It matters. People don't like being in prison for even small amounts of time; and being a registered offender means you're more likely to be identified the next time 'round. Then there's the social pressure: for those in not-entirely dysfunctional families and peer groups, exposure is a nasty penalty in itself. All in all, if you know you've been caught once, you're not as likely to risk it again - even if the penalty isn't harsh. A small penalty with a high chance of punishment that follows quickly is generally more effective than a high penalty and a small chance of punishment that isn't executed for years.
You don't lock up kids for years when they do something wrong either - right? And yet, they still seem to learn. And while adults may not change their ways quite so easily, there's no point in overdoing it either: that's just wasting state resources and wasting the time of a member of the public who despite the minor infraction can still be an otherwise productive, social member of the public.
Using more punishment than necessary isn't just immoral, it's plain stupid: it wastes both the state's and convicts resources and can invite police fraud. Even fines (which aren't as inefficent as prison time) have a huge overhead, and can be pointlessly small or overly harsh depending on the convicts resources. Better to invest in prevention.
So, *if* CCTV's would actually enable higher probability lower severity punishments that'd be great. Unfortunately, the article doesn't actually quite go that far. I wouldn't quite trust the officer in charge of the camera's (who's job or prestige might make him a vested intrest) to make a completely unbiased report; and that's who the BBC is quoting. And what he's saying doesn't quite mean the CCTV's *improved* policing, just that they were *part* of policing in those 2500 cases. Perhaps spending the police resources elsewhere rather than on CCTV's would have been even more effective; and perhaps some of those 2500 cases would have been solved anyway and the CCTV's were merely incidental. In any case, the article doesn't demonstrate they were actually useful.