Ezra posts with little commentary that political ads [may] just waste money, particularly early on.
That seems an overly simplistic look at things. In the same way that a slight nudge over lots of time can shift an asteroid's path, a string of ads over a long period of time could well produce a meaningful shift in perception, but one that, in the wildly varying political world, is difficult to tease out from other background shifts.
This can be compounded if, say, an ad is released and a poll is taken during the high impact time that shows a shift that is then broadcast by news media, which makes others believe that perception/favor is shifting from one candidate to another.
Finally, the fact is that advertising really can't overcome a landslide either way. It only matters in cases where the election is very close. If 1% of the vote either way can make a difference you don't want your image dictated by your opponent for 5-12 months leading up to the final push. Especially since you know he/she is going to be spending extra that last week.