Why you should always assume the driver has not seen you

Can you develop that negative assumption into some positive behaviour change that would have helped that particular cyclist, without demanding perfection of perception or judgement?

I would have been a) further right, since the Audi clearly isn't positioned to take a right turn, b) sitting with my fingers tensed on my brake levers anticipating the Mondeo wouldn't stop and c) maintaining my speed or braking slightly until I saw that the Mondeo was stopping and it was clear to accelerate. I'd have done this in a car, and I'd do it on a bike.

DJ, andyp etc are all completely right that the cyclist has not broken the rules of the road. He's positioned correctly, has right of way to enter that roundabout, and yet got knocked off and bloodied. Despite this, others are also right that he's note done the best job of staying on his bike.

The argument that the rules of the road are enough assumes a perfect system: perfectly consistent rules with perfectly rule-abiding peds, drivers, cyclists, bikers and so on. Each of these groups has fallible members, and the rules themselves aren't perfect, so as we ride/drive/cross the road we have to make continuous assessments of the risk associated with our current situation. Your own personal level of risk aversion probably has a strong effect on how right or wrong you view the cyclist here, but it's clear that he either underestimated the risk of the Mondeo failing to stop, or he sensed the risk but took a punt on pulling through, and lost.

The car should have stopped, and had he been a different cyclist with a different set of priorities, he could have avoided the accident. Either way, the poor bastard has my sympathy.