This is partially a complaint directed at my fellow Wazers and partially a suggestion/data analysis conundrum for the developers. It might be a feature request, but since it's more than that I'm putting it in the main forum. Moderators, free to disagree and move it.
It seems to be human nature, at least everywhere I've lived in the States, to say, "Ooh, bad traffic. Must be an accident." In Waze, this turns into thirty people reporting accidents when they're actually just sitting in bad traffic.
Most bad traffic is just that: bad traffic. Some moron was driving 40 MPH in the passing lane an hour ago, and thanks to breaking laminar flow he or she has caused a full traffic stoppage.
The Waze app should be clear: unless you can actually see the accident, do not report an accident.
On the developers' side, it would be interesting to try and puzzle out the real accidents from the data. Get a dataset of reported accidents and find a way to filter it to confirmed accidents. Then, look at the relationship between the submissions and the false alarms.
Complicated; maybe impossible. Sounds like a fun challenge.