirtyEight doesn’t say to whom Ball lost, but you can probably guess: Hubert Humphrey.

When it comes to predictions, I’m with the English novelist George Eliot, who advises in the omniscient narrative voice of Middlemarch that prophecy is the most gratuitous form of human error. Yes, the forms of human error are many, as are the sources of gratuitous human error. Mankind is fallible and prone to error. But among the sources of gratuitous human error, dear readers, prophecy is the most gratuitous. (UPDATE for the commenters below: Typographical errors figure in somewhere as well.)

Throwing caution to the wind, I’m going one better than FiveThirtyEight and predicting without qualification that Landrieu loses today, to a metaphysical certainty. I should leave room for the probability that vote fraud will save the day for her — it must exceed 0.2 percent in Louisiana all by itself — but what’s the point? I’ll leave it for an after action report, if necessary.