[EM] Advanced Voting Systems

[Abd ul-Rahman Lomax]

At 09:04 AM 12/26/2008, Michael Allan wrote:
>Abd ul-Rahman Lomax wrote:
>
> > "Alienated" should be considered a relative term. Compared to what?
>
>The standard for an *alienable* medium is money.  Spend it, and it's
>gone forever.  A vote is like that.  Cast a vote, and it's gone.  Not
>quite forever, but for a long time.  (Caveat elector.)

Yes. I've come to the conclusion that the power of a single vote is 
actually a motivation for sincere voting, with the right system. It's 
been inaccurately claimed that optimal strategy in Range is the 
approval style vote, but this neglects part of the situation, most 
notably possible uncertainty in the voter's assessment of the voting 
environment and the value to the voter of a simple, sincerely expressive vote.

A voter who gets a poor outcome with a sincere vote will experience 
less emotional regret than one who gets the same outcome, facilitated 
by an insincere vote. Standard utility analysis has completely 
neglected this weighting. The result of accurately simulating it 
would be a voter preference toward the sincere vote, with distortion 
only arising when strategic considerations are *strong* -- which 
requires good knowledge of the environment.

>The standard for an *inalienable* medium is human thought, or perhaps
>free speech.  I can hold an opinion to myself, and I can also express
>it.  Either way it still belongs to me.  If new information comes to
>light, I can always change my opinion, and even "take back" what I
>said.  Compared to that, voting falls short.  I cannot take my vote
>back and I cannot change it, not unless the voting is continuous.

That's right. However, a compromise is found with short-term Asset 
Voting, where the voter may vote with *maximal* sincerity, for a 
candidate the voter knows personally, or relatively directly. This 
may indeed flip, from some misbehavior, but it is far less likely to 
do so, to result in serious discord between the electorate and the 
Asset electors. Then, the actual assembly seats are elected by a 
process that *may* be reversible.

There are two paths to reversibility. One would be a process whereby 
an elector revokes a vote assignment; the other would be a process 
whereby the elector directly casts a vote on a matter before the 
assembly. The latter process, if in place, would make discrimination 
possible: I trust the seat I voted for, *generally*, but she is a bit 
wacko on this subject!

The former process would take place when there is a major failure of 
overall trust.

I would build some hysteresis into the process. If the electors can 
name proxies for the exercise of their votes, the loss of default 
voting power for a rejected seat may be immediate, and all that 
happens as continuing harm is some wacky representation in 
deliberation, which is pretty much harmless. A non-seat proxy could 
exercise the votes of those who have withdrawn.

(If voting is a matter of entering a user name and password on a web 
site, a voter could entrust their user name and password to someone 
else; revoking that by changing the password, and the real elector 
could at any time request a password being sent to their email 
address, this is standard process, actually. It's pretty secure. And 
the few exceptions could be dealt with by a process that verifies 
actual identity; the real elector would file a statement under 
penalty of perjury that they have lost access illegitimately, and the 
voting right would be suspended until it was resolved, which would 
normally be quite quickly. And it would be fraud to exercise the vote 
of an elector without the permission of the elector, i.e., after 
notice that the permission was withdrawn. This is, I think, pretty 
standard proxy law.)

>(The distinction is important in social theory.  Alienable media are
>associated with intrumental/strategic action, and non-alienable with
>communicative action.  Not sure if there's anything in that...)

Makes sense to me. This is part of FA/DP theory.

> >> But if the votes were open to recasting in real time...
> >
> > It's Delegable Proxy. That is the principal difference between Delegable
> > Proxy -- which is continuously reassignable -- and Asset Voting. ...
>
>Can you point me to the original description of DP?  I'm looking for a
>source I can cite.

Well ... I'm perhaps the number one source on it. What do you want me to say?

Seriously, there was an article on Wikipedia. It was deleted 
primarily due to lack of published sources, confused by 
misunderstanding my involvement in the article. I hadn't actually 
edited the article in a long, time, once I realized the issue of 
conflict of interest; I'd originally edited the article on Liquid 
democracy, which later was changed by someone else to Delegable 
proxy. Now, in fact, there are some sources, but it wasn't called by 
that name.

You might look at:
http://en.wikipedia.org/wiki/Wikipedia:Articles_for_deletion/Delegable_proxy

The article on Delegated Voting wasn't deleted, but was Redirected to 
Proxy voting. Because it wasn't deleted, you can still read the article at

http://en.wikipedia.org/w/index.php?title=Delegated_voting&oldid=197094901

(Wikipedia is quirky. Merge was clearly the appropriate action for 
the Delegable proxy article, if there was *some* value to the article 
and if it might become notable someday. Merge is an editorial 
decision, and is often accompanied by much less fuss than a Deletion. 
Both actions, Deletion and Merge, leave the article in the database, 
the only difference is that with Merge, the article still exists, but 
if you enter the name of the article, you are redirected to the 
target article. At the top there is then a link back to the original 
article ("redirected from ... [link]). Then, in History, you can see 
all the old versions of the article, and the Talk page remains as 
well. Absidy and I argued long and hard that Deletion should be 
reserved for true garbage. But there are fanatic "deletionists," some 
of whom showed up in the debate I linked above. Material in a merged 
article is only seen in History, and that isn't searchable by the 
googlebots, so all the "vanity" arguments are totally moot. It's a 
common accusation in deletion discussion when the creator or editor 
of an article has something to do with the topic. It's really moot, 
and there are Wikipedia documents saying so, but the argument is 
still made. The question should solely be the article itself, not who 
made it. Deletion isn't punishment, or it shouldn't be. The deletion 
discussion, though, did come up on Wikipedia because Absidy did 
propose delegable proxy for Wikipedia, which is an application crying 
for it if I ever saw one. The application was either (1) roundly 
misunderstood, and that is clearly true. Delegable proxy wasn't 
really being proposed as a voting method, but as a *representational* 
method, for bidirectional communication. (2) it *was* understood, by 
those who would feel threatened if the average Wikipedia editor, who 
doeesn't have the time to get involved in the very tedious debates 
that go around and around in circles on Wikipedia, were empowered, 
which delegable proxy would do. I vote for the former. It's 
ignorance, not avarice for continued power. They really don't 
understand, which has been my general experience. It seems to take, 
for most people, at least a year of exposure for the idea for it to 
*start* to sink in.

I might ask to recover the deleted Delegable proxy article and its 
Talk page. There was some stuff added there which may be of use. A 
copy of the article was ported to the Election Methods wiki:

http://wiki.electorama.com/wiki/Delegable_proxy

The BeyondPolitics wiki that was linked there is largely gone, due to 
bugs in the update process for the TikiWiki installation that existed 
there. I haven't had -- or haven't taken -- the time to fix it. 
Meanwhile http://beyondpolitics.org goes to a new MediaWiki 
installation, with little content.

The wayback machine for BeyondPolitics.org should have most of the older stuff.

http://web.archive.org/web/*/http://beyondpolitics.org

There is also James Armytage-Green's page on delegable proxy:

http://fc.antioch.edu/~james_green-armytage/vm/proxy.htm

Where I have differed from his work and the work of others is in 
having a vision of how to get from here to there.

Delegable proxy can be used within what might be called a judicial 
system, or advice system, where what is important is the 
trustworthiness of the advice, and, as pointed out by Montesquieu 
long ago, mixing advice (judgement) with power causes the corruption of both.

Hence my concept of using Delegable Proxy within Free Associations, 
which are organizations which do not collect power (beyond a certain 
kind of power that is represented by being a trusted advisor, a power 
that is easily lost by giving bad or corrupt advice). Such 
associations, I believe, could very easily function to negotiate 
broad consensus very efficiently, even on a large scale. Delegable 
proxy sets up what should be an efficient, intelligently filtered 
bidirectional (inward-outward) communication network, where every 
member chooses the filter.

Some writers have been concerned about proxy loops, but I consider 
them unavoidable, in the first place (the alternative implies a 
superproxy, one person who represents everyone, which *might* be 
dangerous!), with the harm limited to lack of representation in some 
discussions, a situation which is easily remediable if the members of 
the loop want to remedy it.

There is no harm at all if at least one member of the loop participates.

Using delegable proxy within Free Associations thus finesses the 
security problem, and other problems having to do with the untried 
nature of the system. It is actually only a formalization of what 
already happens to some degree. Those who participate represent those 
who don't, and when a serious imbalance results, more participate 
(which can create quite a mess!). DP prevents lower participation 
from becoming a mess. It self-adjusts.

Then, if a Free Association, which is designed with rules that remove 
impediments to practically unlimited growth -- beyond the normal 
impediment of inertia and cynicism, which will disappear as FAs of a 
certain size arise -- becomes large enough, its coherent advice 
becomes a powerful force. If it cannot find coherent advice, the 
influence of its factions tend to cancel out. Thus it only has power 
if it can find some level of consensus.

With large FA/DP organziations active, existing voting systems become 
simply methods of ratifying the consensus. Plurality works! (But why 
not use better methods, some of them are cheap!)

And, of course, there is Asset Voting (Warren Smith). Also called 
Candidate proxy (Mike Ossipiff and Forest Simmons). Not only does 
this implement a kind of secret-ballot delegable proxy, an FA/DP 
organization of electors could serve to efficiently negotiate the 
vote reassignments needed to create assembly seats, and, as well, to 
maintain communication between electors, and to advise those with seats.

I do not consider "delegated representation," where the votes of 
proxies are supposed to represent the opinions of the clients, to be 
a good thing. Rather, it is far better for clients to entrust proxies 
with the power of independent decision; besides, the delegated voting 
concept doesn't scale well. A proxy representing 100,000 voters 
should vote what he thinks the clients want? Obviously, client views 
are important, but the whole point of choosing someone trustworthy 
for the office is to choose someone who is in a good position, 
participating in deliberation, reviewing as much of the evidence as 
possible, to make a trustable decision. Otherwise, we would simply go 
to mass voting by robotic representation.... *That* is an idea that 
would be highly dangerous, easily manipulable through mass media, etc.

Delegable proxy, where what is delegated is the representational 
power, not some specific issue position or even a set of positions, 
is as safe from this as I can imagine. And so would a mature Asset 
Voting system, which could become, effectively, delegable proxy, 
while still retaining the power in the hands of the electors, who 
*can* vote continuously, in effect. But they will rarely need to do so....

> >                                   ... people understanding that if they
> > give their votes to a massively popular politician, they get far less than
> > they do if they give it to someone they can sit down and talk with on
> > occasion. In the latter case, they gain a communications channel, in the
> > former, they simply support an image they have been presented.
>
>Do you have a citeable source for that, too?

It would be me. I wrote that. To my knowledge, I'm the only one who 
saw the implications in Asset Voting, that once votes would not be 
wasted, even if given to a "candidate" who only gets one vote (the 
voter himself?), voters can make the choice without any constriction 
at all, beyond the natural constrictions of the situation.

(Under "difficult conditions," it might be necessary for candidates 
registering to be electors to secretly name a proxy or the like, to 
receive their votes if they get less than N votes. I'd like to see 
this to deal with the problem of unexpected elector incapacity, but 
it also might be needed when there is massive possible coercion. 
Then, electors who got less than N votes would simply know that they 
didn't get that number, they would not know *who* "betrayed" them. 
And N is large enough that the resources of the state can be devoted 
to protecting the remaining electors. It looks more like an extended 
proportional representation system here, or what is called Candidate 
List, where candidates publish a list of vote reassignments to be 
used if they aren't elected. But I don't think that, for example, 
such modifications -- which create new security problems, who watches 
the watchers? -- would be necessary in the U.S. for example. Iraq, maybe.)



> >>   [1]. Lewis Carroll.  1884.  The Principles of Parliamentary
> >>        Representation.  Harrison and Sons.  London.
> >
> > Carroll was the first I know of to propose votes transferable by the first
> > preference candidate...
>
>And if I understand, that entails *recursive* transfer?  I'll need a
>source for that, too.  I'd better read Carroll...

No, it doesn't imply delegable proxy, itself. The idea was that the 
transfer was under the direct control of the candidate receiving the 
vote. I've proposed that electors (candidates with votes, that's all, 
who may recast these votes publicly) might use delegable proxy to 
facilitate their own process, which becomes more necessary if there 
come to be *many* electors, but I wouldn't do much more with proxies 
than allow electors to name a proxy, not delegable (possibly). It 
might be delegable, but what's important is that the elector can 
revoke the single assignment at any time.

When the number of electors becomes very large, delegable proxy, used 
for direct votes in the Assembly, could become quite useful.

But I wouldn't write it into the first applications. Simple: Asset 
Voting to create a proportional representation assembly. Probably 
fixed term as a first step, no revocation, etc. The elector body 
still exists to advise the seats. No direct voting at first. But it 
becomes an obvious reform that would decrease the distance between 
the Assembly and the people.

And that's the problem that I'm most concerned about addressing: the 
sense of distance and separation between people and government. 
Government, too easily, becomes "them." It should be "us." When we 
have open channels for participation, which a mature Asset system 
with small-scale selection of electors and large-scale concentration 
using delegable proxy, into an Assembly where each seat has fixed 
default voting power, it would, I'd predict, be "us."

Proxy democracy has been suggested, Absidy found some references, for 
political application, about a century past. The idea was that each 
member of a city council would have votes on the council proportional 
to the votes they received. While it's an interesting idea, it 
suffers from a very serious problem. The focus of power can become 
too great. When there is an excessive focus of power, there is an 
attractive target for corruption. By distributing power among a large 
set of peers, corruption becomes far more expensive.

The possibility of corruption has been suggested as a hazard for 
delegable proxy, but what these analyists have missed is the 
supervisory nature of the structure, the clients supervise the proxy. 
They can't control him, but they can advise him, and they expect 
responses that make sense. If they don't get those responses, they 
can and will change their proxy assignment, or if it's just on a 
single issue, they will vote directly.

Imagine some corporation, desiring some outcome, bribes a prominent 
proxy. In an Asset system, if the "proxy" has a seat, this gets them 
one vote, it would be expensive, but maybe it might be worthwhile 
under some circumstances. However, there is a risk: the proxy says to 
them, "Thank you very much, I'll enjoy the money." And he says to his 
direct clients -- who would, themselves, be highly trusted electors 
--, secretly (he knows them personally and trusts them), "I've been 
offered a huge payment and I can use the money. So, be advises, I'm 
going to support this corrupt proposal, I'll try to convince you, in 
our public communications, that it's a good idea. I don't think so. 
If you don't like it, I'd suggest voting directly."

The corporation gets a mouthful of hair. Alternatively, perhaps the 
bribe is conditional on success. (Not a great idea for the seat to 
accept it, how is he going to enforce it?) The high-level electors 
will still smell a rat, most likely. They are massively trusted, it 
won't be a very rare thing for them to take an interest in actual legislation.

No, I think corrupting a delegable proxy system, when the massive 
concentration of power in a single stage is unusual or nonexistent, 
would be very difficult, so risky and expensive that corporations 
would be better off, it would be more profitable, if they spend their 
money trying to better serve the public. Which is what we want, right?

> > I haven't been able to find the original pamphlet yet, it's expensive to
> > buy the collection it is in. Eventually, I'll get it.
>
>My library has originals.  If I can make a copy, I'll send it to you.

Please. I'll send you my address by separate mail. Or you can email me scans.

> > I don't know that he realized the deeper implications, that this tweak to
> > STV could become the whole show, and lead to quasi-direct democracy. Once
> > there are electors holding votes, and those votes are cast publicly, the
> > problem of scale that afflicts direct democracy and is generally 
> considered
> > insoluble, is solved -- or reduced by an order or by orders of magnitude.
>
>None of this has been explored, I don't think.

My sense is that my work has been original. Delegable proxy, itself, 
was widely and independently invented, over the last decade or so. I 
was working on it maybe thirty years ago, but not in writing. 
However, when Warren Smith showed me his Asset Voting system, I 
immediately recognised it as a form of delegable proxy, and I already 
knew that DP would work best when the proxy assignments were 
small-scale, where the proxy and client could easily communicate directly.

And who would want to be represented by someone who one can't raise 
on the phone, or sit down and chat with in an office or elsewhere? 
Not me! Not if I have a choice. Sure, as an ordinary voter, I won't 
ordinarily be able, just from that, to easily talk to my 
representative in an assembly for, say, the State of Massachusetts. 
I'll have to be content with a staffer. (By the way, this is a bit 
how the existing system works, particularly when the staff is good 
and really does digest and understand what the citizen is saying. 
Delegble proxy would incorporate this kind of functionality, but 
would be a bit more efficient at the movement of information. It can 
be very frustrating to call up the office of a representative and 
meet with a blank wall. Very polite, to be sure, but clearly no 
understanding. I once wrote to Jerry Springer, who had stated on the 
radio that direct democracy was impossible, giving the classical 
reason (scale). I pointed out that this was not exactly true, because 
*chosen* representation was possible, and could accomplish the 
necessary reduction of scale, all it takes is some layers so that 
deliberation is possible at all levels. I got a nice mail back from 
some staffer thanking me for my interest. No response that showed any 
understanding at all.

My proxy would understand! And would know whom, on the next level up 
(levels informally appear with DP), to talk with, perhaps, but not 
limited to, his or her own proxy. And so on. An idea would rise until 
it meets *cogent* objection, which would then go back down, always 
through relationships of rapport. I'd get an answer, quite possible 
an answer that would show me why my great idea wasn't. But if my idea 
is really sound, I might be able to answer the objection, convince my 
proxy -- if I can't convince my proxy, who in the world *can* I 
convince? -- and thus the idea goes back up, this time with a 
persistence marker which will slightly increase attention.

It's little less than mass intelligence, a device for "thinking" on a 
large scale. And that's a lot more important than voting! Voting is 
just about the outcome....