I found this piece by Alex Hutchinson very interesting, here is one excerpt on the issue of incentives:
One reason marathoners are running faster is that road racing is more lucrative. When the Sheikh of Dubai put up $1 million in prize money plus a $1 million world-record bonus in 2008, the Dubai Marathon instantly became one of the world’s fastest, despite its desert temps (average high in January, when the race is held, is 75°F). In fact, prize money for road races more than doubled since 1998, while track racing purses have gotten smaller (see below). As a result, runners are increasingly heading straight to the marathon. But big money can also draw the fastest runners away from the fastest courses, and the standard winner-takes-most prize structure favors cat-and-mouse tactics as runners race each other instead of the clock. When the Amsterdam Marathon switched to time-based prizing in 1999, four different runners immediately smashed the course record by 90 seconds. The sub-two-hour solution? A big pot of money that runners can win no matter where they race, and that is shared equally among all who break 2:00 in that event.
It is not obvious to me why a big first prize is not a good incentive, for instance why does it militate against speed to “race each other instead of the clock”? Is it that the runners stay on too few courses, thus lowering the variance of performance outcomes?
In any case the hat tip goes to Vic Sarjoo.