Over the first full week of running leaderboards (to replace partly banners and partly skyscrapers), it seems to be worthwhile.
In this week impressions were little changed (up 3%) so reasonably comparable. Average Click-thru rate was up almost 25% from the previous rate, CPC was down 5%, total improvement in CPM (revenue per thousand impressions) was up 18%.
Perhaps the most interesting stat is it seems to be less volatile. Although days still vary greatly for no obvious reason, the ratio of highest:lowest daily CPM has gone from 2.7 to 1.7, in other words the worst perfoming day in the previous week was a bit above 1/3 of the revenue of the best day, now it's better than half the best day within the week.
So it works for me so far, although I've no way of knowing if this improvement would have happened without any change. It does seem to have started with a bang and tailed off though so it may not even retain this gain next week.
728 * 90 still performs at a nearly triple click-thru % over 468 * 60 for me and the percentage doesn't change much day to day. This seems to say that many people who decide to click on the ads will click on more ads if given the opportunity. Perhaps, a nice Christmas present from Google will be the introduction of the six ad banner?