-----Original Message-----
From: wikimediauk-l-bounces(a)lists.wikimedia.org
On Behalf Of Chris McKenna
Sent: Friday, June 29, 2007 1:00 AM
Subject: Re: [Wikimediauk-l] Editors' exclusion due to privacy
One thing that you need to ensure somehow is that a
person does not gain
more than one vote by being a member of WMUK and a member of the
anymous trust.
Say I am a member of WMUK as Chris McKenna, and then apply to join the
anon trust as Thryduulf. Obviously a sane first check would be to see if
I've published my name anywhere on a Wikimedia site (in my case I have,
but for the purposes of argument assume I haven't).
The anon trust can't say ask WMUK "Do you have a member by the name of
Thryduulf" as they don't register people by username they wont be able to
say. The trust also cannot say "do you have a member by the name of Chris
McKenna?" as this proves that I am a member of the anon trust - which
removes my anonymity. This also works in reverse with WMUK asking the
trust.
A lot depends upon the process and requirements for normal registration.
What's important for the anon editors trust is not so much that we can't
verify them. It's that there should be no formal record which can be used to
trace back to an identified individual.
If this was computerland, we could for example require some date of birth
photo ID from everyone - then SHA1 hash the full name and date of birth and
keep the hash only, not recording the rest. That's probably impractical
outside the world of I.T., though. The question "how would you prevent
multiple registrations by a person prepared to pay multiple times" does need
sorting out. But there needs to be found some answer, even if it's simply
that WUK retains a lawyer who under legal privilege holds identifying
information for them as a client, for the sole purpose of confirming
eligibility and non-duplication against Wikimedia UK's members and anon
members list, and would-be applicants contact said lawyer with their ID
directly, and Wikimedia UK and its representatives are never allowed access
to it. That should also resolve the problem for all but the most concerned
editors.
But yes, without real ID entering the picture somewhere, or only permitting
voting in person, it's hard to avoid that loophole in any voting body.
Also you need to sort out what the trust does when its
members are not
unanimous about how they want to vote. Say the trust has 5 members, and
the question is "Should the logo be red, blue or green?".
If all 5 members agree it should be blue, then there is no problem, the
trust casts 5 votes for blue.
If 4 members think green, and one abstains then it cassts 4 votes for
green.
If 2 members think it should be blue, two prefer green and one want the
red logo, can the trust cast 2 votes for blue, two for green and one for
red?
If two members abstain and each of the other three prfer a diffierent
colour, does the trust place 1 vote for each colour, or does it not both
to vote as it would be irrelevant?
As conceived, each person has one vote. The anon editors trust might have 20
members, but if only 5 specify how their vote is to be used for poll on the
colour of the logo, and their votes are 2 blue, 2 red, 1 green, then the
block vote is used to vote for 2 red, 2 blue and 1 green, (?and possibly 15
abstains), on the basis that it votes as its members each direct it. Your
middle alternative works most universally.
Although mathematically it's the same thing, it's not really up to the anon
editors trust or WUK to judge "irrelevance" of a vote that might be
cancelled by another. In usual voting, we still count all votes even though
one side will usually outnumber the other, even just for accurately
recording the size of support for each if no other reason.
Good comments both. But we do need to find an answer to this one somehow.