The Gotham Gazette's Wonkster reports the request from the site for crime data were
denied by the city police force. NYC's ComStat (typo corrected per comment) program is widely praised, but the data unlike that from Chicago is not provided at the block level. The next step? A Freedom of Information Act request.
That post linked me to yet
another interview with Adrian Holovaty, the fellow behind EveryBlock. Here he mentions one the open source mapping tools:
We use an open-source library called Mapnik to render the maps, so that library does the heavy lifting for us. Paul is also working on a how-to article, in the spirit of giving back to the open-source community, that explains how to use Mapnik.
I know of
Mapnik, but this forced me to finally confirm what I'd always suspected: the fellow behind it, Artem, was at Cadcorp with me back in the day. (Interview with Artem at Nestoria
here.) The world is small my friend!
Also of note, an article by Paul Smith, EveryBlock's mapping guy, that
explains the why and how of the solution.
...except that is, by that great show The Wire (on HBO) which often criticizes CompStat and those police (usually the bosses) who "game" it to seemingly reduce stats. In this season the Mayor finally had enough and fired his police commander.
Live by the stat, die by the stat.
The relationship between crime mapping, profiling and surveillance (especially of minorities) is still problematic as I pointed out in my post last year (http://ubikcan.wordpress.com/2007/11/09/la-police-mapping-muslims/).
Isn't it just a computer version of the pin-map, a tried and true tool of police departments? Seems to me, the real innovation is in the area of management, i.e., FORCING people to use data. As ubikcan says, that has its own drawbacks.
Social problems don't usually have (purely) technological solutions. The human factor always returns to bite you in the backside.
That is not always the case. While there are some things which will always be with us, many times "purely technological solutions" fail because they don't go far enough or are done on the cheap, with the former often being the case because of the latter. There are times technologists should Just Say No, that a system can't be built for the price sought.
Thus, in the case of crime statistics, like health statistics, three features are necessary for their proper collection and use. First, their reporting needs to be standardized, made mandatory, and based upon tangible measures. Some of these measures ought to be carefully chosen proxies for intangibles. Second, their collection and reporting needs to be periodically audited by a group independent of the organization subject to scruitany. Third, a team of statisticians needs to be consulted on their presentation, and one or more should be on call to assist in formal interpretation.
The trouble with pin-maps is that they are a crude means of estimating a spatial density, and there are far better ways of doing that, particularly digitally. If risk of crime is to be presented, surely the kernel size used to estimate these densities needs to consider the mobility of prospective criminals and potential victims.
This isn't stuff that's abstract and in the clouds. Field biologists use spatial statistics like this all the time.
All the same, I understand the sunk cost problems of trying to change the practices of big organizations. For example, I don't believe it was until very recently that the FBI's Bureau of Justice Statistics (BJS) took an epidemiological view of crime and its reporting, and ceased relying upon voluntary reports from local and state police. Before that, one got anomalies about, say, homicides, because one set of numbers came from BJS, and pathologists reported their determinations up through Centers for Disease Control (CDC), and they didn't match.