At 05:18 PM 8/20/2008, Juho wrote:
There is a difference between methods where only voters can modify their votes at any time and methods where the candidate that got some votes can redirect these votes. The latter case may cause larger and faster changes. And such changes may lead to reactions also among those voters that gave their vote to this person (if the voters do not like the change). These properties may mean higher instability.
One of the reasons why I've avoided speculating much about things like "stability," with delegable proxy systems is that we have very little experience with them. However, they shouldn't be *terribly* different from the behavior in relatively small organizations of standard proxy. (Proxies have almost always been delegable, in theory, but the use of delegable proxy to handle large-scale organization would result in *routine* delegation, whereas normally, under smaller-scale situations, a proxy is given to someone expected to participate directly, and delegation only takes place in the case of unexpected incapacity.
However, I don't think that Juho's speculations are particularly likely as effects. A great deal depends, though, on the conditions under which the proxy network is formed. I'm going to start out, though, by noting that present representational systems, if run as fine-grained STV, should produce similar results if the electorate is awake and informed, as, say, Asset Voting. Asset Voting simply allows the sorting out that happens in campaigns to happen more spontaneously; Asset should, in fact, make campaigning not only unnecessary but actually rejected.
Currently, representatives (in an STV assembly, approximately) are already trusted with our "vote." I.e., if we consider an assembly to represent the people, those elected cast our votes in it. They can "redirect" those votes, i.e., vote differently than what we might have expected from them when we voted for them. And, in fact, this is a good thing! For, quite likely, we would ourselves change our minds, sometimes, if we were to participate in committee sessions that consider evidence and alternatives, negotiate for broader consensus, and so forth, all the things that real representatives do.
In other words, we already have a system where "candidates" -- certain ones, those elected -- can redirect their votes. Does it lead to instability? Sure. The kind of instability we want, intelligent, flexible decision-making that is not bound to some prior agenda.
We have discussed in the past what an Asset Voting Assembly with a penumbra formed around it of electors, public voters, those who received votes directly from the electorate in a secret ballot election, and with election to the Assembly being a standing result, subject to revocation in two ways. One is that electors could vote directly on issues before the Assembly, including Assembly rules, regardless of how their "seats" vote. These votes, then, would be deducted from the seat votes according to the percentage of such "free" votes cast from electors who elected the seat. My prediction: it would be rare that such votes would shift an outcome, but the fact that they would be possible could make it more difficult for those with seats to decide in a manner wildly different from the position of the electors; one of the jobs of a seat would be to keep their electors informed and well-advised. This is really, *almost*, direct democracy, with a twist: voting is direct, except proxy voting is allowed and, indeed, encouraged. And, of course, deliberative rights are limited to those with seats, thus keeping the scale of the Assembly manageable.
The other way that votes could be "retracted" would be revocation of the assignments that create a seat. I don't, expect, again, that this would be common; though an obvious application would be the ability of such a system to replace a seat that becomes incapacitated or resigns. Theoretically, even one vote of withdrawal would result in a loss of the quota, but I'd probably want to see hysteresis built in so that minor fluctuations didn't have major effects. I'd probably want to see a new seat put together from votes, before the old seat is replaced. There is no harm ad-interim because the electors involved could still control all the votes. (It should be allowed to name a proxy to exercise elector votes; thus, for voting purposes, almost all the votes of a seat could be controlled by a single person, not the holder of the seat.)
I think that many of the problems we'd expect simply would not arise, but others would, and the system I've described would be self-modifying, by permission of the electors. The electors would be a broad group, and I've mostly thought of them as volunteers (but I've also gone a little way down the road of considering what compensation of electors might look like ... not for now). If the system settles as I'd expect it, most voters would be voting for someone who isn't going to be directly elected. They will vote for someone they personally know, quite often, someone whom they can talk with. So at the base, we would be setting up a system that collects public views, in one direction, and informs the public, in the other. And then the process of amalgamating the votes would encourage that communication to continue, up the hierarchy, to seats. As a voter, you would know who you voted for, and you could talk with this person. The person wouldn't know if you voted for him or her, for sure, but *it wouldn't matter. Remember, you vote for someone you can talk with *already*. And from there on up, your "proxy" is handing over an identified vote, the seat knows exactly who elected him or her, and presumably will listen. And it is this communication that's crucial, that makes this kind of structure into an intelligent decision-making system that takes full advantage of the resources of the entire society.
I actually think that it would be quite stable; that when it reversed direction rapidly, it would be because it became obvious that this was the best thing to do. If it is routinely making decisions by supermajority (which a good and appropriately scaled Assembly should do, not as a rule, but as a desirable condition), then small shifts in seat assignments, or small numbers of direct votes, aren't going to make much difference. An Asset Voting assembly would make a party structure unnecessary for election purposes, but parties would still exist, I'd think, but they would simply be caucuses of seats with similar ideas. Currently, the electoral system in the U.S. fosters two major parties that strive for an appearance of difference, and intraparty politics tends to move each of the parties away from the center (of the whole electorate, toward the party center). The parties tend to form around two "wings" of the electorate, and thus they tend to stay more or less in balance. This, then, leads to instability, as "course correction" by changing the party with a majority is too broad for finer control. You would never engineer a control system this way, it simply grew as a response to conditions.
---- Election-Methods mailing list - see http://electorama.com/em for list info
