Reminds me of the infamous John Forester's "Actual Sidepath Test" from his book Effective Cycling demonstrating the inherent danger of riding a bike path (by riding like a speed demon jackass on it.)
Extract from http://www.johnforester.com/LAW/Bike...ontroversy.htm :

"I know of only one valid test of a sidepath system, my own. Palo Alto instituted its mandatory sidepath system along my route to work, which I had used for several years with no problems and no incipient collisions. After I had been convicted of continuing to ride on the roadway, I was hounded by bikeway advocates saying that this system had been instituted for the safety of cyclists and that my ill opinion of it was unfounded. Therefore, I decided to ride that system using the same speeds and right-of-way that I had enjoyed on the roadway. After all, if the system was safer, then it would be safer at the same speeds as before. Seven times in five miles I faced incipient car-bike collisions that I was able to avoid only by the combination of expert understanding of traffic with expert bicycle handling skill. Few other cyclists would have avoided any one of these. The cyclist who had observed part of the test was white-faced and incapable of speech when she met me at the end. I tried once more, and in my atttempt to make a left turn the only course I could take that would not certainly involve me in a car-bike collision was to ride head-on in the reverse direction into an oncoming two-lane platoon of cars, riding the lane line and hoping that no motorists was in the process of changing lanes. I terminated the test because of its excessive dangers.

Since I had had no such incipient collisions in several hundred days of use of those roadways, and had seven in the first attempt to use the sidepaths, I concluded that the risk rate was at least 1,000 times greater on the sidepath than on the roadway. This test has been reported in all the various editions of Effective Cycling and Bicycle Transportation.

Bikeway advocates deride this test: "infamous ... simply ignored by non-vc types as a biased and unscientific anecdote against sidepaths that has little relevance to the real world as they see it." (Geary, Riley; public e-mail comments on the draft of this paper.)

How is this test biased and unscientific?

As to method, I rode the sidepath at the same speed and with the same expectation of right-of-way that I had used on the road. I could not overstate the severity of the test, because I was physically incapable of riding much faster; I could not be given greater right-of-way than I had on the roadway.

As to evaluation? When I wrote that it took both unusual traffic comprehension and unusual bicycle handling skills to avoid the eight incipient car-bike collisions, was I overstating the case? Considering that bikeway advocates complain that I have excessive skill, and that I terminated the test after so short a period, that speaks for the dangers of that test for the ordinary cyclist.

As to replicability? One test, properly done, is valid scientific evidence, but repeated tests are better. In this case, there has been only one test. You cannot arbitrarily throw out the data from the only test that exists because you don't like the results. The logical people to repeat this test are not people like me who won't repeat it because we won't risk our lives in what we believe to be an extremely dangerous test, but those who believe that urban sidepaths make cycling safe for beginners. They are the ones objecting to the results of the test, and none of them has tried repeating that test. They can't plead ignorance of my test, for it is "infamous" in their circles."