[EM] proxy representation with "dissenting votes"

[Abd ul-Rahman Lomax]

Mr. Green-Armytage is one of the people working on the many 
possibilities for the use of delegable proxy. My own realization has 
been that the implementation of delegable proxy in governmental 
practice is unlikely unless and until a substantial number of people 
have experience with it in other settings.

While delegable proxy looks very good indeed to me, the devil is in 
the details. I would greatly prefer to see the (unanticipated) kinks 
ironed out in actual practice before even attempting to propose it 
for public election use (though Asset Voting is very attractive and 
is specifically designed for elections).

At 06:50 PM 2/25/2006, James Green-Armytage wrote:
>[...] now I'm
>thinking about ways to construct the legislature itself by means of a
>proxy system.

My own opinion is that delegable proxy, implemented as a 
bidirectional communication/filtering structure, will bootstrap 
itself into methods of operating that we can only dimly imagine now, 
if we can imagine them at all. In other words, once there is DP 
functioning in an organization which is considering political 
structure, it will do a much better job of arranging the details than 
we are likely to do, thinking only theoretically. It is useful, 
however, to consider possible details simply to recognize that the 
objections that can be raised to DP have possible answers.

>BASIC SYSTEM:
>         Each voter may assign their voting power to a proxy, or 
> representative.
>Voters may change proxies at any time (although it is not expected that
>people will change proxies very often on average).

The whole point of a proxy is to allow the individual to attend to 
other matters, i.e., to get on with their own life. Frequently 
changing proxies would be self-defeating. Further, I am highly 
recommending that DP structures only involve relatively small numbers 
of direct proxies collected by any individual. The number is 
essentially the number of people with whom the proxy can manage 
direct communication. My proxy should return my phone calls. Personally.

I don't think it has been sufficiently realized that delegable proxy 
is a solution to the problem of scale in democracy; it essentially 
could make direct democracy practical. In the assembly 
("legislature") I would see, every citizen could vote. But *not* 
every citizen could address the assembly or enter a motion. I think 
that this separation between voting rights and full participation 
rights is a new invention.

>         If you are chosen as a proxy, you may delegate both your 
> own voting power
>and that of others to another proxy in turn.

In other words, this is DP.

>         Seats in the legislative chamber are held by individuals 
> who hold a large
>number of proxies. The members of the legislative body do not have equal
>voting power; their voting power is determined by the number of proxies
>that they hold.

Asset Voting is a form of delegable proxy which creates a peer 
assembly, where every member has equal voting power, and may be more 
suitable for assemblies with fixed membership and terms of office.

>DISSENTING VOTES:
>         Members of the legislature should be empowered to vote in a 
> heterogeneous
>way. For example, imagine that I am a legislator, and that there is an
>upcoming yes/no vote. Once a vote has been scheduled, I may announce the
>vote to my constituency, and indicate that I intend to vote "yes".
>However, I invite dissenting votes from my constituency. Let's say that I
>have one million constituents, and one hundred thousand of them indicate
>to me that they will vote "no". I can then cast 900,000 "yes" votes
>against confirmation, and 100,000 "no" votes.
>         When an important and controversial issue comes up, there should be a
>natural procedural break between the announcement of a vote and the vote
>itself, to give proxies a chance to check in with their constituents, and
>allow time for dissenting votes to be submitted.

I've considered this, but came to the conclusion that routine 
vote-splitting by DP representatives would not be such a great idea, 
for quite a number of reasons. Remember, I'm thinking about direct 
democracy. If an individual holds an opinion on the subject, 
differing from his proxy, he can simply vote directly. When a citizen 
votes directly, that citizen is not included in the vote analysis of 
the *vote* cast by the proxy.

I would have proxies simply vote *as if* they were voting one vote. 
After all, they are just one person! *However*, when the votes are 
counted, a proxy's vote would be multiplied by the number of persons 
registered as naming that person as proxy, directly or indirectly. 
*If* a division of the assembly is requested.

Delegable proxy really does suggest a complete rethinking of 
democracy. Among other implications, I see it becoming far more 
possible to act with substantial unanimity. Remember, I'm thinking of 
a DP assembly as if it were a single entity, a superconscious being. 
Consider this: does an intelligent person make decisions, absent 
emergency conditions, when divided in mind? Generally, if we are 
sane, we will consider a potential action of consequence, weighing 
the alternatives and options, until we find one which does not divide 
us; when we are divided, we will postpone decision and continue 
deliberation. I find it interesting that the same word is used for 
thinking things over and for democratic process: deliberation. 
*Voting* is a very small part of deliberation; rather, what I forsee 
is much more like polling, the measurement of the degree to which 
consensus has been obtained.

I also see that DP in the pursuit of consensus will be much simpler 
in Free Associations, outside of government. Because of the peculiar 
characteristics of Free Associations, as I define them, consensus is 
something that will be valued, but it will not be necessary, for any 
caucus will remain free to make its own decisions, and act on them, 
at any time. This is not possible in a sovereign government, or, to 
be more cautious in my language, this has never been done with a 
sovereign government, and hosts of objections arise at the thought. 
But it is entirely possible -- and already exists, informally -- in 
Free Associations.

If the rep is going to poll her constituents before casting this 
proposed split vote, what exactly is the function of the rep? Why not 
just have the constituents vote directly?

I do not see proxies as rubber-stamps, simply collecting and passing 
on votes. Rather, I see the assignment of proxies as a means by which 
society weights the effect of the proxy.

For me, the question is whether or not to *allow* proxies to split 
their vote. It greatly complicates matters. For one thing, it 
requires the proxy to know exactly how many people he or she 
represents at the time of the vote. This is fixed assembly stuff, 
contrary to the idea that proxy assignments can be changed at any time.

I have always answered the objection, "But what if I disagree with my 
proxy on a matter coming up?", with noting that one could then simply 
vote directly on that issue, thus effectively cancelling the proxy 
*on that one vote only,* or one could change one's proxy assignment 
if one no longer trusted the proxy. Changing proxy assignments is 
always going to have *some* delay involved. But voting directly can 
be done at any time.

>         It seems that the internet would probably be a useful medium of
>communication between constituent and representative (although other forms
>of communication would of course be possible, especially at smaller
>levels).

Demoex in Sweden implemented, for a time, delegable proxy, but it was 
thought of as a high-tech, internet method. Now, the internet does 
make a great deal possible that would not have been possible without 
it, but DP is *not* a computer method, per se. It could be done 
without computers. It essentially is a networking of people. The 
communication between the people can be by *any* means, and I assume 
that communications with one's direct proxy would often be verbal. 
Telephone or in person.

If you read what writers on democracy say about its limitations, you 
will come upon the one that representatives cannot consult their 
constituents directly, when they represent large numbers of people. 
DP solves that problem in a way that has often been overlooked by 
writers on this subject: it distributes the communication load. Mr. 
Armytage is thinking of a representative who represents hundreds of 
thousands of people. But, as I would see the system evolve, she would 
only *directly* represent perhaps twenty. *And that is all the number 
of people whom she would consult when consulting her constituency.*

I think that effective assemblies will also not be much larger than 
twenty. Certainly not the hundreds that we see today. The fact is 
that such large assemblies do their real work in committee. Plenary 
sessions are largely ceremonial, places where decisions hammered out 
in committee are ratified. (Or where public contests take place 
between factions, but that is another story.)

>  I suspect that this system would be superior from a security
>standpoint to a nation-wide internet-based vote, in part because the total
>number of dissenting votes would be fairly small on average, and thus
>fairly easy to keep track of. The people who agree with their
>representative on the issue in question (which should be a large majority)
>will not need to vote directly, and thus the count process should be less
>unwieldy.

If they dissent, why not just count their vote directly?

Now, the elephant in the living room: the system proposed here 
requires open voting and assignment of proxies. If a proxy knows 
specifically and personally whom he represents, there is no secret 
ballot. This is not a problem in Free Associations, nor in many 
non-governmental contexts, but it could be a big problem in 
government. The simple solution is secret ballot assignment of 
proxies, but then the proxy has no way to determine who the 
constituents are. There is a possible compromise: a base layer of 
proxies are indeed assigned by secret ballot, but then those who 
receive those basic proxies, in turn, assign them openly. By 
volunteering to serve as an "elector," one might call these people, 
one is agreeing to act publicly.

(This is what I think Asset Voting would evolve to; this could work 
even in highly dangerous and unsettled places. I have recently seen 
proposals for IRV and Approval Voting as supposedly being helpful in 
Iraq. That would be like applying a bandage where a tourniquet is 
required. But Asset Voting could do the trick, pulling together a 
truly representative assembly. Proportional Representation without 
party lists. (But people could vote in party alignment, and probably would.)

>         Different representatives could follow different standards 
> of security
>and anonymity, thus allowing each constituent to have an optimal
>combination of the two. Representatives could send constituents an e-mail
>or letter receipt whenever they cast a dissenting vote, which would help
>to catch most cases of hacker fraud. Proxies will use different media of
>communication, different security systems... in my estimation, this
>decentralization effect should limit the severity of any potential fraud.
>Obviously, representatives have a strong incentive to keep their
>communications channels secure, since they would be in danger of losing
>proxies otherwise.

I think it can be *much* simpler. Base-level proxy assignments can 
be, if necessary under the conditions, assigned by secret ballot. 
Beyond that, though, assemblies will self-organize. They will set 
their own membership requirements. Hey, if you want to sit through 
what happens when a hundred self-important people get together and 
try to make decisions, be my guest. Take a good book with you.

What I *greatly* prefer is to see how delegable proxy actually works 
in real organizations before trying to go too far into the design of 
governmental structures.

In fact, what will design the DP governmental structures will be 
FA/DP organizations of voters. Who will then implement what they design.

>SECRET BALLOT OPTION:
>         For those who want so much anonymity that they don't want 
> there to be any
>records of who their representative is, they could formally designate
>anonymity, which would give them a ticket to participate in the kind of
>secret ballots that are now standard in most elections. (If they do this,
>they might not be able to cast dissenting votes, but that seems like a
>fair tradeoff to me.) Again, a smaller volume of votes here should have a
>positive impact on security, plus a negative impact on election cost.

I don't think that the implications of secret ballot have been 
considered enough. For there to be anonymity for *anyone*, there must 
be base-level anonymity for *all*. Otherwise the coercer will simply 
say, I want you to choose X, and, if you choose to be anonymous, I 
will breaka you face.

>THRESHOLDS:
>         Where does a proxy chain end? That is, who is empowered to 
> occupy a seat
>in the legislative chamber and actually cast votes on legislation? Let's
>assume that we want all the legislators to be able to fit comfortably in a
>designated meeting room that is not enormous. Take for example the US
>House of Representatives, which has 435 members. Assuming that we don't
>want to find a larger chamber, we would need to limit the number of people
>who hold seats in the chamber at any given time to 435 (or at most a
>number not drastically larger than 435). Thus, we would probably need to
>make use of a threshold.

I think that any meeting can and will set its own rules. In a direct 
democracy where the problems of scale have arisen, meetings can set 
rules as to who may speak at the meeting, by right, rather than by 
permission. Anyone else present may *vote*, but not take up the time 
of the meeting by speaking or entering motions, unless permitted. 
(However, one could simply ask one's proxy to enter the motion or to 
ask the assembly for permission to speak. The proxy then has to weigh 
whether or not to honor the request.)

>         At least two basic types of threshold are possible: One would be to
>simply give seats to the 435 people who hold the greatest total number of
>proxies.

Yes, that is the solution I came up with. A meeting determines what 
is the optimum size, N, then the N proxies with the most votes are 
full members. It would actually not be difficult to do.

>  Another would be to choose a minimum number of proxies held so
>that the likely population of the chamber is approximately equal to some
>target number, or that it is very unlikely to exceed some undesirably high
>number. (As Gordon Tullock suggests, it might make sense to allocate
>speaking time in proportion to the number of proxies held.)

It might, but it doesn't. What really makes sense is to make the 
meeting small enough that speaking time isn't that much of an issue. 
Do recognize that DP makes tremendous diversity possible in a 
relatively small meeting, because there might be one person at that 
meeting who represents a huge percentage of the population, and 
another with a big chunk, etc. The *least* qualified member might 
only represent a relatively small number of people....

>INDEPENDENT VOTING:
>         Let's say that I don't have enough proxies to hold a seat in the
>legislative chamber. If this feature is implemented, I would still be able
>to cast a direct vote, provided that I didn't vote by secret ballot in the
>last election. (I must either be openly represented by a particular proxy,
>or I must have openly abstained from voting for a proxy.) This way, if I
>vote independently, and I have designated a representative, the weight of
>my representative's vote on that issue will not include the weight of my
>votes (including the votes of my own constituents).

I think that I agree with the sentiment, but I've already pointed out 
the problem with secret ballot. This ability to vote directly, I 
think, is not compatible with secret ballot. (And, indeed, voting at 
Town Meeting where I live isn't secret!)

>         If I'm not voting from inside the legislative chamber, 
> where am I voting
>from? If I accept that my independent vote will be a matter of public
>record, then security should not be a significant problem, and thus
>internet voting should be fairly reliable. (Voting my phone and by mail
>might also be possible, although I'm not familiar with the security issues
>in phone voting, and in some cases mail voting might be too slow.) That
>is, anyone who casts an independent vote would be able to check on a
>public list to make sure that the vote was received as intended.

Yes.

>[...]
>P.S. As usual, I don't know how much of what I'm saying has been said
>before. Pointing out previous proposals along the same line is actually
>quite welcome; at some point in the future, I should start trying to
>catalogue these sources, eventually making a kind of annotated
>bibliography of proposed proxy methods.

And, of course, put it on the wiki at http://beyondpolitics.org/wiki....

BeyondPolitics.org is itself an FA/DP organization. (That is, it will 
recognize proxies of members, and will recognize the delegation of proxies.)