An outright ban is too harsh. I like MotoMoe's idea of a special quarantine for pseudoscience. Make a link to the FAQ and a disclaimer than the post contains pseudoscience at least as prominent as the post itself, then folks come by and tell us all about the latest miracle product they just spent a hundred bucks on.
You could always set up the forum, then shut it down and ban everyone who contributed if it becomes a hive of bad engineering. Any good scientist knows it's okay to run an experiment, then change his methodology if he doesn't like the results.
