[EM] Re: Issues about "personal data" in releasing results of ballots:

[Diana Galletly]

On Sun, 23 Nov 2003, Donald Davison wrote:

> Hi Diana, you wrote: " Am I way off base here and if so in what form ought
> I to be asking for the data such that no-one's privacy gets breached?  I'd
> prefer the data in the form I requested, but as I'm particularly hoping to
> compare Condorcet methods with IRV here a pairwise matrix would suffice --
> but I don't think I could trust the central administration to draw such up!
> Which is why I asked for what I thought the most foolproof way of
> presenting the data.    C'est la vie" ...  Diana.
>
> Donald here: If you are asking for the ballots of some voters by name, then
> of course you are off base, but you do not say in this letter what form you
> are asking for the data, so we are not in a position to say if you are off
> base.

No, of course I wasn't asking for the ballots of any voters by name!  Perhaps
I should have been clearer: I was asking for the data in tabular form as
given (for a hypothetical example) at

http://www.chiark.greenend.org.uk/~galletly/voting/dagtoac20030902.html

> I am one who believes that all the ballots (without names) should be
> available so that interested members of the public can work the ballots and
> confirm or protest the official results, but the Cambridge central
> administration may be fearing people who will work the ballots under some
> other method and then YELL OUT that the city has given the position to the
> wrong candidate, that candidate `X' is the real winner of the election, and
> of course candidate `X' will agree and proceed to sue the city for fraud.

Since the Cambridge administration (University, not city as it happens)
are well aware that there are problems with the arrangements for ballots,
principally due to my speeches this year although there had been a great
deal of unsatisfactory happenings last year too, I don't think that the
worry is of that nature (or at least not mainly).  Also these are
referenda on issues, rather than elections of candidates (for which
the regulations were overhauled some years back to use "a more modern
form" of STV than the one they'd adopted in 1926), so I don't think there's
much fear of a losing issue suing!

> Now the central administration could present to the public and candidate
> `X' the argument that this Condorcet analysis is not valid because it is
> not proper to use Irving data in a Condorcet trial election.  This is
> because people will vote differently in a Condorcet election.

Not this electorate.  Maybe *one or two* people within an electorate that
now numbers some 3000+, but a much greater problem is that a good third
of those who vote seem unaware of what they are doing, some even still
believing it's a FPTP system, or that a single transferable vote means
they may mark only two options.

> The administration could say that the Condorcet method is deceptive because
> it is asking the voter to make lower preferences that will help elect some
> lower candidate while their higher preferences are still contenders.

Swings and roundabouts, I fear.  This is part of the problem where there
is no true best voting system.  But on issues rather than candidates I
think that trying to give greatest satisfaction to the greatest number
to be important.

> You wonder if you can trust the central administration.  Can the central
> administration trust you??

I think that that depends on one's definition of trust.  Certainly there
is a significant proportion of Cambridge's academics who do not trust
the central administration here.  Unsurprising: controversial things have
been hidden from them and then presented as a fait accompli.

Why should the central administration not be able to trust me?  I am an
academic, albeit quite low on the pecking order, and I adhere to such
things as academic integrity etc.

> Besides, your analysis is suspect, Condorcet advocates have already made
> their minds up, any analysis of Condorcet vs Irving they make is merely a
> ruse to badmouth Irving.

Note the use of the word *advocate* there.  Of course any advocates of
any system are going to use something that appears to be in favour of
their cause to attack their opponents.  (I get the impression you are an
IRV advocate, frex.)

But what I want to do is to come up with the best solution *for Cambridge*.
The results of Graces 5 and 6 as shown at

http://www.chiark.greenend.org.uk/~galletly/results.pdf

cause me profound worry.  I don't think either result is likely to have
reflected the wishes of the voters (not the least because of the way the
matters were presented to them, conflating two issues of almost orthogonal
relevance into the wordings of the options).

> If you really need ballots, use the Real World Condorcet election ballots
> that Alex Small was kind enough to post to this list on October 03, 2003.
> Those voters knew that they were voting in a Condorcet election, so we can
> assume that they voted accordingly.

I'll chase them up; I think I have them somewhere.  But was this an
edge case where any of the paradoxes occurred, or any important criteria
violated?

> The best analysis of Condorcet vs Irving will only come from giving a group
> of knowledgeable voters two ballots, one Condorcet and one Irving, and
> asking them to vote each ballot so that their first preference wins in both
> methods.

I live in hope that one day I might find such a group of knowledgeable
voters, all collected in one place and willing to fill out ballots for
academic study purposes ...

Diana.