[Election-Methods] delegate cascade

[Abd ul-Rahman Lomax]

At 03:59 AM 7/23/2008, Michael Allan wrote:
>Juho wrote:
> >
> > What is btw the reason that there were no arrows forward from the two
> > leading candidates in the election snapshot picture in the 
> references page?
> > Did they abstain or were their votes (not even their own vote) 
> not cascaded
> > forward for some other reason?
>
>   http://zelea.com/project/votorola/d/outline.xht#delegate-cascade
>
>They abstained.

I have not examined this diagram specifically. But I'm getting a 
feeling that there is a conceptual problem here. I'm guessing that an 
election is taking place. So what is being collected is votes in the 
election. Votes are assigned through a cascading process. But that a 
vote is assigned isn't automatically an election or casting of the 
vote. If a candidate has collected votes, those votes are either 
exercised or not. If not exercised, they would be moot, of no effect. 
Rather, I suggest, the system is assuming that collected votes are 
exercised for the one they settle upon. So those candidates are 
considered as having voted for themselves, automatically, without any action.

I think that's an error. Consider, instead, the situation if the 
candidate casts a vote for one of his or her clients. I.e., the 
candidate is saying, "If it's not me, it's X." Which is the same as 
all other voters are saying, effectively, though possibly without 
volunteering for the position, i.e., some may be saying "It's X or 
who X chooses, directly or indirectly, not me, I'm not available." 
What if the candidate isn't available? (Does the system require 
candidacy? Mistake, I'd say. Rather, if everybody trusts X, but X is 
not available to serve, let X choose who serves. Simple.)

If X casts a vote for one of his or her clients, then X's vote total 
remains the same as if X had voted for himself, unless that client 
gets more votes.

It is really quite simple. It's proxy voting, delegable, and the 
*action* of assigning a proxy is different from the *action* of 
voting. Proxies are assigned as standing assignments, presumably. (In 
Asset Voting with secret ballot input, they stand until the next 
election, the base-level votes. Higher levels would be standing 
proxies, if the electors use DP for efficiency). You can combine the 
two, i.e., voting for an office with the creation of a proxy 
structure from the voting process, but I think it is messier, and 
less efficient. Rather, create the proxy structure, and you can use 
it for *many* decisions; and with each decision, every member can 
vote, now for the decision, not for proxies.

Thus it functions like the candidate proxy method being described, 
but the proxy assignments are set up for general purposes, then 
simply used for a specific decision. This, then, allows large virtual 
participation when only a relatively small number of trusted members 
participate; these members will contact their clients if some outside 
action is needed, of if the proxy thinks that a member's direct 
participation would be desired by the member or useful to the organization.



> > The behaviour of voter A in the example above may be quite "sincere". He
> > likes B. If B forwards his votes to some candidate that A considers to be
> > worse than C then A may vote for C directly.
>
>So A has good reason to vote for C, even though C (suppose) is a star
>having many times - thousands of times - more votes.  The reason might
>be:
>
>   i)   A knows C personally or professionally
>
>   ii)  C is recommended by A's favourite talk-show host

Yep. Any of these reasons, we don't know if they are good or bad. 
However, give people a chance to function intelligently, with a 
system that respects and uses their decisions, they will do much 
better than some of us might expect. Town Meeting government works 
quite well, as long as the scale is relatively small, it is scale 
that kills it, not citizen inability to make good choices.



>(i) If communication channels (personal or professional) give A
>influence over C, and not just knowledge (in other words, if the
>channels are 2-way) then A may be voting rationally, in the sense of
>effectively.


But A will generally be better advised to get the vote to C through 
an intermediary, someone who will listen to and address A's personal 
situation and concerns.


>(ii) Otherwise, A is a mosquito voting for an elephant!  It's probably
>not rational.  It would be better to vote for a mouse (the talk-show
>host, M), assuming that M *is* actually voting (directly or
>indirectly) for C.  Then the direct effect on C would be the same -
>she'd still be receiving A's vote.  But now A would have gained an
>open, 2-way communication channel to C, via M.

Right. You got it. Congratulations, this is actually rare. Took me 
several years of repeating this stuff over and over in these fora, to 
find *one* person who actually got it well enough to explain it. It 
seems that you found this all independently. It's not surprising, to 
be sure, this isn't rocket science, all it takes is some 
out-of-the-box thinking; the elements are all actually pretty 
well-known, it is only putting them together in this way that is new. 
And I've seen on the order of a half-dozen independent inventions 
over the last decade or so. I'd say that's a clue: the time has come. 
Until now, I was the first independent inventor to combine DP with 
the Free Association concept, but Michael, you seem to have 
recognized, at least, parts of this. Are you aware of how 
phenomenally successful the rather counter-intuitive FA concepts were 
when applied by Alcoholics Anonymous? They became a world-wide 
movement practically overnight, as such things go. What? Don't 
collect a lot of money for the cause? Don't make controversial 
decisions as an organization? Don't have strong, famous leaders? How 
in the world can you get anything done? Aren't those things necessary?

Well, they've certainly been common! Bill W., the theoretician behind 
AA for the most part, though not without quite a bit of assistance, 
studied what had caused prior movements to eventually fail even 
though they may have been very successful in the beginning. He did 
his work well. In some cases, he was pushed into his eventual 
positions; for example, he *did* try to raise a lot of money, but 
Rockefeller, I think it was, basically said that a lot of money would 
ruin the thing, and gave only enough to help Bill W. survive 
financially while he was working on the book. Wilson was 
disappointed, but later said that this had been the best thing that 
could have happened, the money would, indeed, have ruined it.

>Or A could band together with other, like-minded mosquitoes (perhaps
>by soliciting their votes) and vote en-masse for C.  Then C would be
>more likely to pay attention to their demands etc.

Except that this isn't how A would do it if A wants influence with C. 
A and the cohort would choose a common proxy who would choose C. 
Without this, A and other mosquitos would only be heard as an 
incoherent buzz and maybe a few annoying bits.

"Demands" though, is a poor description of what would happen in an 
FA/DP organization. If you don't trust someone to the extent that you 
have to "demand" something, you don't trust the person, period, I'd 
say, and you should simply choose someone you trust to, perhaps, 
sometimes, know better than you do!

And if a client made demands of me, I'd withdraw acceptance of the 
proxy, almost certainly. By the way, that should be part of any proxy 
structure in an FA, acceptance. And I'd consider acceptance as being 
equivalent to permission to contact. If I accept 1000 direct proxies, 
I'm inviting each and every one of them to communicate personally 
with me about their concerns. Do I want to do this? It would depend 
on the nature of the organization, so I would set no rules. But, no, 
generally, I wouldn't want that. I'd rather have, say, twenty clients 
who, on average, each represent twenty directly, with some of their 
clients representing the difference, to make up a thousand total. 
Much easier to handle, much more efficient.


>The votes flow individually.  Although received votes (black inbound)
>are normally carried back out along with the delegate's own vote
>(black outbound), there is one exception: if carriage of a vote would
>result in a cycle (the vote being received a second time), then that
>vote stops.  The stopped vote is held where it is (red), and the other
>votes continue on their way.  That's why the held votes deposit
>themselves evenly around the ring.  (The exception in the figure is
>the one vote injected from outside of the ring.)

I haven't examined this. What a proxy structure sets up is possible 
vote flow. Actual vote flow in an election or choice, then, is 
determined by the structure plus the actual votes in that election. 
Whenever someone votes directly, all proxy assignments from that 
person are moot, not considered further.

Imagine the proxy structure, and then mark all those who directly 
vote. Then look at all those who did not vote. Look up the structure 
to proxy assignments until (1) a member who has voted is found, or 
(2) no proxy assignment exists for a member who has voted and who 
holds a proxy, directly or indirectly.

If (1) the voter's vote is assigned as found and, since this voter 
voted, the member's vote is added to the total.
If (2) the member has abstained. In some organizations, this would be 
reported; DP could make it possible and practical to require an 
absolute majority for some decisions (even in an FA, some decisions 
must be made, about the FA itself.)

If a member has abstained, this will be public record, and anyone 
could choose to notify the member (subject to some restrictions 
because of privacy and traffic considerations). The system might 
automatically notify, and I'd make this a configurable option, 
publicly known. Someone on automatic notification could be considered 
to be more highly participatory than one who has set their options as 
"Don't bother me."


>As a consequence, casting a vote has no effect on votes received.  If
>any one of the voters in figure 9 withdraws her vote (black out), it
>will not affect her received votes (black in).  And since electoral
>standing is determined by votes received, and not by votes held (red),
>the casting of a vote can never increase ones standing.

Yes, standing is a characteristic of the proxy structure, not of 
actual votes. The problem here is that "votes" are being confused 
with "proxy assignments." I'd suggest that they be rigorously 
discriminated, except for special purposes. Proxy assignments *can* 
be used to assign some participation rights; for example the right to 
post to a high-level mailing list might be automatically restricted 
to those with a certain proxy standing, though I'm inclined to avoid 
such complications as part of the software. Let every "meeting" 
define its own participatory rules, it is much better and far more in 
accord with long-standing tradition. The software would provide proxy 
standing information, it's a mere calculation, resulting from 
analysis of the proxy table.


>I doubt Figure 9 will ever occur in a real election - it's very much
>an edge case - but if it does, it shouldn't cause any instability.
>Unless I've overlooked something...

We don't really know much about what will actually happen, beyond 
theoretical speculation. I predict that "superstar" voting, though, 
will become less common as the structures mature and people come to 
understand what they can expect from a proxy. And that's what is 
important: to develop the proxy relationships, which build 
persistence and stability in the structure; and this is why it is so 
important to separate the identification and documentation of those 
relationships independently from the problem of each decision being 
made. The proxy structure will become, I predict, quite stable. 
Decisions are content, transient, chaotic, depending on many 
variables. Whom you trust most, though, of those who participate, 
shouldn't be, for most people, a rapidly changing thing. Proxy 
relationships, if they were not initially, will become friendships, 
relationships between people who communicate directly, outside the 
formal structure.

It's been that with me!