Well, compared to Bruce's Blowfish, it too is symmetric with a variable key length. What is different about ColorBlynd is that the block length can vary and the number of blocks within the key varies as well.<p>What this does is add a "meta key" beyond simply knowing the key. The meta key includes how the key is broken down into blocks. And the number of blocks is used during each round relative to the block length.<p>So if you know the exact key, but don't know how many blocks there are or how many bits per block, you wouldn't know that you had the entire key.<p>The whole color deal is a neat way to transport the key and the cipertext, but it isn't integral to the algorithm.<p>The actual algorithm can be succinctly described in 4 sentences.<p>I'd definitely like to have Bruce look at it, but I'm not sure how to convince him to check it out. If anybody here knows him personally and thinks this might be worth his time, please forward him this thread.
I believe the basic algorithm is truly novel. The usage of colors is a little bit of a gimmick, but is an interesting way to store data. <p>I'm just not sure what I could use it for or if anyone might be interested in it. <p>One interesting aspect is that the algorithm is very dependent on the bit length of each chunk of the key as well as how many chunks there are. Almost like there is a meta-key.<p>Without this knowledge, even if you were to guess the entire key correctly, decryption fails without any indication that you were close.<p>As noted at the bottom of the page, I came up with the basic algorithm in 2002. I was researching several other cryptosystems and it was one of those "eureka" type moments. <p>So, any ideas, comments, criticism, or whatever is welcome.