As far as I know changing one bit of the input should cause a "random" %50 of the bits in the digest to flip as well. It should be fine.
Incidentally I like the idea of using the stringified form of {} as part of your hash. Although maybe you might want to use substr({},7,7). Hmm. well I just checked this and on two runs I got the same value. Have you double checked that this works the way you think it does? I suspect that it might be reusing the same value each time.
Yves
--
You are not ready to use symrefs unless you already know why they are bad. -- tadmc (CLPM)