Thread by @henryfarrell

1. washingtonpost.com/politics/2020/… My @monkeycageblog take on the plausible story behind the decision of the NYT/networks all to jump at once after days of waiting. Brief summary below.

2. The role of major newspapers and networks in the U.S. "saying" that the presidential election has been won is weird, and as far as I know highly unusual internationally. It obviously isn't a law - but it has become a collective norm/expectation.

3. While I don't know of any research on this (but IANA Americanist), my presumption is that this is a contingent byproduct of a decentralized vote counting system, where there isn't any immediate official decision as to what has happened overall.

4. So news organizations rather than officials create the common knowledge that the election has been decided. And they have to do this based on statistical best guesses and other forms of knowledge. They can be wrong, or overconfident, leading to awkwardness or even disaster.

5. This time around, Fox called Arizona pretty early. Ex post, it looks as though they went out on a limb, though the limb will likely hold, despite some dangerous creaking and cracking. This presumably made other networks/newspapers pretty nervous about getting too far ahead.

6. This in turncreated a complicated strategic situation. Over the last two days, news organizations have become increasingly certain that Biden was going to take it. But there wasn't any obvious point at which near certainty transformed into absolute certainty.

7. And no-one wanted to be the first to venture out and say it definitively. It's a specific kind of collective action situation - what game theorists call an assurance game. I don't want to do something unless I am sure that everyone else will do it too.

8. So what happens? As far as I can see (this may turn out to be wrong) - people find a Schelling type coordination solution. When you are trying to solve a complicated strategic problem, where there are different possible solutions, you look for a "focal point"

9. That is - you look for a potential solution that stands out to everyone, perhaps for arbitrary or semi-arbitrary reasons. So the coordination game here is - "when can news organizations officially say that this is over?" See @NateSilver538

https://twitter.com/NateSilver538/status/1324835859812798465

10. And over the last 24 hours, gossip begins via Twitter about how news organizations might jump when Biden's lead exceeds the 0.5% cutoff for a recount in Pennsylvania. This created a kind of self-fulfilling prophecy, that news organizations _would_ jump at this point.

11. This was quasi-arbitrary - there are many other potential complications besides the need for a recount. But this arbitrariness was secondary, so long as _everyone knew that everyone else knew_ that this was a plausible cutoff point to make a decision.

12. And since everyone, including the professionals, was clicking the same Twitter feeds with the relentlessness of conditioned monkeys in a Skinner box, everyone _did_ know that everyone else knew.

13. Hence, plausibly the outcome. First - the risks of being first mover in calling are greatly lowered, since the first mover can be pretty sure that they will be followed. And second, everyone following in an incredibly short space of time, once the first call was made.

14. This explanation might turn out to be wrong. But I suspect that it is right or mostly right, and provides a nice illustration of how simple game theoretic reasoning can usefully explain social processes. Finis.