On Tue, Aug 25, 2009 at 1:59 PM, Gregory Kohsthekohser@gmail.com wrote:
I wonder what takes so long to upload a small data file?
http://meta.wikimedia.org/w/index.php?title=Board_elections/2009/Votes&o...
Let's see... August 25 minus August 12 equals nearly two weeks of delay (and subterfuge?)...
Not much room for subterfuge once they are posted: You'll be able compute the full pair-wise table and confirm that it produces the same results. Voters will be able to look through and identify that at least one ballot identical to theirs made it in.
Since the officials can't know who will and who won't go checking for their own ballots in the pile the only real avenue open for election rigging is through sock/meatpuppet accounts. The edit count and activity requirements mean that there should be sufficient public information available on each of the voters for anyone to go sniffing around for funny business. Since making a meaningful impact on the election would require on the order of 100 accounts concealing that kind of activity would be difficult.
The voting process could be improved — but I think it's one of the most resistant to outright manipulation of any online voting system I've seen.
Though this level of confidence is predicated on the raw ballots being available, at least eventually.
I provided the election committee with a sorting script on August 10th. This script addresses Thomas' "anonymising and randomising" concern and does so better than actually randomizing[*] because sorting is deterministic.
---cut here---
#!/usr/bin/python #Raw ballot information leak remover #input is ballots, one per line, I.e. #O,NHKCJILMGBFEDA #OMN,GBLKIJADFC,E,H
import sys for ballot in sorted([",".join(["".join(sorted(x)) for x in y[:-1].split(',')]) for y in sys.stdin]): print ballot
---cut here---
[*] http://web.archive.org/web/20011027002011/http://dilbert.com/comics/dilbert/...