It is also contrary to the way the law was enforced when I learned to drive, meaning that what counted was the status of the light as you entered the intersection. Plenty of drivers will learn that things have changed and learn it the hard (and expensive) way.
Now consider the effect shortening the yellow light duration will have. Not only do you need the reflexes of a NASCAR driver but a heavy foot as you zoom through to get across the far line in time. This is insanity.
Meanwhile, studies have shown (again I need to find a source link) that increasing the yellow duration - and in many cases that means restoring the recommended minimum!!! - has a dramatic effect on dropping the number of infractions. Yeah, giving a vehicle enough time to clear the intersection under normal operating conditions reduces technicality violations, imagine that.
There ought to be a law - against tampering with the traffic control system.
Here are several posted studies (back when I was looking at them).
http://www.motorists.com/issues/enforce/studies/TRB2004-001228.pdf
http://tti.tamu.edu/documents/4027-2.pdf
http://www.ltrc.lsu.edu/TRB_82/TRB2003-000136.pdf
http://www.ltrc.lsu.edu/TRB_82/TRB2003-000943.pdf
http://www.ltrc.lsu.edu/TRB_82/TRB2003-000285.pdf
http://stc.utk.edu/htm/pdf%20files/red.pdf
http://www.uritc.uri.edu/media/finalreportspdf/536146.pdf
However, I do not believe they back up what you say. The way the literature says they are supposed to be set up is to have the camera come on the moment the signal turns red. Anyone entering the intersection after that gets a ticket. Anyone entering the intersection before that should not get a ticket.
Lengthing the yellow indication to longer than the ITE formula (generally 3 to 5 seconds, depending on several physical factors) does NOT increase safety. In most cases, it decreases safety. Unless the yellow indication time is set wrong to begin with, lengthing it will not increase safety.
The science of traffic control uses statistics more than any other form of engineering (and I have worked in several of them). Statistics can be twisted and misused, but it is starting to prove out that in most cases, Red-Light-Cameras do slightly increase safety. It greatly reduces T-bone accidents which are the kind that are most often fatal. However, it increases rear-end accidents. These kind of accidents usually have fewer deaths, fewer and less serious injuries, and less property damage.
I wish it did not have to be done because the reason it is being used is for revenue, not safety. In addition, it will make it easier to use more cameras resulting in further loss of privacy. But, it is coming. The groundwork for this was laid 10 years ago. It is just coming to fruition now.
I have been saying here that “congestion-pricing” in the next big thing on the horizon. It is now where Red-Light-cameras was 10 years ago. Now is the time to kill congestion-pricing. it is too late to kill Red-Light-cameras, but most here do not agree. In 10 years when they have to start paying money for absolutely NOTHING (which is what congestion-pricing is) you will hear their yells.