If you take the ideas that historically, humans ran barefoot and ran on trails, you start to wonder how this impacts stride length. Christoper McDougall has discussed this recently: "Modern running shoes let people run with their foot in front of their hips, picking up two feet of stride. You can't do that with the naked foot—it hurts."
I think both parts of historical running, trails and bare feet, are important. For example, as Will commented, a person running on trails with thick modern shoes is not really having to deal with the trail. They do not have to worry much about stepping on harmful objects. Running in shoes on trails may be better than running in shoes on pavement, but there is still something missing. If you are running barefoot on a trail, you have to run in more of a defensive manner.
Also, if you run barefoot on a manicured surface, you are also missing something. For example, last winter I ran inside on a track barefoot. While this was better than nothing, it missed out on the variation that comes from natural terrain.
It seems like trails and bare feet would create a natural "speed limit" for running. In these conditions you would undoubtably take a short stride length to protect yourself from injury. It may then turn out that very short stride lengths are "natural", even though they may not lead to the fastest race times.
Below is a video of Erwan from MovNat doing some barefoot running. Notice that he can speed up sometimes, but other times he has to slow down due to the terrain. This type of running experience strikes me as almost completely different than running at a constant speed on a track or flat surface.
