Chris, the traffic light controllers have redundancy and safety built in. There are mandatory hardware set minimums and there is a device called a 'conflict monitor' that closely watches all aspects of the programming. All yellows in a controlled intersection has a minimum set standard of two seconds for a yellow. If any of these monitored points (which are many and varied depending on all the functions/options of the individual specific intersection) go out of its preset or minimum parameters, the whole intersection goes into safe mode. This results in one of two things. Either all intersections go to flashing red, or every one of the intersections goes flashing red with a through intersection going flashing yellow.

Time every yellow you find. Within milliseconds, they all will be a minimum of two seconds (given the lag of seeing the light change and hitting the on/off timer). Depending on the intersection, it may be programmed for a longer yellow, but it will never be less than two seconds. The hardware will not allow it to be programmed sooner than two seconds. The internal program and redundancy will trigger a failsafe intersection lockdown (flashing red for all, or flashing red on most with flashing yellow on one through street only).