It could be set to trigger at whatever distance, measure the sound wave pressure coming from your vehicle and convert that to dB independent of distance, maybe.. then you'd have to be much more careful. Whatever way they do accomplish it, seems like there would be a large margin of error; what if another loud car was in the area at the time you triggered it? I don't know if a directional mic would be able to distinguish that
Two big concepts here, sound pressure (measured by a sound level meter at a distance) and sound power (total sound energy produced by a source). Sound power is what should be regulated, but to simplify the test to make it practical, current standards use sound pressure at a set distance, direction and engine speed. Regulating sound pressure levels without specifying a distance or direction is mostly politics and not based on science.
The current system will have a bias for giving tickets to vehicles closer to the sytem, vehicles moving away from the system and vehicles operating at higher rpm. Even with stock vehicles and mufflers, redlining them would likely exceed any currently established noise limits. This is the point Rob has been making about this whole mess being easily destroyed in court. Also there is a whole near-field vs far-field problem that is not accounted for here, basically anything travelling within a few lanes of the system can have problems with the measurement (distance correction will not work if you are too close).
Off the top of my head, there are a few ways to improve this, none of which are mentioned by either the news or the creators website, so I doubt they are included. All of the below systems can help take distance and direction out of the equation, but without an engine speed correction, this whole system is destined for the garbage bin.
1)Array of microphones could use beamforming to pick out a specific car and calculate it's distance (disadvantages -expensive, a pain to setup, would be influenced by other noises in direction of beam (ie cars in other lanes along the same line).
2)Rangefinder plus current system. Cheaper than system one but influenced by any noise in surrounding area.
3)Probably the best compromise is multiple microphones to form a beam perpendicular to the road, the beam never moves and a rangefinder determines which lane the closest vehicle is in. To work properly this whole mess should be mounted on a fence or some other surface back from the road or it is destined to fail.
4)The best system would use multiple radar/rangefinder/beamforming systems to individually track cars (picture air traffic management) and try to calculate the contribution of each vehicle, but damn would it be hard to get working properly.
The idea isn't bad, but there are trying to do way too much with way to little here (and charging $200K for a system that could be build for less than a 10th of that). I just hope people realize the amount of fail before some city blows taxpayer dollars on this.