[joshvm]

One of the arguments for mass surveillance is that you could hand over to another system if the suspect goes off-camera. So in this case, presumably they knew exactly what time the thief left the building and, provided everyone talks to each other, you could figure out where he went next. Ultimately you might be able to determine where the suspect lived or transition to a camera where better footage is available. Problem is that while we are watched en masse, most of these systems are separate. I'm not arguing that we should have this capability, but this is the obvious use-case.

The quality of CCTV footage (or lack of) is well known. How many times have you seen a news report with a blurry photo of a suspect and a plea for information? This is vague proof that mass surveillance is ineffective unless you throw serious manpower at it.

This is what happens in the case of serious terrorism cases, where the police can wave RIPA around and subpoena footage from everyone, but presumably tracking between cameras still requires a lot of manual intervention (and therefore money). Catching a locker thief isn't worth it.