Au contraire mon frère.
A busy intersection sees thousands of vehicles crossing every weekday. Say 3000 for an imaginary "dangerous" intersection. At a 1% failure rate that would be 30 cars in accidents, assuming each accident is a two car that would be 15 accidents a day.
That simply doesn't happen at any intersection. That would be an epidemic. I used 99% as a place holder.
Six Sigma processes are 99.99966% defect free. That's considered the rarely reached upper realm of human perfection. In our imaginary intersection 15 accidents per 3000 car crossings is an accident/crossing rate of .005, still too high for a real life "dangerous" intersection.
Now you research and find the average accident (and remember most accidents are simply fender benders with no injuries) rate at any red light monitored intersection.
Then your point will be made...or not.
Just two weeks ago, I was almost creamed by a lady talking on her cell phone and blowing through a red light at about 50 mph. Now, I'm not particularly interested in the number crunching like you have done. I'm satisfied with saying someone could be involved in an accident while running a red light because it happens. Maybe not 15x per day at the same intersection, but I noticed that you didn't take into account the number of cars hesitating on a green light and waiting/watching for the red light runners and inattentive drivers to clear the intersection. I'm not sure how you could calculate that figure but it's certainly a factor in preventing those red light runners from being a bigger accident statistic.
Cheers, my dear.