Free association
From BeyondPolitics
Contents |
Free Assocations with Delegable Proxy
>[Why private communication >is necessary for reaching consensus]
There are lots of reasons. This is FA/DP theory. (Free Associations with Delegable Proxy). FA/DP was my main focus before I started working with Wikipedia, I developed it over the last thirty years or so.
Other people have done similar work, but not with such a coherent vision of what's possible. FAs work, for certain applications where most of the communication takes place on a small scale, locally, where the power remains local, and a large organization only handles overall coordination and certain shared functions without the power being centralized.
Alcoholics Anonymous
Classic FA is Alcoholics Anonymous, most of the FA characteristics are drawn straight from the writing of Bill Wilson, the main architect of AA. AA, without centralized power but a high level of overall coherency due to serious consensus process, rapidly expanded until it almost totally dominates its field, there isn't anything equivalent competing with it on anything like the scale of AA. Without money, without centralized power, without centralized control, someone seeking support in dealing with alcoholism can find a meeting in every small town in North America, and in larger towns, often a meeting every night, in big cities, many meetings every night. All free, self-supporting, and probably the most effective "treatment" for alcoholism. One of the big things any treatment program will try to do with an alcoholic is to start the AA habit. It's not religious; early AA was an offshoot of the Oxford Group, later known as Moral Re-Armament; certain concepts were retained, but Bill Wilson generalized and de-sectarianized the Oxford approach, and hammered it out with real situations and real groups.
The Twelve Steps still mention God, though lots of members translate or say it as Higher Power, and "higher power" means, essentially, that it's realized that to "return to sanity" we need something more than ourselves. The group serves as the higher power, in practice, especially in the beginning, but most long--term members find deeper resources, i.e., what is called "conscious contact."
None of it is obligatory, there is no dogma, only a general consensus, which members can disagree with. Classic case: new member declares that he can take a drink any time he likes, as long as he doesn't drink too much, i.e, he doesn't believe this crap about alcoholics must avoid all alcohol. He'll be told, "Let us know if it works, and how you did it, most of us would like to do that as well, and we tried, and some of us made it back, others didn't." And the new member, if he listens, will hear those stories, he doesn't necessarily have to go through the pain himself.
I didn't do 12-step work in prison, when I was a Muslim chaplain, but I talked with other chaplains who did, who were familiar with the program (often for themselves, and that may have been how they went into this volunteer work). It was common among people familiar with the program, who understood the structure, to think that it would be great if the world could run this way, but ... how?
The AA traditions work due to special conditions that are often absent in real-world organizations, but I noticed something. One of the traditions written by Bill W. is "AA as such ought never be organized, but we may create special boards or committees directly responsible to those they serve." "Special boards" refers to, I believe, classic corporations, typically non-profit. AA does not endorse any "outside organization," but it has, through its members, special relationships with various organizations, especially treatment centers. AA doesn't accumulate property, beyond what's called a "prudent reserve," typically a few month's expenses at any level. But treatment centers can become multimillion-dollar operations. AA doesn't endorse them, but feeds patients to them, not by an official endorsement, but through personal contact with AA members who've been in a center, or through running AA meetings in space provided by centers (theoretically, the meeting pays rent, it's considered essential to maintain autonomy). When I say that "AA" is running a meeting, actually, there isn't any oversight or control. Any two alcoholics, meeting for the purpose of mutual support "may call themselves an AA meeting, provided that, as a group, they have no other affiliation." Theoretically, you could have a Baptist AA meeting, but only by "accident," i.e., without any requirement that members be Baptists, and it would never advertise itself as "Baptist."
FA as deliberative structure
So I began to think of FA structure as a kind of independent deliberative structure, for communication and voluntary coordination, and power would remain in traditional structures; what the FA can do is to facilitate consensus among its members, and those members can exert, individually or on concert, whatever power they have in the "outside organization."
If you look at it from a political perspective, this is the separation of the judiciary from the legislative and executive branches. The faculty of judgment, of the formation of cogent and intelligent opinion, is kept separate from the levers of power, because power corrupts it. FAs do nothing but communicate and advise, they do not control or coerce. Bill Wilson: "Our leaders are but trusted servants, they do not govern." And it really works that way. Sure, AA members are human, and sometimes an individual comes to dominate an AA meeting, demanding that it be run according to his or her personal belief. Wilson knew this would happen, and what he set up was a structure where meetings fission easily. There is no property to fight over. You don't like how a meeting is run, you start another one. The saying in AA is "All you need to start a meeting is a resentment and a coffee pot." So resentments are turned into proliferation of meetings and the only uniformity is that which forms by real consensus.
FA/DP theory was developed to solve the problem of democracy in general. It requires no change to existing laws or structures, but, if it works, it could change those laws or structures with ease, when the time is right. Essentially, if you can find consensus, you can do just about anything that large numbers of people cooperating can accomplish. Consensus is powerful. Majority rule is a quite good principle, but it's only a bottom line, a dangerous edge of a cliff, on the other side of which is minority rule, oligarchy, or dictatorship. When we are content with a bare majority, we are living on the edge of that cliff, and we will often, because of process variations actually be over the edge.
Wikipedia
We see this on Wikipedia, which is highly vulnerable to local participation bias. DR theory is excellent, but most Wikipedians actually don't understand it, and there are parallel processes which can torpedo it, such as the noticeboards. Noticeboards work where the issues are clear and there is an established community consensus, but they are easily manipulated.
Why private communication is needed
Okay, the original point: private communication. FA/DP theory would suggest that Wikipedia editors and other interested persons (who represents the readers?) be connected with a communications hierarchy. This hierarchy is formed spontaneously from the bottom, though some growth can also take place top-down. It's formed by every interested person choosing another FA member whom the person decides to, at least provisionally, trust, and that linkage is complete and operational when accepted. This is a personal relationship, nobody else can interfere with it, but, to become maximally effective, it's documented, i.e., the "client" names the "proxy" and the "proxy" accepts. What this means, at foundation, is that they agree and consent to mutual communication. The proxy is revocable at any time. The term "proxy" can be a little misleading, there is not necessarily any power transfer, the essential element is a representation of trust and, hopefully, the development of rapport between the proxy and client. In many ways the relationship is symmetrical, but it's easier to conceive of the effects if we, at first, think of it as unidirectional. The traffic between the client and proxy is self-limited, there is no outside control or monitoring.
On Wikipedia, editors can be sanctioned because of how a communication between two editors is viewed by the community, and it's often a hostile view that has no sympathy, and it may depend on the accidental meanings of words. Editors should be free to communicate with each other according to their mutual needs. We are so concerned about canvassing that we forget that canvassing is simply communication, and we have confused unwanted communication with voluntary communication. The canvassing rules are mixed up with spamming rules, with no clear boundary between them. If people are going to coordinate their efforts, they will, they will simply do the communication off-wiki.
Canvassing rules, votes and decision-making
We need canvassing rules because we routinely violate our own guidelines and sound theory. Decisions, supposedly, aren't made by votes, but by arguments. Okay, if that is true, what's the problem with canvassing. There are two possible effects of canvassing, which would be the multiplication of votes without an increase in cogency of arguments, or an increase in cogency because of participation by more editors who have more ways of looking at a situation. The latter effect is salutary, and, if we discount votes, the former harmless.
But, in fact, we are, under present conditions, swayed by votes. So, because we have set up processes where voting is considered important, we then have to restrict canvassing, which then acts to restrict the intelligence applied to a situation. Bad idea.
Now, toss proxies into this mix. If a significant number of proxies have been created, it's possible to expand an estimate of consensus by looking at votes -- real votes, we should be honest about what we do -- and adding to expressed votes the number of clients represented by a voter, who didn't vote themselves. This would not be to make a decision on the basis of this estimate, per se. Rather, what it could do is to actually counter the effects of canvassing, because it would be seen that, sure, there were many individual editors who piled in to, say, delete this article, but the few who voted to keep can be seen to represent many editors. It's also possible to weight proxies in the analysis by any factor the analyst wants to use, such administrative status (of the voter and the clients of the voter), or edit count, or any other factor the analyst chooses to use. It's more information, and it could be automated; my recommendation is that it not be centralized, though. Rather, the data would be redundant and subject to independent analysis at any time.
Sockpuppetry
That's an expressive function, and it would be used to advise, at the discretion of a closer, as to community consensus. The objection was commonly made about sock puppetry. Absidy (Sarsaparilla et al) named me as his proxy. We were both immediately hauled before SSP. It was pretty stupid. The last thing a puppet master wants to do is to make the relationship explicit! And proxy information is purely advisory. Remember, cogency of argument. Two real users cooperating have twice as much sensory and analytical and expressive power. One real user pretending to be two is still one person, all this person can do is multiply votes, not actual participation.
We've gone over the theory fairly extensively elsewhere, sock puppetry is a rather stupid exercise that, in a Free Association, gives the puppet master a mouthful of hair, because, remember, no coercive power. And I can look at data like registration date, edit counts, articles edited, etc, and personally discriminate between what looks like real editors and what looks like fake. And I don't really even care; the purpose of estimating consensus is to try to anticipate what will happen if push comes to shove, so that we can avoid coming to pushing and shoving. If it looks like there is this really stubborn and large faction that just doesn't seem to respond to reasonable negotiation and compromise, and if I know I have serious support (which I certainly won't know if I've created my own sock puppets), I'll simply say No.
In political organizations, real cooperating groups of members can raise large sums of money, rapidly. Creating socks doesn't amplify your fund-raising ability at all, it really just wastes time for a very small gain, a possible appearance of more support than you actually have, for what? The only worthwhile purpose in an FA is to gain an audience for ideas, and FAs are structured already to provide maximized freedom of information flow without the nodes (the individual members) becoming overwhelmed with noise or excessive traffic.
Filtering of information
To get an idea considered at a higher level in an FA/DP organization, you only need convince your proxy it's worthwhile, and you chose this person for trustworthiness and rapport, ideally. If you can't convince your proxy, maybe it's a bad idea! And your proxy will explain why; with traditional structures, we often have ideas that we'd like considered, they gain no traction, and we can't figure out why, except to blame inertia, stupidity, bias, etc. Our proxy will explain why the idea isn't being passed on: it's because one of the natural duties of a client is to not bug the proxy with useless garbage being sent from others. If I had a client who was sending me emails on the nature of "I don't agree with this idea, but I have a client who thinks it's hot stuff, and I want to make him happy, so I'm sending it along to you," I'd probably fire the client! Please! I agreed to open communication with you and not to everyone you might want to please. If you don't think it worthy of my time, don't pass it along to me, instead explain to your client why you don't accept it.
Filtering of information is essential to intelligence. Filtering must be intelligent. Proxy networks set up filters for information flow, and the filters are chosen by those who need both to pass on information and to be protected from too much information.
These networks would have many functions, but the bottom line is that with this, existing Wikipedia structures would work far more efficiently.
Decision-making
There are also possible structural changes that proxies could facilitate. Suppose that we have a proxy-naming facility, such as the one that was set up with WP:PRX (all that this proposal did was to set up a simple file structure, for voluntary use). Okay, first select an editor base according to some standard: autoconfirmed, perhaps, more than X edits, registered at least M months. Then do a delegable proxy analysis of the editors and proxy assignments. It is possible from this to determine a proxy status for an editor, i.e., if nobody else voted, how many clients would this proxy represent directly and indirectly? From this, it's possible to determine the top N editors as to proxy rank. Set an optimal assembly size, and allow the top N editors by proxy rank deliberative rights there. With the existing software, which doesn't allow page rights, per se, with sufficient flexibility, you'd set up the assembly deliberative structure off-wiki. Only those editors can edit the deliberation pages. However, when there is a vote, any editor can vote. The votes of the members with "seats" in the Assembly are expanded by the votes of those they represent, and who did not vote directly. (If a proxy who doesn't have a seat votes, those votes are likewise expanded).
It's hybrid direct-representative democracy. A Wikipedia Assembly should seek consensus, because it can't actually control the editors, and a decision there by a mere majority could be disruptive, but, on the other hand, requiring;; more than majority vote, in fact, is known to become minority rule wherever the status quo favors a minority. Wikipedia may be rather stuck on that at this point.
No elections, as such. Composition can shift as proxy assignments shift; an Assembly can also make certain seats permanent. That confers no actual power beyond a recognition that the advice of a member is valued, for voting power isn't dependent on holding a seat. It simply gives that member a right to "address the assembly." It would be revocable or subject to restrictions, if needed.
(My guess is that the vast majority of votes cast in an assembly would be cast by those with seats; but allowing direct voting would have a major salutary effect, it makes it plain that editors are equal and that the restriction of access to the Assembly pages is not a confinement of power to the Assembly. If any editor wants to put a proposition before the assembly, all they have to do is convince one editor with a seat that it's worthwhile, and for the one with the seat to find a second.)
(Yeah, standard deliberative process: a motion isn't debated unless seconded. Raw discussion, as dominates most Wikipedia process, would only take place in subcommittees or pages that are "committees of the whole", and these never make binding decisions, they are only used to prepare recommendations.)
Ahem. That's what could be. Until then, what's important, as I see it, is that editors start to organize for direct communication. There is now an Association of Established Editors being attempted, and one of the purposes is mutual defense. Good idea, wrong procedure. They are trying to do it on-wiki, and will probably get crushed or neutralized.
Much more important, however, than coordination for defense (or, I suppose, for offense) would be division of labor. Wikipedia must become more efficient, it's burning people out. It's like a big Ponzi scheme, you can invest many hours of work and, poof! it's gone and you have to invest more time to maintain it, if it's even possible. There was a naive assumption that articles will only improve, but I've seen much article devolution. The people who created the article, and who often knew the subject, aren't maintaining it any more, and articles can get really mangled by people who don't understand what they are writing about, it's worse than newspaper reporters.
[specific example snipped]
Canvassing
>[re WP:CANVASS]
Ultimately, it's a restriction and inhibition on social intelligence, which limits us to the views and opinions of a small subset of editors. Paradoxically, having a proxy system in place would probably reduce the number of actual votes at AfD, etc., with an increase in variety and cogency of arguments.
>[more re WP:CANVASS]
Spamming is a problem.
Network of proxies
Proxy choices won't be random. It's a directed graph is how delegable proxy, I think, has been described. What you'd be looking for, with two editors, would be communication between circles of editors. If everyone names a proxy, you should realize, there will be created "proxy loops." These are another kind of "natural caucus." A single "superproxy" is possible, but unlikely. However, when loops are small, there is an incentive for them to connect themselves with larger loops, because members of the loop will then gain access to a larger circle. All it takes is one member of the loop (a loop member, one where, if the loop is broken, that member represents all members of the loop. There are loop members who don't represent anyone, these are "inbound connections" which may be individual editors or which may represent many members, but, in a single-proxy system, they aren't themselves loops but rather chains and branches that connect to a loop)
The important thing is that it's easy to document who is represented in a discussion and who is not. With some processes, automatic notification at some point of unrepresented loops or members might be generated. "We are holding a vote on whether or not to distribute as a lump payment to editors the 3.5 billion dollars that was recently donated to the Foundation, in gratitude for service. The present vote analysis shows that you have not been represented in this vote; you may wish to vote directly, or to suggest to your proxy that your proxy vote or otherwise take steps to allow your caucus to be represented. The discussion is at X; analysis of the discussion has been collected at Y, and you may wish to ask your proxy about this, or review it yourself and come to your own conclusions." Etc.
>They don't have to be called "proxies". They could >be called "friends", "associates" or "intermediaries".
Michael Nordfors called them Advisors. We both thought they could be called Advisor/Proxies. Advice could be taken to represent outbound information and Proxy to represent inbound information, toward the "center." I.e., the virtual council that consists of all high-level proxies.
>Then discussion on the issue could occur between >each pair along the path.
Only as needed.
One proxy per person
There are ways and situations where there would be more than one. Special Proxy lists, which override a general proxy for a specific situation, can be set up by anyone. So if you are interested in an article, the article may have a special proxy list, where you name which editor you most trust, or have the best communication with. Proxy lists are not assignments of power, they merely document relationships, and if there is an editor who is trusted by, say, ten editors, and an edit war starts with an editor who only represents himself, well, who do you think will win? A priori, who should win? Edit warring is how Wikipedia handles majority rule, in a very rough way. But an editor who is trusted by ten editors who are themselves well-connected is probably going to understand the importance of consensus and will only wield that editing power stick (i.e., requesting clients to assist) if it's needed to bring the lone editor to the table to negotiate.
Now, consider the lone editor. Seeing the proxy assignments, this editor knows not to get into a pissing contest with the editor. But, hey, one of those clients, hes know that editor! Good wikifriend. For some reason this lone editor has been unable to get through to BigShot. So the editor contacts his friend. If he can convince his friend, he's got a shot at BigShot, someone whom he can assume will be able to at least receive a fair consideration. And if not, he'll hear why, from his friend. Much better than being told by someone else.
But visualizing and understanding DP networks is hard enough with single proxies. The point of single proxies is to amplify trust. If I choose someone whom I trust the most of all those available to me, and that person chooses someone similarly, there is what I anticipate is an amplification of trustworthiness. Highly trusted people are more likely to make sound choices of whom to trust. It's not infallible, and this only needs some general average increase in trustworthiness as one goes up the hierarchy to create circles of highly trustworthy people at the top. My sense is that when people directly know and interact with other people, they usually make decent choices of whom to trust, it is assignment of trust to people whom they don't actually know that is where most abuse of trust and power take place. I.e., have I ever voted for a politician whom I actually knew? (Once or twice, in the small town of Cummington, Massachusetts, the finest small town in the universe.)
That's why acceptance of proxies is so important, and why it's important that acceptance reflect permission to communicate, and why clients should be educated to expect response from proxies; and this expectation then naturally limits the number of direct clients that a proxy will take on. There have been proposals to limit the number of direct proxies that anyone can collect, but, in fact, I consider it better for it to be totally self-regulating, at least in a Free Association.
