[EM] Rob Lanphier's hierarchical scheme

Abd ul-Rahman Lomax abd at lomaxdesign.com
Tue Aug 29 09:39:53 PDT 2006


At 05:45 AM 8/29/2006, Rob Lanphier wrote:
>I realize I wasn't being fair to you on the applicability of Delegable
>Proxy (DP).

You are not unusual, except for one thing: you noticed. That's actually rare.

>I'll be referring to my system as "PH", for "proportional hierarchy".

Reasonable enough. The differences between it and DP are that the 
group size is fixed, more or less, group leaders, those who represent 
the group, are elected instead of being chosen. This latter is a huge 
difference, in fact, making PH, as described, into a majoritarian 
system that would suppress minority opinion at the base level, a 
dangerous place to do it, though there might be ways around this 
problem. Perhaps groups are assigned based on some affinity, or 
perhaps groups may send more than one rep to the next level if they 
cannot agree on one. This latter situation would convert PH into a DP system.

>Ok...this is the part that I hadn't previously understood about DP.  I'd
>like to understand the mechanics of this better.  Specifically, how does
>a popular leader subdivide their constituency?

This is great! These are obvious questions, but few ask them. If I 
try to explain it all in one post, or even in a few, I'm too verbose, 
it irritates people. But, in fact, there are hundreds of details to 
explain, and each one is a mine field, ready to demolish the idea in 
people's minds. "It won't work because A," people will think, and 
that interrupts, for many or most people, further consideration, even 
if we have an understanding of A and how and why it is not an 
obstacle. DP is one concept, not terribly difficult to understand, 
though there are many implications that can take months or years for 
people to notice; add to this FA, which is *also* a simple concept, 
particular for those with experience in Free Associations, which *do* 
exist, but which radically alters the implications of DP, it is no 
wonder that I'm seen, quite often, as a crank.

Okay, set up a proxy system, there are going to be some people who 
are the immediate choice of very many people. Some people think that 
DP is "populism," or, in the words of one writer, "lemmingism," 
because of this. Indeed, if this happens, proxy democracy could be 
extremely dangerous. Which is one reason why I want to see DP 
implemented in safe environments before it is tried in government or 
serious control structures. I think that once people experience DP, 
they will never again be content with less. But how to get them to 
that point? Not simple, let me tell you. Or if you think it is 
simple, be our guest. I'd even send money if it were needed and the 
application seemed reasonably likely to succeed.... But it shouldn't 
need money, or at least not much. And I'm not sure we are at the 
point yet where money could be efficiently and effectively spent. We 
might get there, actually, within months, or it might take years... depends.

Alright, already, I'll answer the question! In mature organizations, 
people will expect service from their proxy. They will expect that 
they can send an email to their proxy with questions or suggestions 
and get a cogent answer. They will expect that their proxy will 
understand their interests and will contact them when something comes 
up that they might miss, because they are no longer following the 
traffic. In short, they will expect a personal connection with their proxy.

That is, their *direct* proxy. The proxy will have their phone 
number, often, and vice-versa.

This is not Clint Eastwood, not matter how much I might want to name 
him. (Actually, for me it would be Bob Dylan, perhaps, but never 
mind....) So ten thousand people, perhaps, nominate Clint Eastwood as 
their proxy. Now, Clint could accept, but these people will quickly 
come to realize that their mails don't get a personal answer, and 
Clint hasn't the foggiest idea who they are, they are just names on a 
large list. Clint *might* be able to manage this with staff, but 
that, of course, is expensive. In a Free Association, the money is 
not there for something like this. In an FA, Clint would do something 
else. He would have a circle of friends or acquaintances who he *has* 
agreed to communicate with. They trust him *and* he trusts them. This 
mutual trust is essential to the concept! It is not going to work 
well without it....

So, when a new proxy nomination comes to him, he will pick one or a 
few of his clients and suggest that they respond to the nominator. 
Perhaps this person would be content to be able to communicate with 
someone who is in direct contact with Clint. If the new client has 
something to say to Clint, the new proxy will filter it, and, if the 
new proxy is giving good service, an the proxy is *not* going to pass 
the message along, the proxy will explain why to the client. They 
will be able to discuss it. This process, of course, can be expanded, 
so that even if a billion people name Clint, a proxy assignment can arise.

However, here is my guess. People will realize that they don't 
actually know Clint, and Clint doesn't know them. He might represent 
something to them, to be sure, some ideal that they find congenial, 
but he is not a personal connection for them. Except for very few. So 
they will, instead, name someone close to them. They might be able to 
find a proxy who lives close, they could actually meet face-to-face. 
The development of rapport between proxy and client is, again, very 
important. The proxy should have a good idea, if the client has an 
emergency communication, whether or not this is likely to be a nutty 
or paranoid response, or is likely to be something truly urgent. 
Consider the FBI agent who noticed the flight students not interested 
in landing. From the Middle East.... That agent filed a report 
through the top-down-organized bureaucracy, and was presumably handed 
to a manager with too much on his plate. In the near-term world that 
I imagine, the top-down bureaucracy still exists. It's a control 
structure, FA/DP is a communications structure. So the agent calls 
her proxy in the FBIFA, a Free Association of all persons with 
interest in the FBI, most especially employees. The employee, being 
an employee, has a proxy who has the necessary security clearances, 
I'd expect, but the major point is that the employeee has chosen this 
person, and can make a different choice at any time. The agent can 
talk to her proxy directly, she *will* get a response and a 
sympathetic ear. And then the proxy will decide whether or not to 
pass the information up. What do you think would have happened?

I'd suggest that within a day the information would have reached the 
top levels of the FA, and that people functioning on that level would 
have the ear of the Director of the FBI, if not of others as well. 
They don't have any authority at all over the Director. (But they 
certainly could make life hell for the director if they chose, with 
general agreement among them, to do so!) What do you think would have 
happened? I'd suggest that 9-11 would not have the ominous meaning it 
now has for us, it would simply be a way to make an emergency report 
or request.

So: proxies who are too busy to provide good service will generally 
pass requests on to a suggested person to serve in that capacity, 
which might be someone in their constituency, though it might also be 
someone outside that. It would be someone trusted by the proxy. If 
the new client and new proxy hit it off, fine. Otherwise the client 
is free to chose someone else, just as the proxy is free to say 
"Enough! I don't want to get this mail, I withdraw my consent. You 
can ask So-and-so, if she will accept!"

The DP network is formed by the mutual seeking of proxy and client. 
It is not controlled centrally. It has been suggested that when the 
proxy list shows small loops, leaving people without high-level 
representation, the central organization might notify the members of 
the loop, but it would not control what they do. High-level loops are 
expected and normal (a high-level loop is a proxy loop that has at 
least one member at a high level).

There are other suggestions regarding special-purpose or issue 
proxies, alternate proxies, etc., but I prefer to avoid these for the 
time being. In the FA/DP context, anyone can create lists of such and 
anyone can use them. The *whole process* is decentralized. There may 
be central services, but the organization uses them for convenience, 
it does not depend on them.

> > What I see as a serious shortcoming of Mr. Lanphier's proposal is
> > that far too many members must put in far too much effort. DP
> > distributes the effort. When it is implemented in the Free
> > Association context, we have assumed that all members retain the
> > right of direct vote, when polls are taken, but -- and this is
> > crucial -- not all members necessarily have the right to address a
> > high-level meeting.
>
>PH from doesn't really talk about the right to vote in polls and other
>measures. PH is purely for organizing the discussion itself, making it
>so that everything gets read by someone (or is at least far more likely
>to be read by someone), without everything needing to be read by
>everyone.

An essential part of deliberative process is some means of gauging 
consent or agreement. One of the common frustrations in informal 
associations and groups is the lack of formal process. So someone has 
an idea that has no chance whatever of being accepted, indeed, 
everyone else opposes it. Perhaps strongly. The person makes the 
suggestion, and then an extended argument takes place. Sound 
familiar? Under Robert's Rules, a motion cannot be debated until it 
has been seconded. This is a *huge* time-saver, with little loss in 
openness. If a person can't convince one other member in order to 
gain a second, there is a vanishingly small possibility that 
discussion would, in a reasonable time, overturn this. It is not 
worth the effort.

And if there is a small group, so there *is* a second, but most 
people don't want the bandwidth eaten by the discussion, and they 
have a supermajority of the latter opinion, they can suppress debate. 
Under standard RR, two-thirds vote can do this.

Even a pure communications structure must be able to make decisions 
about its own process, or it is bound by the initial conditions, it 
is not flexible. Meetings can make their own rules, it is crucial 
(and it is standard Robert's Rules, which are only a suggested 
starting place, a default set of rules known to work well in 
democratic process.)

FA/DP is a rigorously, thoroughly democratic/libertarian structure 
and process. Just like Robert's Rules, it has no opinion about 
matters of controversy. That would be content, not process. FAs 
strictly avoid taking any position which could divide the membership 
in a way that makes membership unattractive. This *requires* that 
they avoid property, for example, more than necessary for performing 
the minimal communications function. Which is quite small, so 
negligible that fund-raising is essentially irrelevant for FAs, they 
can pass the hat, literally or figuratively, and easily collect 
whatever is needed. Generally, the amounts spent by FAs are so small 
that many individual members could shoulder them alone, though the 
tradition is also strong that nobody gains control through this, 
except for, of course, what is natural. It is understood that if 
someone pays, personally, the rent for a meeting, they can say who 
can attend. If they want. Mostly they won't want to, because members 
of FAs have plenty of options regarding meetings to attend, in 
general, and where there is no other meeting, they can simply start 
one. Power in an FA is thoroughly distributed among the membership, 
it is not centralized. FAs are quite robust. There may be an FA 
central office, but, as I once wrote when involved with an 
international FA, the central office, which was in Boston, could fall 
into the Atlantic and it would have little effect on the actual 
function of the FA. Local groups would simply take over whatever 
functions had  been taking place in Boston, and they would distribute 
these functions voluntarily. Boston depended entirely for funding on 
contributions from local groups; the FA traditions forbid the 
acceptance of large donations and bequests, precisly to maintain this 
dependence. If Boston started issuing proclamations at odds with a 
consensus among the groups, they would cheerfully ignore it. If 
necessary, they would change their name so that they would not be 
associated with Boston any more. But all these local groups would 
remain in communication with each other. They don't depend on Boston 
for that.....

I must say that my comment was not appreciated by the hired staff in 
Boston.... But, of course, while I was proposing a modest level of 
decentralization of some functions, I was certainly not threatening 
their jobs. What ensued is another whole story, something which 
taught me a great deal about power in organizations. Essentially, 
what was missing from that FA was DP. There was still far too much 
central control; in particular, this FA, like many, held a national 
Conference with delegates sent from local groups. The board organized 
the Conference, and essentially ran it. Oops. The Conference was run 
in a way that kept it from effectively functioning, making the 
organization, in my view, over-conservative. That is, the status quo 
preserved the status quo, the well-known conservation of power 
inequities phenomenon I've written about so much. This happens even 
in thoroughly nonprofit, volunteer organizations. So while there was 
independent communication between the groups, it was not organized in 
a way as to make it easy to develop an organization-wide consensus 
outside of the Conference, and the Conference consisted of delegates 
who mostly served only once, so there was no institutional memory. 
Outside that held by the staff and a few members of the board..... 
Long story, I won't go into more now.

The lesson is that if one wants to rigorously preserve the 
libertarian freedom of an organization, communication must be 
decentralized while still connected overall. That is DP and, in fact, 
is largely why I invented it. Politics was merely a possible application.

>Since I don't understand DP well enough to know how meeting size is
>arrived at, I'm assuming that's entirely up to the leader.

More accurately, it is up to the interaction between the proxy and 
the clients. A proxy has absolute authority to refuse proxies. It is 
up to an analyst whether or not to consider unaccepted proxies in 
tabulating votes. Frankly, I'd be inclined to discount them, because 
they don't represent a real, solid connection.

A meeting, in its own rules governing its own process, would make the 
decision. What would the initial, default, suggested rules be? Well, 
that's what BeyondPolitics is about, to develop some kind of 
consensus on what initial DP rules would be.

There is a distinction between a "meeting" and the constituency of a 
proxy. Meetings, generally, are, I expect, open. Membership is 
self-defined, presumably, on-line, by registration. In FAs, it may 
not be necessary to validate memberships beyond confirming that the 
email address exists, and that the addressee is actually willing to 
receive mail at it, standard wiki/mailing list stuff. There may be a 
meeting, however, that consists of the proxy and his or her 
clientele. Because it has a "membership rule," however, this is a 
caucus of a kind, not an open FA meeting.

>   The problem
>I have with leaving it entirely up to the leader is that, as a
>constituent, I may have a different opinion about what constitutes a
>good group size than the leader does.

Of course. And you are free to form such a group. Maybe you can get 
the "leader" to join it.

The AA tradition: "Our leaders are but trusted servants, they do not 
govern." But they are not slaves, they are free to make decisions as 
they see fit, within the scope of their charge, or otherwise as 
respects their own personal rights. If they don't want more than 20 
subscribers to their personal mailing list, which they are committed 
to read, who is going to force them to do something else?

If the clients don't like it, they are individually and collectively 
free to go somewhere else. Perhaps 40 of them would form their own 
list around one of them who is acceptable to all as a proxy. A list 
moderator who is also the owner of a list is quite like such a proxy. 
The only harm in list moderation traditions is where alternatives are 
not readily available.

General FA meetings would be a little different. They would have 
leaders, i.e., moderators, but they would operate, I'd assume, 
according to deliberative process, with a chair. My own opinion is 
that the majority has the right of decision, but that it will wisely 
understand that broader consensus is worth pursuing where possible. 
But general FA meetings must be able to limit their own traffic as 
they grow in size. They remain democratic if the power and authority 
to do so depends on majority opinion, and if dissenting members have 
other options. DP provides a generic alternative option: where a 
meeting, which is generally equivalent but not limited to mailing 
lists, has closed participation -- i.e., the right to address the 
whole meeting -- individual members still have access through proxies 
whom they choose.

What if a proxy attempts to add clients by promising to forward their 
posts unfiltered? Well, no harm done if no harm done. If this 
irritates enough other members of the meeting, they can revoke, by 
vote, the direct posting privilege of the offending member. Present 
standard mailing list practice is that the moderator does this, and 
all too often without consulting the list. It is not necessary that 
moderator decisions all be debated, but Robert's Rules actually 
covers all this in a way that balances the desirability of open 
access with the need of the meeting to filter out time-wasting noise.

>   The leader may overestimate how
>many people he can deal with at a time, whereas a constituent has a fair
>expectation not to get lost in the crowd.

Certainly. But they will work it out. Clients can regulate the 
process, if they "feel lost in the crowd," by finding a proxy not so 
busy or not so burdened. If a proxy takes on too many, as well, the 
proxy may revoke the acceptances, some of them. To be polite, of 
course, the proxy would suggest to the extra clients a proxy as a 
replacement, and might still be available to some degree. The member 
being dropped might well retain the read-only right to follow the proxy's list.

I could go on and on with stuff that really is obvious fall-out from 
the concept. It is libertarian theory in actual application. I'm not 
a Libertarian, capital-L, because I don't know that libertarian 
concepts and government are ready for each other. But I have no doubt 
about their application in peer organizations. Which may, by the way, 
be power and control structures if the members consent.... I'm just 
starting with FAs because it is safe. A political DP organization 
missing the crucial rapport between proxies and clients could be 
*very* dangerous. I'd rather see existing structures used, initially, 
for control structures like PACs, with FAs advising them. Which is, 
by the way, how AA handled the problem. In the Traditions:

"AA as such ought never be organized, but we may create service 
committees and boards directly responsible to those they serve." So 
AA's central operation is in the hands of AA World Services, Inc. The 
board is, I think, elected at the annual AA Conference, but it has 
full authority over the assets and, indeed, the legal responsibility. 
They can ignore even a full consensus of the Conference, though they 
would only do so, I think, if the Conference asked them to do 
something illegal. If needed, AA as a whole could simply drop AAWS, 
stop supporting them, and replace it with some new organization. The 
money, given the size of AA as a whole, is trivial, pocket change. 
AAWS, completely dependent upon continued support from the 
membership, usually through intergroups, would close down. But no 
such conflict is likely to appear. The limitation in the Tradition on 
collecting funds beyond immediate needs plus a "prudent reserve" 
keeps AAWS in the dependent position; conversely, the members provide 
AAWS with whatever it needs. Just not individually, no individual, I 
think, may contribute more than $1000; similar restrictions exist on bequests.

>This is the problem I'm trying to solve.  I'm trying to create a system
>where people get the most merit based way of moving from "unknown" to
>"qualified".  What I find with Daily Kos is that you can spend a lot of
>time writing, and not rise above the noise, without being sure if its a
>quality thing, or just luck, or headline writing skill, or what.

In a DP system, people do this by being massively trusted.

Yes, you have identified the noise problem. In particular, you don't 
know why you do not seem to be heard. Indeed, you don't know if you 
are heard at all. Many readers may not rate what they read, even 
where they agree with it. In a DP system, you have the kinds of group 
participation systems that currently exist, *but* you *also* have 
your proxy. Your proxy, generally, has agreed to communicate with 
you. If you post to your proxy's list, and your post falls like a 
lead balloon, and your proxy does not spontaneously reply, you can 
write to the proxy personally and ask. You can expect an answer; if 
you don't get an answer, your proxy is not giving you good service, 
and you advisedly would find another proxy. Whether or not you ask 
your proxy for a recommendation, I presume, would depend on your 
judgement as to the reason for the lack of response.

Suppose there are people who are cantankerous, they offend their 
proxies, one after another, they have trouble finding someone to 
agree. A list would exist for such people. They could inflict their 
excessive communication or abuse on each other, and perhaps out of 
this leaders would emerge who could represent these people. If not:

If you can't participate yourself, and you can't find someone to 
represent you, .... you might ask yourself why  and try to understand 
how to change your own behavior in order to be able to connect. I 
don't see other options, beyond that cantankerous list, kind of a 
self-help group for Cantanks, or whatever they are called. :-)


>I also want messages to go to more than just a single "qualified
>member", because this isn't about setting up absolute gatekeepers.
>There needs to be oversight on the gatekeeper operation...so, in
>addition to the leader seeing and approving, the group would get a shot
>at it.

The client oversees the proxy's gatekeeper function. Generally, in 
the organizations I'm promoting, FA lists are public, so a client can 
see the views and positions of those who *can* post, as well as those 
whose posts have been forwarded or otherwise accepted, so the client 
has means of finding potentially congenial proxies.

The alternative "oversight?" Central control, or what amounts to 
something dangerously similar, control by fixed rules, perhaps 
implemented in software.

I presume that general lists would, by default, have a moderation 
queue, and there would be members of the list generally trusted to 
have the right to approve posts in that queue. Some lists would not 
have that queue, probably, so all non-qualified members would be 
"banned," which simply means that any post from them would be 
automatically rejected *without* going into a queue. Instead, the 
rejection message would suggest to the rejected writer how to get a 
post on the list: look at the open proxy list, pick a proxy, etc.

>I believe my system also respects minority opinion.  By being a
>proportional system at all levels, one would be able to vote oneself
>into a small group of likeminded people.  That group would filter into a
>larger group of somewhat likeminded people, and that's where the
>deliberative synthesis would occur.

Sure, you could do it that way. Or you could simply allow people to 
directly organize themselves without the assignments. Indeed, any FA 
could have in place a mechanism for assigning people to a default 
group. Simple and trivial. Just not an essential part of the system.

Could be quite useful, a way for new members to break the ice. You 
ask for it, you get assigned to a new list with no more than N 
members. Mailing lists can get pretty large, compared to face-to-face 
meetings, before noise becomes a problem.

In other words, take the FA/DP, distributed structure we have been 
working on, which is *extremely* simple, and add to it a mechanism 
for automatically assigning members to a Newbie list. It could be as 
simple as a membership message that notifies new members that they do 
not have direct posting privileges to the list they just subscribed 
to, but that they may join the FAlistNewMember27 list, the current 
new member list. Periodically, admin would create a new list and 
update that list name in the message. Easy-peasy.

I do see some problems with this, but I'm confident that the people 
involved at the time can work them out.

I do *not* think it necessary that we anticipate and solve all the 
problems in DP democracy before it is implemented. We are working for 
the implementation of basic DP in FAs because it's a safe 
environment, with the flexibility to change as needed, quite quickly. 
Later, it is possible that "heavier" DP implementations will come, 
but by then we would have substantial experience with it, most of the 
problems would be known and solutions developed.

This, of course, is entirely different from what we normally see on 
the EM list, which is an attempt to anticipate and deal with all the 
problems in advance, often with no practical experience behind a 
proposed method. I'm only focusing on the obvious problems of DP, 
which is, of course, far more than an election method, though it can 
be used as one.




More information about the Election-Methods mailing list