[Ghoulish Delight]

As Alex said, it's not a conscious decision, it's a programmatic quirk.

The way these route-calculating algorithms work (in very simplified terms) is that every road, every turn, every transition is assigned a weight that indicates a relative "cost" of taking that route over another. Assigning that weight is a complex process, and that's what Mapquest pays those providers for. So for whatever reason, in the data the Mapquest buys, those transition lanes carry a lower "cost" of travel than the freeway-proper.

Knowing that transition well, as I drive it nearly every day, it could be that the average speed along the transition road is faster than the average freeway speed. Or it could be that it might cut out a few dozen yards of driving distance. Or some other combination of the many factors that go into those value assignments.

The thing is, the exercise of driving-route-optimization is a highly complex and nigh unsolvable problem in the world of computer programming. There is actually no way to solve perfectly without requiring an amount of processing power that grows exponentially with the number of routing options. So these programs are forced to take shortcuts and make assumptions. They're pretty good at it, but there will inevitably be some holes in the logic. Google maps is sure to have its share somewhere. Though of course gmap's on-the-fly route changing features make them a little easier to account for when they do happen.