MIKE OSSIPOFF wrote:
Kristofer:
I'd said:
The solution: Choose someone honest, as your proxy. That's a big difference from
ordinary representation. You, and you only, choose your proxy.
If you don't think anyone is honest, then don't use a proxy.
I should add: If proxy-honesty is a problem, then remember that your proxy
could be anyone--
Your best friend, your spouse, your uncle, your father or your uncle or
grandfather. Maybe
your college instructor. It could be a leader of your political party, or a
candidate
of that party, or any member of that party, etc.
You replied:
So the problem is really one of abrupt change (whether a ballot should
be public or secret)
[endquote]
No, the question of secret ballot isn't a problem. Your voter ID number is
anonymous. No one knows what person has what voter ID number. Yes, the system
knows if you designate a proxy, but it only knows that the person with that _number_
designated a particular proxy. It doesn't know the name of the person who voted
a certain way or designated a certain other voter ID number as proxy. Of course
neither
does it know who has that number that you designate as proxy.
How are you supposed to know who you're subscribing to (and thus the
trust you put in him) if you don't know who he is? You would have to
know the mapping between a person (say a candidate of the party in
question) and his ID (which you'd need to subscribe).
The proxy could get around this by only giving his ID to the people he
trusted, but then the proxy would have to balance between the concerns
of the additional votes he might get by saying "I'm 327923007, subscribe
to me if you like my opinions", and the attack resistance he gets by not
saying anything.
The ballot is entirely secret.
I'll grant that it is if you're not a proxy. It's easy to make a system
that is proxylike yet secret for non-proxies: just have a traditional
secret ballot where the proxies simply give suggestions the voter can
follow (or not).
You continued:
You would
want a proxy with lots of power to be accountable
[endquote]
Accountable to whom? When designating the proxy, you do so because you agree
with hir,
and are willing for hir to vote for you on everything on which you don't vote.
Accountable to the people, in the same way that a politician is. The
usual argument for transparency at the high end is that the principals
(the people) should know what the agents (the representatives) are
doing, so a voter can vote for someone else if his representative fails
to follow up on election promises, or reverses position for personal gain.
Ideally, if we follow this reasoning, the proxy should be accountable to
those who give him power in proportion to how much power they give him,
so that in the long run, the subscriber's ability to correct the proxy
is no more and no less than if he decided to vote by himself.
That is a gold standard. Representative democracy can approach it
somewhat, because it's very clear who has a lot of power and who has
not. The representative democracy approach isn't perfect, though,
because the feedback is rather coarse-grained.
Proxy democracy has a more fine-grained feedback mechanism. However, it
can't draw a line between "has power enough to be open to scrutiny" and
"doesn't have enough power to be open to scrutiny" beyond just the
voter/proxy dichotomy. There isn't such a line, even: proxies come in
all magnitudes of power.
So this may keep people from supporting proxy democracy, because the
degree to which your actions are made (or can be made) public as soon as
you're a proxy is out of proportion to how much power you wield. (At
least ordinary voters have privacy, so it's not as bad as I thought.)
I hadn't considered this matter before, but, in order for the system to know
how your proxy
voted, of course you must designate hir by number. But then you know the voter
I.D. number of
a voter (your proxy), so hir secret-ballot is no longer secret. But that isn't
as bad as it
might at first sound: Have you ever told anyone how you voted. Your electoral
system may have
a secret ballot, but you probably don't keep your voting or your political
opinions secret from
your best friends and relatives.
Sure, I have told people how I voted, but they have no way to verify
that I'm actually telling the truth. Were I in the United States, I
could choose to tell my family that I voted, say, Democratic. I could
also lie. If my family was of the religious right and I wasn't, I might
feel forced to say Republican even when I voted differently. Similarly,
I might not want to be a liberal proxy in a very conservative town,
unless I could somehow shield from the conservatives that I was a
liberal proxy.
As for accountability, since it was necessary to designate your proxy by
number, and since the system
knows how each number voted, then it could be allowed for you to inquire about
how your proxy
voted on a particular issue. You don't name hir by name. You can't do that for
any voter ID number--
only one that is your proxy. Of course you don't make the inquiry under your name, but only via
your number. Then, the vote of that proxy is posted somewhere, where you can read it anonymously.
I could see two sorts of proxy structure. I don't know which is the one
in your system.
In the first one, you have a broadcast system. Every proxy makes his
suggestions public and then anybody who wants to copy the decisions.
Obviously, proxy suggestions are visible to everybody here.
In the second one, you have a subscription model. Voters subscribe to
the proxy's list and get proxy suggestions in return. If computerized,
their software copies these suggestions to their own direct democracy
response queue - so the subscribers automatically copy the proxy
decisions. The proxies can't actually know that the subscribers do copy
their suggestions, though.
Either arrangement makes it impossible for third parties to know who are
subscribing to a particular proxy. If the system is of a broadcast
nature (one-to-many), then the proxy can't know how popular he is,
either; but if it's like a mailing list, the proxy would know who are
subscribing to him.
In any case, the suggestions would be informal. If they are formal
(binding), then vote-buyers could buy votes by telling the voter to
become a proxy of one, to whom one of the vote-buyers subscribe to
verify that the "proxy" votes as desired.
You could demand public ballots, but with voter IDs rather than voter
names -- but then it would only take one slip to link ID to name. Unless
the voter IDs changed regularly, everybody who were a public ("subscribe
to me, I'm xyz") proxy at some time would have his identity revealed,
and other identities could probably be narrowed down by analysis of
election responses. If the elections are sufficiently frequent, buyers
could tell sellers to vote this way on some minor question, then that
way on another minor question, and so on to lock on to the name-ID
relation through its "votes-over-time signature", and then start buying.
Anyway, your proxy may or may not have lots of power. It might just be your
spouse, who isn't anyone's
proxy but yours. And if you don't trust your political party or your candidate
enough to choose them as
your proxy, then you might want to reconsider about whether you want to vote
for that party's candidates.
That would work for a small scale private proxy. A large scale proxy
would be visible by his actions (though that's what we want), as would a
public proxy. Would you say this is a tradeoff a proxy has to make
anyway? That is, either to be public and thus accept potentially more
voters, but open oneself to vote coercion/bribery - or to be private and
secure?
Yes there could transparency at the "high" end, in the sense that you can
anonymously read how a proxy
with a certain number voted on an issue, by requesting that that information be
posted.
But if that were a problem, it would ok to drop the provision for checking on a
proxy's vote. After
all, you can choose someone you trust as proxy. On the other hand, the request
for an opportunity
to anonymously read about how the proxy with a certain voter I.D. number voted
on a certain issue
doesn't seem to violate anyone's privacy. Maybe there could be a limit on how
often the request
could be granted.
You'd need to have serious crypto to make the proxy not know what voters
are subscribed to him, the voters not know how the proxy voted, *and*
still have the votes count. Blind signatures, perhaps.
Even if you could, that would cut the feedback that lets the subscribers
know if the proxy can still be trusted. I don't think the problem (or
difficult part) is that the proxy system is everywhere too transparent
as much as that it's trying to span a wide range of influence with the
same amount of transparency. Representative democracy has two sets: the
voter, where the vote is private, and the representative, where the vote
is public.
over privacy at the low end (because it can't be otherwise, unless
I'm missing something).
The system would be entirely anonymous at the low end too, via the anonymous voter
I.D. number.
Hm, I suppose that's correct, as long as it's theoretically impossible
to link IDs to voters, or if the IDs change so frequently that any
attempt to nail them down will fail.
----
Election-Methods mailing list - see http://electorama.com/em for list info
