Diminishing returns kick in after your first step. Your second step is less beneficial than your first step. The question isn't when diminishing returns kick in as much as when negative returns kick in.
I realize you're probably exxagerating to make a point, but I don't get that. Unless you are running a one step race, the first step is irrelavent without the second. And to get the benefits of a long run, you actually have to run long. So I think the returns don't start diminishing until much later. I may be thinking about it wrong, but I'm not asking about when negative returns kick in. If I was training for a race in December 2008 I could think about that but not in two months. I'm concerned about the gains from going longer interfering with potentially greater gains from harder hill/threshold workouts.
As GTF mentioned, though, this point is a very individual factor. Personally, I do well with long runs that most people don't really consider doing. Even when focusing on races no longer than 8k or 10k, I like to do long runs of 20 or more miles. I just feel like they help me be able to sustain a higher effort level throughout the race. Of course, that's just the right level for me. Your mileage may vary.
I never was very good at translating GTF to the common language. I think I do well on longer runs as well. In the trail races I've done in the past, I've beaten many people with more “pure speed” than me because of my greater base. It is just that given my current base, the amount of time until the race and the nature of the course, I wonder if I'll realize greater gains from pushing the other workouts after reaching a certain base level.
Double…thanks for the encouragement. I wish I had a little more of your just go do it in me.