[EM] Fwd: proxies Was Mae West

[Abd ul-Rahman Lomax]

At 11:47 AM 4/9/2007, raphfrk at netscape.net wrote:

> > (Perhaps newcomers should be
> > encouraged to choose a proxy about a week after they join, so they can
> > continue to be represented even if they go inactive. It's typical that
> > only about 10% of newsgroup members are "active" in any sense.) And
> > there should be some mechanism for conducting polls.
>
>I think it would be hard for newcommers to select a proxy.  Also, they
>likely would consider it providing them no benefit.  After 1 week, would
>they really know enough about the other members to pick a proxy?  On the
>other hand, if the forum was specifically setup with that in mind so
>people knew it would be required, then there would likely be less
>resistance.

It shouldn't be required. It should be encouraged. Proxy assignments 
can be changed at any time, so it isn't necessary in a Free 
Association to be certain that you've made the right choice. Just 
someone who seems to be trustworthy and, very important, who is 
willing to take on the responsibility. That responsibility is not 
much, we would expect, but there should be a voluntary relationship formed.

>I wonder if a possible solution would be to automatically assign proxy
>and then allow the user to opt out.

No. Not. Bad idea. There are proposals to do that in some of the 
systems being developed by the top-politics people.

If automatic assignment is done, we have no measure of trust. We 
cannot assume that the proxy will in some way represent the client.

Quite simply, it is not necessary that all members be represented by 
a proxy. It's desirable. But until members are comfortable appointing 
someone, it isn't a good idea to make a proxy assignment by default.

Now, a "greeter" or "buddy" assigned by default, that's fine. It's 
possible to do that. This would be someone who is assigned to a new 
member, who greets them and helps them get going in the group. I'd 
actually discourage that this person be named as a proxy. Maybe. (The 
introductory material would explain the difference between a greeter 
or buddy and a proxy, that a proxy is chosen for general 
compatibility, rapport, or trust; a greeter could be anybody who can 
be polite and helpful. Different jobs.)

>   For example, there could be a system
>where users can rate posts in the forum.  For example, "strongly agree,
>agree, neutral, disagree, strongly disagree".  If the new user marks
>say 5 of a single poster's posts as agree or strongly agree, then the
>system would recommend that the user selects that poster as his proxy.

I suppose. I don't like centralized systems like that because they 
can be corrupted. That particular idea might be reasonable.

However, this point is very important: any Free Association sets its 
own rules. There can be defaults, but those defaults would, I assume, 
be developed as the actual practice of a real Free Association. Quite 
simply, I'm willing to consider possibilities, particularly when some 
alleged problem is asserted, but merely to suggest that the problem 
could have a solution, not to fix the solution as even being desirable or not.

I'm trying to keep the idea very simple. Complexity will be added by 
people as they need it. If it ain't broke, don't fix it. If simple 
works, don't make it complicated.

>The user would obviously have the ability to completely override the
>system if that is what he wished.  However, the idea would be to make
>it so that the user wouldn't want to as they system tends to give
>effective proxy links.

Rating systems have been proposed (and are actually being used in 
some large discussion systems) as a method of vetting participation 
at a high level. Again, I'm generally opposed. It gives to much power 
to a central entity. I see FAs as associations of people, and as soon 
as you make this dependent upon software, you've got a whole new 
world of problems. Demoex in Sweden abandoned Delegable Proxy, not 
because it didn't work, but because they thought of it as a software 
thing and the software was buggy.

>Also, the system could give an estimate on what the user would think
>of a specific post.  The user could then make adjustments as needed.
>As time passes the system should be trained better.

It's interesting. But this is like imagining automated repair systems 
for automobiles before we have any automobiles.... Maybe it would be 
useful, maybe not.

>Actually, this problem is more generally that each username does not
>represent the same level of commitment.  If the point of the FA is to
>coordinate real life actions, then a person's vote should be related to
>how much real life effect they will have.

Actually, the point of the FA is communication, and especially 
communication across factional lines. Where people want to take 
action, they will use the connections made in the FA to organize 
independent organizations which would then coordinate action. The FA 
creates the networks, it doesn't use them for anything except 
communication. But the networks are networks of people, and the 
people are, unlike the FA, free to advocate, to take controversial 
positions, to collect assets, etc.

A proxy is, effectively, the leader of a "natural caucus," which 
consists of the proxy plus all clients direct and indirect. 
Functioning caucuses can and will be broader than that, of course, 
caucuses will form as needed; they are effectively independent 
organizations united by some position or special activity, but they 
only actually form independent organizational structures as they need 
them for their purposes.

>If the point of the system is to negotiate compromises, then the default
>situation needs to be known.  If you have 1000 users but only 10 of them
>are likely to take part in any protests/counter protests/actions, then
>only those 10 really matter.

It would seem. What I've done, in my vision, is to leave this 
analysis to the caucuses. There can be means of estimating true 
support. First of all, there can be means of estimating the presence 
of sock puppets. Proxy lists are public, and so are participation 
records. Sock puppets would show different patterns than real people. 
It could get quite sophisticated, looking at IP addresses, etc. 
Membership validation schemes are possible; an FA might do this 
officially, but definitely individuals could organize it independently.

When you are seeking consensus, it is not essential to know the full 
strength of the various parties. One simply assumes that a party 
exists if someone is representing themselves as part of it. The unit 
of organization in the FA is the "meeting," which could be 
face-to-face at some location, or it could be a mailing list. When 
meetings get large, then rules start to form which control access to 
the "floor." Mailing lists can get quite large before this is 
necessary, much larger than f to f meetings, but, ultimately, they 
face the same problems. Anyway, we want, ideally, every view to be 
represented at an FA meeting. It is not critical to know if a faction 
is large or small.

So how do we know when there is sufficient consensus to go ahead and 
act? The FA does not need to answer this question. A caucus can act 
whenever it decides to do so. If it needs to estimate the size of 
opposition, it will use whatever tools it can find or create. 
However, if the factions have thoroughly discussed an issue, if 
consensus has appeared to the point where dissent seems to be little 
more than the ego of isolated individuals, caucuses will, I assume, 
go ahead safely. Where it is less clear than that, they might be 
advised to continue deliberation and negotiation.

It is *entirely* up to the caucuses when they are ready to act. The 
FA provides a structure that can be used to make the development of 
broad consensus possible, but it does not guarantee that such 
consensus will be found, nor does it insist upon it.

By not needing to make fixed decisions, the FA avoids a whole world 
of security problems....

>Ofc, if the point of the organisation is to coordinate voting in
>the actual election, then everyone matters equally.  You get just
>as much effect on the election by convincing 100 people that your
>candidate is a good one as you do by convincing 100 people that
>your candidate is worth dying for.

In the election, that's true, and this would be significant if 
everyone belongs to the FA. But if the FA is an early stage one, most 
people *won't* belong, so the caucuses will be involved in political 
discourse, political advocacy in the traditional sense. Such as 
buying advertising, knocking on doors, etc.

The FA isn't going to coordinate voting. A *proxy* will quite likely 
suggest to his or her clients how to vote.

> > If the forum attracts many members and is representative in political
> > makeup of the province, _and_ if the polls are honest, then the forum
> > could become quite influential, because it would be a strong statement
> > of the will of the people if a poll resulted in 90% of the members
> > being for some particular position or solution.

>There are alot of ifs there.

Sure. But they aren't implausible.

>   It is a basic bootstrap problem and if
>it actually does start to become influential (so it is worth corrupting),
>then sockpuppet prevention becomes more important and difficult.

The vision here is more of a traditional organization with 
corruptible structure. I'd say that if anyone takes the poll results 
of a non-verified FA membership as being authoritative, all by 
themselves, they'd be foolish. But, on the other hand, if a proxy 
walks into a politician's office and claims to represent 10,000 
people, the politician will either accept it or question it. So if 
the politician questions it, the proxy simply has a few thousand 
people show up the next day. From then on, that proxy has 
credibility. And clout. That can't be faked.

(And if the proxy asks clients to show up, and they don't show, well, 
the proxy has learned something about his or her clients....)

>One option would be to link every user to a RL registered voter.  Users
>in the forum could be general users and confirmed voters.  All polls
>could report both results.

Yes, that's certainly possible.

> > If the forum does become influential, then there is some incentive for
> > people to corrupt the polls to their advantage. How to avoid that?
> > - Make it part of the culture of the forum that proxies are
> > responsible for verifying that any clients they accept are who they
> > say they are. In order to avoid one person masquerading as several, it
> > would probably be necessary for the proxy list to use people's real
> > names.

The way I see it is that people can join and participate anonymously. 
But such memberships are known as such. When polls are analyzed, this 
can be taken into account. Polls are *not* "anonymous" in the FA, 
they can't be if they are going to be expanded.

Systems are being developed which are designed to allow secure proxy 
assignments, but, once again, I consider all that premature and 
hazardous. Centralized mechanisms are far more easily corruptible.

>I know I suggest it too, but this could be a biggie.  One of the
>'benefits' of the internet is that people can be anonymous.
> > For a bit of additional security, add the person's street
> > number or the last three digits of their phone number, etc. If a proxy
> > is found to have accepted a fake or duplicate person, just post a note
> > to the forum reporting the evidence.
>What evidence ?  Also, posting user's addresses isn't going to go down
>very well.

I don't think it's really a problem in an FA. It certainly would be 
in a traditional organization that controls assets. However, if I'm 
wrong, my assumption is that the FA can develop solutions as it needs 
them. It's enough to know that solutions are possible.

>What is needed is some way to prove that a user is a RL person without
>giving away to much identifyable infomation.

All this is trying to solve the security problem (this is an aspect 
of security) far in advance of any necessity.

> > The credibility of that proxy
> > would go down if it happened more than once, and people would start to
> > discount votes associated with the proxy's supposed constituency.

However, yes, this is the general FA mechanism for dealing with 
dishonesty. Caucuses will organize validation and analysis, on their 
own. The FA doesn't have to do it. However, it can establish certain 
basic tools. The membership list, which would show when a member 
joined and possibly IP address or other related information. There is 
a question as to whether or not email addresses should by default be 
public in the system. Probably not. Definitely, when you appoint a 
proxy, you will be providing your contact information to that proxy, 
and when the proxy accepts, you will be provided with contact 
information for the proxy. This can be and should be direct contact 
information, not dependent upon the FA mechanisms.

Then a proxy list or lists would be public record, and the history of 
these lists would also be public, so one could analyze how proxies 
moved, when they were assigned, etc. A sudden blossoming of proxy 
assignments to one person would arouse interest and possibly 
suspicion. It's possible it's legitimate....

> > - Make the polls open (no secret ballots). Then anyone can interpret
> > the ballots however they like.

That's essential, actually. The primary poll results are just a list 
of how identified members voted. It's possible that the FA will have 
an automated expansion facility, but it isn't actually necessary. 
Anyone can do the expansion with public data.

>This is a good thing.  However, I am not so sure how practical.  There
>would need to be a way to automate it.  Few people will read the entire
>vote and re-weight all the votes.

Anyone can set up a system to do it and report the results. And 
anyone else can verify it. It isn't necessary for everyone to do it! 
Indeed, as a member of an FA where I'm not extensively active, I 
leave all this to my proxy, who tells me when and if my attention is 
needed.... the proxy serves me and, potentially, I serve the proxy. I 
don't control the proxy and the proxy does not control me. But the 
proxy listens to me (or I drop the assignment) and I'm likely to 
trust the proxy's recommendations; at least the proxy has the best 
shot at convincing me to do something.