About CCTV "impacts"

Reading this report on "the impacts of CCTV in the UK, I was struck by some points in the conclusion:

"crime rates appeared to the authors to be a poor measure of the effectiveness of CCTV. The problem about measuring outcomes in terms of overall crime rates was that they disguised some important successes with particular types of offence. Moreover, in some cases (although not many) an increase in crime was an indicator of success, and this needs carefully teasing out. Similarly, mechanisms increasing recorded crime rates can work alongside those that reducecrime, and these can cancel each other out. Recorded crime rates were subject to a great deal of background noise from other factors, such as other crime reduction initiatives in the areas being studied, regional and national crime trends, and changes in methods of crime recording, any of which could mask the small impact that CCTV might have. (...) there was a lack of realism about what could be expected from CCTV. In short, it was oversold – by successive governments – as the answer (indeed the ‘magic bullet’, Ditton and Short, 1999) to crime problems (...) there was a tendency to put up cameras and expect impressive results, ignoring the challenge of making what is quite a complex measure work (replicating the findings of Ditton et al. 1999), and failing to define what exactly the CCTV system was expected to do. (...) the installation of CCTV requires more than the production of a technically competent system; generally, project designers did reasonably well in this regard. However, systems have to be monitored properly or recordings made and stored properly; but the quality of this work varied considerably from one control room to another."

Why do I blog this? was reading this at the airport after having found the reference in a newspaper, I find interesting the arguments given above as they can also apply to lots of other ubiquitous computing projects. The expectations towards such camera-based system may interestingly apply to other sensor-based deployments. There's a lot more to draw on the report, especially regarding the humans who are in control rooms and who have trouble keeping up with the huge amount of data that is collected.