[gwern]

> Most (all?) of those decisions were going to come down the same way regardless; the judge (or clerk or amicus brief author or whoever) had already decided and the citations are just a way of making up a post-facto justification.

It's true that since they find that the gain appears to be concentrated in 'positive' citations, used as justification, they probably didn't flip many decisions immediately (if any). But they also do followup linguistic analysis to show that (like the earlier studies) judges are borrowing language in describing their decisions. So you are going to have an accumulating effect here where the citations at zeroth order are used for justification, but that makes those cases better known later on, and they will be described as the article describes, and increasingly interpreted that way when read by later judges due to being precedent, and who will then copy it (that's how common law is supposed to work!). And that may well start begin flipping cases.