Cameras have a significant impact on criminals, but they take effect slowly. I speak from experience.
I own a business across the street from an unused building. For years it had been a site for heroin dealing, vandalism, aggresive panhandling, multiple assults and batteries, and at least one mugging in broad daylight. This, of course, was not good for business.
The cops wouldn't do much. They frequently had higher priorities. ( When a crime was being committed and we called, they would often ask if a weapon was present. This, I learned after awhile, was dispatcher-speak for: "We're too busy with more important things.")
So I got a high quality netcam, put the camera feed live on a web site, and informed anybody who cared to listen ( this included neighbors, cops, drug sellers and buyers, etc ). It took several months for the problem to go away.
First to disappear were the most obviously affluent buyers. Then, over several months, there was a gradual decrease in traffic, and the last to leave were the most desperate looking ones of the entire crowd - the stone junkies with half their teeth rotted away.
That was about two years ago. Every month or so, someone shows up, still thinking that it is active. They sit there for an hour or so, then give up and go away.
What initially surprised me was how long it took to have an effect. I've thought about it a bit since then, and below are my conclusions.
First, a camera, even at its best, is merely an influential tool. It puts a drag on the crime business only if someone follows up on what they see, or if the criminals think that someone might do so. It doesn't actually stop crime immediately
The combination of a person and a camera is very effective.
This last one I have tested up close and in person. After seeing the success of the netcam on one storefront, I started carrying a camera around the neighborhood and taking pictures of drunks, junkies, etc. They hate it. The most common response is to protest innocence and leave. The second most common is to threaten me with bodily harm and then leave. Oddly enough, nobody asks why I am doing it. ( This really surprised me. Do they understand why, or are they completely incurious? I have no idea. ) But nobody stays except those too drunk to move.
Second, I expect the attendance of criminals at high-crime-rate sites to sort of follow a bell-curve distribution, much like any group of attendees at any informal gathering place with no posted hours; a few are there every day, the majority show up something like weekly or monthly, and there are a few who make it every year or two. That is roughly what we saw before the camera was installed.
So there will be people who don't know about the camera for a long time.
Third, there is not a very wide variation in the response to a camera.
PS: Please note that any pro-camera comments above are about a privately owned camera. I share the concern, expressed by many, that cameras will be misused by governments. The solution is to make them all public netcams with no password, available to anyone with a browser. The cops can use information, but it is less likely for them to misuse it, because anybody could have copies.
PPS: The astute reader will note that there is an implied 'broken windows' theme embedded in the above text. Yes, I think that Keller's broken windows theory is correct, and that numerous municipalities - particularly NY - have implemented it badly. In my experience, if you engage the sub-criminal or semi-criminal behavior like panhandling or littering, you discourage many of the people who would otherwise commit serious crimes.