On to today's topic. Will "Brett" be confirmed? Probably. How likely? Here's the page with all the betting over at PredictIt. As of typing, shares of Kavanaugh's confirmation are trading at 55 cents on the dollar, with shares of Murkowski and Collins each trading at 53 cents on the dollar.
Huh?
There is always some weirdness with the aggregation of the betting, and if there weren't limits on how much you could trade, one could take advantage of arbitrage. The probability of "Brett" being confirmed, mathematically, cannot be higher than both the probability that Collins and Murkowski vote yes without some weird cross-party coalitions, and that ain't gonna happen. No, the simplest, "reduced form" way to think of this is as follows. Every Republican not named "Susan Collins" or "Lisa Murkwoski" has a probability of supporting Kavanaugh of 1, unless another shoe drops. There is currently some question of another allegation in Montgomery County, and at this point, you can't completely ignore the possibility of things getting even crazier, but I have a really hard time seeing any Republican not named "Lisa Murkowski" or "Susan Collins" voting no. They've said, quite directly, that they don't care what Ford or Ramirez have to say, they support Kavanaugh. What more do you need to assess probabilities?
Then there's Collins and Murkowski. If both flip on "Brett," what happens? In principle, if a couple of Democrats, like Manchin and Heitkamp flip on their party, that cancels out and "Brett" still gets confirmed, but as I said, that's unlikely, and that's where I'm going here. These probabilities are not independent.
"Brett" gets confirmed if there are 51 votes, or, 50+Pence. The mathematical way to solve this, then, is to figure out how many different ways there are to reach 50, right? When they are probabilistic events, you've got a multiplication problem, essentially, but you can sort of visualize how it works, right? The problem is that the factors that would push Collins, Murkowski and Manchin are potentially similar. If something were to happen tomorrow that made Murkowski sour on "Brett," she'd probably take Manchin with her. Maybe even drag that useless idiot, Collins, with her. (Their voting scores are quite similar using NOMINATE). These people are not rolling separate dice in their heads.
Here's how it works. If Murkowski were to vote no, it would be because Kavanaugh screwed up under questioning tomorrow and said something sufficiently obviously rape-y to make her unwilling to vote yes. That event, plus Murkowski's influence, takes Manchin and Collins out of the yes column. You can't just say, "well, Murkowski has a 53% chance of voting yes, and Collins has a 53% chance, so there's a 28% chance of them both voting yes, confirming 'Brett.'" How much of the probabilities are based on the potential for intervening events, and how much is based on current indecision? Are these probabilities correct?
Right now, without another intervening event or particularly poor showing by "Brett," I have a hard time seeing how the Republicans don't confirm him. Murkowski has given several indications that she doesn't trust "Brett," but Collins, well, I've been pretty harsh to her since long before this started, and nothing about her handling of "Brett" has given me any pause to reconsider my low opinion of her. The GOP could lose Murkowski and still confirm "Brett," as long as Collins stays stupid. Collins and Murkowski don't always vote the same way, and I have seen nothing from Collins to suggest that she is taking this much more seriously than Grassley. The uncertainty I have here is based on the potential for more news breaking, either a mess tomorrow, something from Montgomery County, or something else. Without that, the GOP keeps Collins, and if they keep Collins, they confirm "Brett." Murkowski? She may wind up the lone GOP dissenter, but every indication Collins has ever given is that she is on Kavanaugh's side. Why is the betting on Collins so uncertain? Bluntly, I think this is about reputation. Collins has a reputation as a "moderate," and I have written about this recently. A lot of people have expectations about "moderates," both normative and predictive. Yes, Collins has a voting score in the middle of the NOMINATE space, roughly zero on the -1 to +1 scale, but on party-based votes, she is more likely than not to support her party. Also, being a "moderate" does not mean being intelligent or moral. Those looking at Kavanaugh from the perspective outside the GOP bubble look at Collins, with positive expectations because of her reputation as a "moderate," but... she's an idiot. Murkowski isn't. She cuts deals where she can, and is more conservative than Collins, but Collins is not smart, and not principled. The "moderate" label is creating expectations that may not be warranted. If "moderate" is good, and voting for a rapist is bad, then Susan Collins must be disinclined to vote for a rapist because she is a moderate, right? That falls apart once you understand that Collins is just an unprincipled idiot. Nothing in her words or actions regarding Kavanaugh justifies the assessment that her vote is a coin-toss.
Something new may happen. See The Onion. However, you can't just multiply those probabilities together, and while I usually think the betting markets get this stuff right, I think right now they are getting a few things wrong based on expectations of "moderates."