–
[George is in his office at Yankee stadium, he pushes 'play' on a cassette recorder. The voice on the tape sounds exactly like George.]

VOICE: Chapter one. In order to manage risk we must first understand risk. How do you spot risk? How do you avoid risk and what makes it so risky?
–
http://www.seinology.com/scripts/script-140.shtml

(Sorry to make such a trivial contribution to your excellent and thought-provoking site.)

From a risk compensation point of view, a car driver suffers from very little risk of being hurt if they accidentally collide with a pedestrian, while a pedestrian suffers a great risk of being hurt in a collision with a car. Thus the pedestrian naturally takes a lot more care than the car driver does, and the car driver always wins the game of “chicken” when negotiating who should give way to whom.

Looking at this another way, “shared space” works when the apparent risks to all road users are balanced. Just removing signs and kerbs, in the hope that everyone will suddenly take turns because the boundaries have gone, is trying to change an effect without affecting the root cause.

If you want the motorists to share, you have to make them feel as vulnerable as the cyclists and pedestrians, and that’s not easy to achieve. This can be done, though, by making the motorist feel as though they’re a tiny minority. Or you can make the motorist realise that they’re about to become vulnerable pedestrians themselves, as in large car parks.

In my experience, shared space works between cyclists and pedestrians much better than between cars and pedestrians or between cars and cyclists. In a cyclist+pedestrian space, the risks are much more evenly balanced than when there are motor vehicles present, so cyclists (the more dangerous mode) do moderate their behaviour significantly.