This is a special post for quick takes by Ace Delgado. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
Previously wrong in general? How similar do two estimates or predictions need to be for the earlier wrong prediction to be informative about the later one?