Ed Felten half-seriously suggests that we create censorware that doesn't overblock and has an open list, so that US libraries who want to receive federal funding can do so without unduly harming free speech.
Up for the challenge, I've written a simple censorware program (the censor code is only a couple of lines): censor.py
It's easy to run: $ python censor.py
Then set your web browser's HTTP proxy to http://127.0.0.1:8000 as it says.
By default it blocks all URLs starting with http://www.microsoft.com but you can adjust this by editing blockprefixes.
I don't plan to maintain this seriously unless there's interest, but I'm happy to hear your comments.