
Blowfish has not been broken in my opinion. I wonder if Perry is thinking of MacGuffin, the block cipher by Schneier and Matt Blaze based on an asymmetrical Feistel network. It was broken, and I think it was at Eurocrypt. Here is a message from sci.crypt a month ago where Bruce discusses the status of Blowfish. A weak key attack is known against a weakened version, but I think the weak keys are rare.
From: schneier@parka.winternet.com (Bruce Schneier) Date: 1996/03/14 MessageID: 4i907g$9lj@blackice.winternet.com#1/1
The most successful attack against Blowfish to date has been against the weak keys (two identical entries in an S-box). These can be detected in a 12-round variant, but not in the full 16 rounds. I still believe that random S-boxes are better than chosen ones, and think that more rounds is better than fewer rounds with better S-boxes. There are a few things I would do differently if I was to write the algorithm from scratch right now, but on the whole I am still pleased with the results.
Bruce
Hal