JonWienke@aol.com writes:

I have no disagreements with this. I merely proposed using the compression function as a means of roughly estimating entropy and preventing the seeding of the hash/PRNG with potentially "weak key" type data.

It's not a useful estimate of entropy, and I don't see what you mean by "`weak key' type data". There are no keys or PRNGs involved here, just a hash function. Now, if you've got a sufficiently compressible data stream, compression may be a fast way to jump-start the distillation, but you absolutely need a priori information on the entropy of the source.

Would anyone like to propose a means of measuring entropy that we can all agree on?

If your definition of entropy is at least as strong as Kolmogorov complexity, it's infeasible to compute. The way to measure entropy is to spend ten years trying to understand the data source, and hope that no one else can afford to spend twenty. -- Eli Brandt eli+@cs.cmu.edu