I have a counterexample to make: Skripal's case. AFAIK the police pieced together CCTV footage from multiple cameras, which allowed to track and identify the su

I have a counterexample to make: Skripal's case. AFAIK the police pieced together CCTV footage from multiple cameras, which allowed to track and identify the suspects. Given that the suspects are likely trained hitmen and the stakes were pretty high, I can only assume that the main reason most crimes don't get solved that way is the complexity of the search. ML is poised to change that. I'm not sure I'm for surveillance per se, but I do believe painting CCTV as useless and inefficient is wrong. It can be efficient in aiding high-level investigations if applied properly.