Are stoplights really necessary?

Via Tom Vanderbilt, the London borough of Ealing is about to conduct an experiment in which they observe whether traffic flows better at some intersections without stoplights:

Ealing found evidence to support its theory when the lights failed one day at a busy junction and traffic flowed better than before. Councillors have approved a report which recommended that they “experimentally remove signals since experience of signal failure showed that junction worked well.”

I observed a similar phenomenon sometime around March of 2009 at an intersection near where I work.  For about a week, the intersection shown below, which is usually regulated by stoplights (including a left turn arrow), was regulated solely with stop signs as a 4-way stop. My personal observation was that traffic flowed no worse than usual, and if anything may have been better. Ultimately, I think the best solution for the intersection shown below is multi-lane roundabout.



View Intersection where stoplight was out of service in a larger map