Bad Neighborhoods as a Unifying Theme

The other day I was discussing the relative importance of climate change, peak oil, automation/globalization driven unemployment, and the financial crisis to society.  It seems to me that at least that last three of these things might be expected to converge in the phenomenon of an increase in bad neighborhoods in certain places:

• The financial crisis has obviously led to large numbers of foreclosures in certain areas, which has presumably driven neighborhood quality down.
• Peak oil and oil price shocks are likely to make certain outlying neighborhoods less economically viable, which might show up as higher unemployment, dropping house prices, and more poverty in those neighborhoods
• To the extent that automation and globalization render a portion of the population unemployable, we might expect some of those people to end up moving to the most affordable places, and reducing the employment level (and thus economic activity generally) in other neighborhoods.
• Even climate change can get in on the act - we might expect neighborhood quality to go down in places prone to floods, hurricanes, etc.

Of course, this raises the question of what we mean by a bad neighborhood - to what extent do income, employment, educational achievement, crime, etc, really cluster together?  And is it really the case that the trends above are having any noticeable effect on bad neighborhoods at the moment?

Ideally, I'd be in a position to make interactive maps and videos of key demographic variables so we could look at those trends for major urban areas, at least in the US.  The data almost certainly exists - the U.S. Census takes massive amounts of data - but so far it doesn't seem that the Internet has solved the problem of making census data conveniently available to the curious Googling blogger.  In searching around I found some half-finished very buggy free tools, and various very expensive tools, or tools that only run on MS Windows.

The map above was done in Google Earth from a map created by gCensus, and shows median household income for San Francisco Census tracts in the 2000 census.  gCensus seems to be awkward to use (eg no proper labels to scales, only handles one county/city at a time), buggy (maps don't line up properly in Google Earth), and unfinished (work stopped in 2007 with only parts of the 2000 census done).  It seems clear that at some point I'm going to have to have to find some free weekends and invest some effort into actual coding and scripting to get something usable for my purposes.  If anyone knows of any better free/cheap tools, let me know.