* John H. Robinson, IV <[EMAIL PROTECTED]> [010331 17:54]: > but it is harder to scan for 33 numbers than (say) 8. but other than > that, the results are the same: a unique, reporoduceable identifier that > is difficult to surreptitiously change.
Well, the md5sum/sha1 hash could be truncated to some smaller range if the problem is *that* bad -- such a truncated hash is used in IPSEC; the only real problem is an increased likelihood of collision -- out of 313 voters, even 32 bits ought to be plenty to avoid casual collisions. However, it may be preferrable to convert the bits into alphanumerics -- a-zA-Z0-9, perhaps add two punctuation marks to bring the total to 64 characters (six bits each) -- this gives us 128/6 = 23 or 160/6 = 27 characters for md5 or sha1. Another option, used with S/Key, is to convert the hash into several short english words. Although, scanning for the 32 numbers shouldn't be that bad: double click in one xterm, '/<middle_click>' in the other xterm.... Yeah. I like Raul's amending of John's suggestion. I'll think about it for a few days and see if I can think of any problems with this idea. It manages to capture ID, vote, and random data all in one easy bundle. :) -- Earthlink: The #1 provider of unsolicited bulk email to the Internet.
