If you watch the Bradley Wiggins Hour Record, one of the things the commentators mention over and over is how steady of a pace he rode. A whopping 33.88 MPH for an hour.

Would he have been better off with a sprint/coast, say varying from 30 MPH to 36 MPH. The answer is absolutely not. At that speed, the additional wind resistance of sprinting to 36 MPH, then coasting back to 30 MPH would have killed him.

Now, at a much more leisurely 10 to 12 MPH on the level, there may be some benefit of sprint/coast riding, as the effort to accelerate, say to 14 mph, then coasting back to 10, then sprinting to 14 again is relatively minimal, and the lower wind resistance would mean longer rest periods.

Back in the Wiggins case, it might only take a few seconds of coasting to drop from 36 to 30, where the casual rider would have more recuperation time coasting from 14 to 10, and much more effort for Wiggins to regain the 36.

Now, add hills into the mix. Yes, you get some freewheeling coming down the hill, but the added energy expenditure to get up the hill more than makes up for the coasting. And, thus it is easier to maintain a high speed on the level than going up and down mountains.

Part of the problem with the hill climb is a natural assymetry in speed.

Say you ride at 15 MPH for 1 hour, and at 25 MPH for 1 hour. Your average speed is 20 MPH (15 miles at 15 MPH + 25 miles at 25 MPH = 40 miles in 2 hours, or 20 MPH).

On the other hand, if one considers distance. Say one goes 15 MPH for 15 miles, in 1 hour. Then goes at 25 MPH for 15 miles, it only takes 36 minutes. So, one goes 30 miles in 96 min, or only 18.75 MPH. And, one gets hit with the higher wind resistance when coming back down the hill.