[EM] proxies and confidentiality (six variations)

[Abd ul-Rahman Lomax]

At 07:34 AM 3/30/2006, James Green-Armytage wrote:
>         Overall, I'll say that delegable proxy, or indeed anything 
> approaching
>direct democracy, is not a very natural fit with very strong
>confidentiality. Perhaps accommodations can be made to provide this, but
>the system probably works best in a society where this kind of
>intimidation is not a major problem.

This is one of the reasons for my proposal to start DP (Delegable 
Proxy) with NGOs, and in particular with Free Associations, which, as 
I define the term, implies no collection of central power beyond the 
power to persuade by generating consensus (which can be very powerful 
indeed, but which is not necessarily centralized); so it is 
relatively safe. This is possible in any reasonably free society; 
indeed, it is my opinion that it is possible in China now, 
immediately, for another major aspect of FAs is that they don't take 
positions on issues of controversy. So an FA would *not*, as an 
organization, take any position likely to offend the dragon. Instead, 
it would work *with* the dragon, it would simply be patriotic 
citizens, for the most part, communicating and cooperating toward a 
better China. So far, I've seen no sign that the central power in 
China wants to discourage such communication; it simply wants to 
avoid the chaos of uncontrolled mob power, it was quite frightened by 
Tiananmen Square. (And I don't know how much the readers of this list 
know about that event: there really was a massive rebellion of the 
citizens of Beijing, and it took bringing in troops from other areas 
of China *who did not speak the local language* to quell it. Local 
troops sympathized with the workers, as one might expect in a 
Communist country!)

It is my opinion that DP organizations will select for 
trustworthiness and that such organizations will be quite capable of 
dealing with local conditions and all the necessary subtleties. They 
really will create democracy without offending the dragon. But, of 
course, it is for the Chinese to work all this out. I'm working on 
FA/DP here in the United States and for the English-speaking community.

Direct democracy, in the past, has required open participation. I've 
not seen a town convert from town meeting to elected mayor-council 
based on fear of intimidation. Rather the reason is that the meetings 
have become tedious and tendentious, and mayor-council is perceived 
as more efficient. As it is, compared to raw direct democracy without 
proxy voting and without appropriate participation rules.

Given that proxy representation is a common-law right, it is amazing 
to me that I find no historical record of attempts to apply it to 
government in a direct democracy. Proxy representation *is* used, for 
example, in the New York legislature, where representatives may cast 
votes in committee by proxy, if I've got it right. I'm not convinced 
that that application is a fair test of proxy representation, it is 
happening at a high level in a fairly dysfunctional system, in my 
opinion, that leaves most people unrepresented.

Anyway, I think we need to test DP in NGOs before we can seriously 
propose it for governmental use. Fortunately, this is easy, at least 
in theory. In fact, it is quite hard to get people to try it. Why?

I have my theory: it is a variation of what I called, in the past, 
the "Lomax effect," more descriptively "the persistence of 
inequities" effect. If a power structure is inequitable in some way, 
restoring it to equity will remove power from those who currently 
enjoy excess power. They will see this as a threat and will act to 
prevent it from happening. And, by the terms of the problem, they 
have excess power. It takes no evil conspiracy to explain this. Have 
you ever noticed that those who are active and hold positions in an 
organization often believe that they know and understand more about 
how the organization should function than do the rank-and-file? I 
have seen again and again such people fear that if the rank-and-file 
gain power, they will ruin everything.

And indeed they might. Which is why power shifts must be done very, 
very carefully. Delegable Proxy is a specific answer to this problem, 
by concentrating trustworthiness. It is representative democracy 
without elections, but through free choice. It is a solution to the 
problem of scale in democracy. If it is first implemented in FAs, the 
problems of corruption, as one example sometimes raised as an 
objection to DP, do not have to be addressed until a time when we 
know much more about how DP will actually work.

>         Also note that there is sometime a tradeoff between 
> confidentiality and
>the transparency, or integrity, of the process. When information about who
>voted for whom is not available, it becomes harder to verify that the
>final count is correct.

Absolutely. In an open DP system, with an open assignment list, 
anyone can at any time analyze the structure and verify vote counts.

However, DP with base-level secret ballot can be pretty safe and 
almost as good. To understand this, I think we have to look at 
exactly what is and what is not dangerous, and what risks can be, and 
routinely are, accepted in secret-ballot systems.

In particular, we have no problem with identifying the members, for 
example, of a city council. Yet these are the people who are actually 
voting on matters of substance, who, one might think, would be most 
vulnerable to pressure. But they are not, of course, for a simple 
reason: there are a relatively small number of them, so police 
protection can be provided for them at an affordable cost. You could 
not provide this level of protection for someone who only represents 
a few neighbors, nor for individual voters.

So the solution is fairly simple: a governmental DP system under 
difficult conditions would be secret ballot until a certain number of 
votes have been collected. Essentially, the secret ballot system 
would function in a hidden way until it collects enough votes to 
qualify a candidate for open identification. Yes, this requires a 
trusted mechanism. A trusted mechanism is essential to secret ballot, 
if it is not to be merely one way in which a dictatorship pretends to 
have the support of the people.

Consider Iraq. Those who actually have power there, I will not 
presume to name them, if they are interested in democracy, could set 
up a system where every citizen is invited to name a proxy, the 
person they most trust. This would be done secretly and privately, 
and in the process the citizen would be asked if they were being 
coerced. If they were being coerced, their vote would be discounted, 
but not in a way that could let it be traced back to them. I won't 
give details, it can be done, such that, if people trust that they 
will not be betrayed, the vote counts will not betray them 
individually, and would only indicate to a coercer that something 
went wrong. And, of course, with a high level of credible reports of 
coercion, that coercer might be arrested....

Such a single assigment can create loops, but a community, 
understanding this, can decide to give enough votes to at least one 
rep to put that rep over the participation threshold for the open level.

The point of such a secret-ballot designation is that it would select 
a representative who actually did represent a significant 
constituency. He would largely know who they are, not in specific 
detail, but he would be local and would generally be relatively 
accessible (though, once elected, security concerns might make him 
much less accessible).

And then the rest of the system functions openly. Iraq is a tribal 
society, largely, and so what we would see at a high level would 
largely be tribal representatives, but chosen freely by their 
communities, rather than through pure power and intimidation (as well 
as more legitimate tradition and respect). The key to delegable proxy 
is that it brings everyone to the table. In open DP, it is literally 
everyone who cares at all to participate. (My own proposals would 
allow direct voting at any level, but not direct full participation: 
the right to take up everyone's time must be limited in a large 
organization, or it breaks down. That's what happens at Town Meeting 
as the town grows.)

>         Variation 1: If you do not choose the secret ballot option 
> (mentioned in
>the proposal linked to above), then your choice of proxy is recorded on a
>public list, which you or anyone else can view at will.

As you have realized, this is transparent to a coercer, who will 
simply require that the voter elect to be open. If you need secrecy, 
and you need a simple system, it must be secret for everyone *except* 
for people whom the society can afford to protect. Which means a 
relative few under difficult conditions.

Under good conditions, coercion has become rare enough that police 
resources can be brought to bear on the few exception.

I don't think, for example, that there would be any significant 
problem with open DP in the U.S. It is enough that extortion is 
illegal. Someone trying to coerce a voter would be taking a huge 
risk, for a very small gain. One vote is just one vote. And if you 
try to coerce thousands of votes, discovery becomes practically inevitable.

But, again, we can find out. We don't have to jump off the cliff. We 
can do it outside of government.

>         Variation 2: Base level voters can choose between secret ballot (no
>direct vote option), having their vote listed publicly, or having their
>vote known by the election authority but not listed publicly. Proxies must
>have their votes listed publicly.
>         Drawbacks
>* Doesn't really solve the problem of people being pressured to vote for a
>particular candidate.

Yes. This leaves the same problem in place.

Direct vote, in my view, is desirable but not essential. It is enough 
that relatively low-level proxies have direct voting privileges. The 
people in general will have good access to those people, because they 
will not represent many. Through them the people will have access to 
the top level. Yes, blocks of opinion below a threshold will be off 
the radar. But a society which needs secret assignments has much more 
serious and urgent problems to deal with. It's enough, under those 
conditions, that every major group be represented proportionally.

Asset Voting, by the way, accomplishes much the same thing, producing 
a peer assembly. Ultimately, though, I think DP is simpler, and would 
transition quite easily into full, open DP. Which is much, much more 
than an election method.

>         Variation 3: As long as you do not declare yourself as a proxy, the
>identity of your proxy is known to the election authority, but it is not
>publicly available. You can request a paper receipt for your proxy
>delegation, should you be unsure that your vote is being recorded
>accurately.

And, as you went on to note, this does not solve the secrecy problem, 
for several reasons. If you request a receipt, you can then prove 
that you voted as demanded. It must be impossible for you to prove 
that, or small-scale coercion can work. Essentially, all 
secret-ballot designations must be secret, and everyone must 
designate by secret ballot. Secret-ballot DP elects representatives, 
it cannot be ongoing. However, those representatives will ideally be 
serving on a relatively small scale; the number of votes they have 
received will be public record, but not who gave them those votes, 
and their subsequent choice of proxy will be public. And changeable 
at will. It is a direct democracy of those elected in the first round.

Simple open DP is *much* more straightforward. I think it has 
implications that few, so far, have noticed, though I keep harping 
away at them....

>Drawbacks:
>* Although the information about base level is not officially public, it
>is known, and thus can potentially be leaked.

The problem only exists if you want to have specific interactivity 
between base-level voters and the proxies they choose, and if you 
want to allow direct voting on issues. Clearly, if you allow direct 
voting, you would need to know who chose whom, so that the votes 
would not be counted twice for some. Direct voting as a routine 
matter, on all issues, which would be one possible solution, with the 
proxies only serving to negotiate consensus, is a possible solution, 
but it does require frequent voting with everyone participating, 
which, in my view, is largely impractical and even inadvisable, for 
reasons that I don't think I have to elaborate.

>* It is still possible for someone to put pressure on you to declare
>yourself as a proxy, and thus make your delegation a matter of public
>record.

Those who act, subsequent to the secret ballot, must be willing to be 
public, and perhaps should be declared candidates. But that they are 
willing to be public should not be enough for their identities to be 
confirmed by the voting agency. All that will be announced by the 
agency is the identity of those receiving a threshold level of votes, 
and the number of votes they received. In one implementation, even 
the agency would not know who actually voted. So pressure on a voter 
to declare consent to serve will not be sufficient to make that 
voter's actions visible. The voter would have to actually *receive* 
the required votes, and since the ballots would be secret and thus 
presumably not coerced, what would have happened is that the coercer 
would simply have created a candidate with votes. Coercion is not 
necessary for that!

However, there is one problem, which is loops. In standard DP, loops 
are easy to deal with. It is commonly overlooked that loops in full 
DP are necessary, for if everyone designates a proxy, then a 
top-ranked proxy designates someone of a lower rank. The 
relationships could get quite complex; but what has happened in this 
case is simply that a top "official" has designated a stand-in, to 
serve in his absence, which is what a proxy is, anyway. But in secret 
ballot DP, how could we deal with "wasted votes"? i.e., votes that do 
not end up being assigned to a candidate with sufficient votes to be 
identified publicaly and to act at a higher level.

In open DP, one simply informs those "trapped" in such a loop. Any 
member of the loop may break or at least enlarge it by designating 
someone outside the loop. Loops are harmful only when they result in 
lack of representation, which is much less of a problem in open DP 
because all can review the record and vote directly, and it is simple 
to join a proxy group by designating the proxy.

(Actually, in the FA/DP implementations I'm working on, proxies must 
be *accepted* to be effective. A proxy is like a professional, and 
the constituents are like clients. Indeed, I'm coming to use the word 
"client" for the one who is served by a proxy. Anyway, it may not be 
so easy to just grab a proxy when needed. The proxy must agree to 
serve, and service means much more than voting.)

>         Comments: Perhaps this is what Abd had in mind? Seems like 
> a reasonable
>compromise if there is some worry about coercion, but not an intense
>amount of worry.

I think it is easy to adjust the level of secrecy to the need. FA/DP 
implementations will almost certainly not require secrecy, though, 
paradoxically, secrecy becomes much easier to obtain with little or 
no harm. This is because FAs don't really make binding decisions; for 
FAs, what is important is communication and measurement of consensus, 
and when you are trying to measure consensus, a few votes here or 
there don't really matter. And even if someone fakes a whole slew of 
votes, even a *majority* of votes, they end up with a handful of 
straw, because nobody will actually salute the flag they raise. FAs 
don't collect power to be grabbed through fraud. Rather, they 
communicate, and someone who creates a whole fake caucus in an FA in 
order to appear strong will be spending tremendous effort for a tiny 
gain, if any gain at all.

So participants in FAs can be anonymous. The proof that one is not 
dealing with an illusion of a large number of people will come when a 
caucus has gathered the support of supposedly 10,000 people to 
contribute to a PAC, with polls indicating they will contribute an 
average of $50 each. And what actually shows up at the PAC is what 
the fraud behind the show can afford to personally contribute.... FAs 
do not leverage power, which is where they differ radically from 
traditional organizations, and which is why they may be able to 
accomplish things that have so far eluded us. The power remains with 
the people, until they individually choose to collect it and apply it 
for a specific purpose. That purpose may create a traditional 
board-controlled organization with a charter as chosen and supported 
by the voluntary donations. And there is a whole history behind this, 
it does work, quite well.

>[...]
>         Variation 5: Non-proxy voters vote by strictly secret ballot, thus
>ensuring anonymity, but also losing their ability to cast direct votes.
>(If it isn't known who your proxy is, then your vote can't be subtracted
>from his usual total if you vote directly, and thus direct voting can't
>work.)

Exactly.

>         Drawbacks:
>* Direct voting can't work
>* It is still possible to pressure someone into declaring themselves as a
>proxy and voting for the favored candidate.

Not if the declaration alone is not sufficient to allow participation 
uplevel. The wasted-vote problem does exist, possibly, but there are 
ways around that, which can get more complex; again, I won't get into it now.

The major thing to realize is that secrecy is not a normal condition. 
It is only necessary under circumstances of great instability and the 
breakdown of the social fabric, where lawlessness is rampant. We 
institutionalized the secret ballot, and I suspect it has been 
useful, but I have never heard of an attempt at vote coercion at a 
Town Meeting. I've never seen a recorded vote at Town Meeting, but 
anyone who cares can see how people vote. Cummington is a very 
liberal town; from what people say publicly, you'd never know that 
Bush got one-third of the vote here. A town resolution to advise 
Massachusetts representatives to act to remove the national guard 
from Iraq passed with, I think, only one or two dissenting votes. I 
sensed no pressure at that meeting. But, then again, there may be 
subtle pressure. But there will always be subtle pressure, and 
periodic secret ballot validation of what is going on may well be 
advisable. Indeed, if FA/DP in Non-Governmental Organizations does 
what I expect it can and will do, the existing structure does not 
have to change at all. FA/DP will have organized the electorate, 
which can then manage the existing electoral structure. You can do 
practically anything you want if you can find consensus (including 
change the constitution, which really only takes a distributed 
plurality), and if it is working correctly, there is no harm in 
periodic secret ballots for the selection of officers. I'd expect 
them to rubber-stamp the extra-governmental consensus (or, if there 
is no consensus, to match quite closely the FA/DP poll results). If 
they did not, something is drastically wrong, and we'd be grateful to find out!

(For close match with secret ballot, the FA/DP organization would 
require identity validation, which is relatively easy. We would know 
if it was necessary by such results. There would also be a skewing of 
ballot results from the difference between the FA voting body and the 
registered and actually voting public in the secret ballot, but that 
should be relatively easy to compensate for.)

>         Comments: I can't really recommend this, as the direct vote 
> option is one
>of the major benefits of the system.

Absolutely. Many critics of DP don't seem to get that what we are 
recommending is direct democracy, but with something added: the 
option to vote by proxy, plus, with DP, the routine delegability of 
such proxy assignments. Nobody has to exercise that option, but 
everyone who does so potentially gains flexibility and access.

>         Variation 6: Give up on delegable proxy altogether and just 
> use STV with
>the entire nation as one district.
>         Comments: This, the actual legislators are the only people 
> whose votes
>are formally recorded; hopefully at least the members of the legislature
>can be protected from coercion! If not, then I guess the next step is just
>to give up on democracy altogether, which resolves the problem by
>realizing its worst consequence.

Asset Voting accomplishes the creation of an almost perfectly 
proportional assembly, if the assembly is large enough. Asset Voting 
is really a variation on DP, creating a set of proxies who have equal 
voting power.

I don't know what legal system we will end up with. It is enough for 
me that I can see how to create the structures that will allow us to 
manage the system. They do not have to be part of the legal 
structure, it is enough that the electorate be organized *outside* 
the legal structure, voluntarily, by free choice, without coercion of 
any kind, including the common coercion of "our way or the highway," 
that is, "agree with us or form your own damn organization," which is 
so difficult that most people will give up. (This is a form of the 
dictatorship of the majority, but it is not necessarily even the 
majority, merely the first to open shop.) But FA/DP, theoretically, 
should make it extremely easy to "form your own organization," while, 
at the same time, creating the potential of a superstructure that 
could negotiate broad consensus. And implement it.

BeyondPolitics.org, an FA/DP organization dedicated to the 
development of FA/DP concepts and the facilitation of the use of 
these concepts by all kinds of peer associations. Wiki at 
http://beyondpolitics.org/wiki