Automatic recognition is the way to go; all necessary data can be extracted from driving patterns. Editing a light at a crossing does not help; the editor should provide Waze with the "phasing" of the light as well. 10 seconds red/50 seconds green is better than 30 seconds/30 seconds; but depending on traffic density a 60 second/60 second phasing could be better...
Waze could, if equipped with the right algorithms and enough data, bring this down to an accurate prediction of time spent on consecutive roadsegments. Having the lights on the map does not help in accuracy, only raises the complexity for the editors, of which there are far too few anyway. Besides, I know of crossings with 30 to 40 lights working in concert. Good luck editing that AND expecting a better accuracy than by recognition (which alas is not done by Waze - yet).
Waze could, if equipped with the right algorithms and enough data, bring this down to an accurate prediction of time spent on consecutive roadsegments. Having the lights on the map does not help in accuracy, only raises the complexity for the editors, of which there are far too few anyway. Besides, I know of crossings with 30 to 40 lights working in concert. Good luck editing that AND expecting a better accuracy than by recognition (which alas is not done by Waze - yet).
How about adding Traffic Lights/Signals