From Metagovernment - Government of, by, and for all the people
Jump to: navigation, search

Collaborative Consensus

What if a Group were able to define its own conditions for Consensus? Such as 'Bob' creates a group within the community with a 75% threshold for declaring consensus. Members joining that group would accept that definition holds for the group as a whole, but would also be able to change that definition if they managed to meet the threshold.

Similar groups within the community could have their own different thresholds, but the community consensus threshold could be determined automatically from a weighted calculation of the group thresholds and the respective membership numbers. Thus if a Quaker group (!) specified a 100% threshold, they could preserve their identity (this is what we, as a group, think) while contributing their fair share of input to the community, and not preventing community consensus from being reached.

A way to implement this, might be to allow a user to simply specify their own option/setting percentage value for consensus, from which the group could derive its threshold.

Interestingly then, you'd be able to compare what the group-collective consensus is on an issue, and what the global consensus is (weighting each individuals vote equally irrespective of their group memberships). I would expect there to be some divergences, reminiscent of the old republican fear of mob-rule. Richard Franks 18:43, 31 August 2008 (UTC)

I think that is pretty much the way we have been leaning. At least, it is the way I have been thinking, but I have not gone into depth with the programmers on this topic recently.
The way I see it, each community would probably run a separate instance of Metascore. Each community could then set variables such as what defines a consensus, whether or not to use user scoring, what influence to give to people "outside" of the community, and other things we want to test and/or give flexibility on.
One corollary of this which I have meant to bring up with the developers is that it would benefit greatly from a user-portability system. I am not sure how we would implement it exactly, but it would be very helpful if users could register only once, then apply that registration to other instances of Metascore. A user's tags would travel with them, then, perhaps with a community-prefix to each tag. — Ed Pastore 03:23, 1 September 2008 (UTC)
User-portability is easier to implement than it is to retro-fit. Especially with regards credentials - if a "good" user was originally from a community which subsequently goes "bad" (e.g. hacking a modified version of Metascore) - should that user be disadvantaged (unable to login/interact as normal) if other Metascore servers stop communicating with the rogue server? — Richard Franks 04:42, 1 September 2008 (UTC)
If I understand my own scenario correctly (which I may not, as I am not really a programmer), then I think we would not have to drop everything which is in any way associated with a rogue server... instead we could just drop all tags which are prefixed with that server's community prefix. In fact, theoretically, any community could weight the prefixes from any other community, giving more nuance than simple accept/drop. — Ed Pastore 22:12, 1 September 2008 (UTC)

Yes, I think we all want a model where every threshold is a variable that can be set by the community. Although this will make a hell of a discussion every time someone want to define new variables. I mean, do you need a different consensus to modify the variable that defines consensus? Like in general you need a much higher level of consensus to change the constitution in a modern country? I think we will have to face that question lately. As part of the general question, are all laws the same or some are slower to be changed.

Let me think on that, and see if we can come up with a nice continuous formula, that includes everything.

Yeah, now that you mention it... it is a bit of a catch-22 to require a consensus to define what a consensus is. But I also think you are right that a sort of super-consensus (i.e., a higher standard) could be used. Perhaps even a multiplier of the current consensus definition... that is, something like halfway between the current value and 100%. — Ed Pastore 22:07, 1 September 2008 (UTC)

Back to the consensus. I like how in some communities (Findhorn comes to mind, although I am not sure), there are more than one voting way: you can say that you are:

  • for something and ready to help out int hat happening
  • for something but someone else should help, as you will not (busy, uninterested, whatever)
  • you don't care (equivalent to not voting, but it changes the number of people who have voted, plus no one can say you are ignoring the issue)
  • against, not in my name (a vote against but not really strong, you would not go out of your way to stop the resolution from happening, in case)
  • strongly against, I want this not to pass (a strong vote against, a person who votes against in this way is practically saying that he is willing to start a hunger strike to stop this vote to pass, or similar)

We have to understand that the "against", and the "strongly against" are not just quantitatively different, but quantitatively different. I can be against many things. But strongly against means that I think is not just a wrong move, but make huge, diverse, uncontrollable damage.

As such I think we should have 2 thresholds. How many people are against, and how many people are against or strongly against. I don't think we should have that consensus means that no one should be strongly against. This is because some people might not have just the best of the community interests to their heart.

Now we should also recognise that according to some psychological studies (cfr. the sociopath next door, from Martch Stout) there is about a sociopath every 20-25 people. That is 4%-5%. Those are people who have no sense of belonging to the community, I mean, to the HUMAN(!) community. They feel no pleasure in connecting with people, no pain in being apart. They can feel boredom, and domination. But they don't feel love, and affection. They actually are able to go through an incredible amount of social pain, and even torture without having psychological problems. They make great soldiers, but are really not the right person to make decisions for the community. They actually feel pleasure in dominating and/or torturing people or animals. Now we can expect the karma to take care of not having those people in charge (btw, when they rise atop of an organization that is when REAL problems arise: if I remember well 60% of normal people are willing to become torturer if a white lab authority orders them too). But still I would suggest a threshold of not higher than 95% among people with equal karma.

SO I would say we should have a (suggested) threshold for the strongly against of 95%, and a threshold for the against which is between 95% and 51%. Maybe about 67%?--Pietro 08:50, 1 September 2008 (UTC)

While I admire the logic behind the categorization, I do think there is some value in the "Yea/Nay" construct. What politicians understand very well, and understand it to the point where they actively exploit it against each other when appealing to the general public.. is that nuance and context is _everything_. Whether they voted for something which they didn't fully agree with because it was a choice between an incremental change for the better and nothing, or whether they voted against something they mostly believed in because something contagious was bundled with it -- evaluating the motive behind the yea or nay is impossible without a lot of research. See Obama and the "present" votes in the Illinois legislature for a current example of this.
One consequence of retaining a "Yea/Nay" construct would be that more people would be exposed to this knowledge through practice, rather than dry theory - which is easy to understand but not emotive until you've experienced it yourself. As Metagovernment would, by definition, substantially increase the percentage population who are voting upon matters.. more people would be forced to explain to friends and colleagues the rationale behind their voting choice.
This process would seem likely to generate a wider appreciation of the nuances involved in the issues being discussed, and surface-paradoxically a deeper understanding that "Yea/Nay" do not automatically equate to absolute ideals. Of course, a wider appreciation of nuances would seem likely to lead towards better proposals.
Adding additional weasel-categories, allowing people to "hedge their bets" does not force them to evaluate the issues any more deeply than they already have. Information like "I would be willing to help", and "I would go out of my way to stop this" would seem to be valuable but meta -- maybe interested parties could attach additional voting categories to issues as and when required? — Richard Franks 16:48, 1 September 2008 (UTC)
If we have some sort of synthesis scoring, then it seems to me that we need to have more than just up/down votes on proposals. I don't see a problem with people saying, "I think this idea is pretty good, that one is excellent, and another is horrible." This is what I had been proposing with the concept of nuanced, recursive scoring, which is what is currently described on the main page (though seems to have been dropped in the proposed revision. — Ed Pastore 22:07, 1 September 2008 (UTC)
Ah! I misunderstood synthesis scoring to be a bit simpler. That is, when a revised proposal comes up, individuals would be able to say with a boolean yes/no vote "I think I speak for the rest of the community when I say that people who voted for X, will vote for this better version which is Y". Everyone in the community would be still be able to vote yes/no if inclined. Individuals who prove not to be speaking for the community, or who are trying to abuse the synthesis scoring process, would then find that their future weighting ability to be diminished -- and those who understand the community better would bubble up their weighting value.
It still gives the potential for abuse (storing up weighting to push individual issues), but it makes it easier to detect that abuse, and synthesis scoring would (presumably) be overridden by actual votes.. which is more likely for issues which would become controversial because of detected abuse.
It has the advantage of separating votes from debate, where the purpose of voting is to reach consensus, regardless of the compromises individuals have to make in their decision making process? — Richard Franks 22:46, 1 September 2008 (UTC)
Furthermore, assuming that:
  • Each user has a box on their homepage showing which proposals they are currently being used as synthesis votes for
  • Proposals which finish the voting process will list all proxied synthesis votes
Then there should be some additional motivation for users to keep an eye on that box. This might mean that it could be expanded to allow more advanced and nuanced constructs such as: "People who voted Yes for X and No for Y, will vote Yes for Z". I'm going to think about this some more but I'm tempted to hack together a quick demo to see if it works as described. — Richard Franks 23:10, 1 September 2008 (UTC)
Hello Richard, it took me a lot to understand what you were saying. But now that I did the work I am not sure that I like it better than when I was not understanding. So maybe I am not really understanding it still. So, please let me see if I got it correctly: You say that in politics where there is only a yea/nay, people vote yes or no depending on many reason. Forcing people to vote yes or no, they would feel, understand and come to appreciate the nuances of it. Which would lead to them understanding the issue better. And would lead to them writing better proposal. Whereas giving them different types of votes would not force them to look at the issue more deeply. This I find be the core of your message.
If this is what you said in your first message, I will now answer, if this is not, please can you clarify it better. In my experience even more complex concepts can be explained much more easily, so I invite you to do just so.
How about - an individual votes, not for themselves, but to serve the community of which they are a member. It serves the community best to reach a firm decision which can be acted upon. An individual who has the thought that "if the community passes this vote I would try to stop it from happening in other ways" should probably spend their time instead finding (or creating) a community which better suits their views.
There is a time for debate and a time for voting on action - the community and the present environment tend to determine the delineation as an emergent process. Why should a community support an individual who is working against its interests? Or why should a community make it easier for the individual to do so? — Richard Franks 15:54, 2 September 2008 (UTC)
And now my answer. I am surprised by how convoluted your reasoning is! Now politicians vote yes and no for different reasons, agreed. Instead of just focusing in making sure that the decision that comes out of the voting process is the best possible, it takes care of all the possible weak spot, and it represents the will of the community; instead of this you want that the new democratic process, instead let people understand the nuances of the reasons that leads (today!) politicians to vote yes and no, and then use this to build better proposal. Why? I mean, who give a f*ck why politicians now don't represent the community? We know they don't, we want to make a system that gives people a voice. If the system is in place the reason why today politician vote in a certain way would not hold anymore. Why? Because when the power is distributed enough you simply cannot buy all those votes. It is the reason why juries in ancient greece had thousand(s?) of people. Our aim is not to understand why politicians are corrupt, our aim is to have a politics which is not corrupt.
Second point, forcing people to understand all this, will force people to look deeper in those issues. Maybe. Maybe! the truth is that the more you force people to work, the less people you will find willing to follow you. Also the resons that lead politicians to vote yes and no will not be the reason that lead simple citizens to vote yes and no. Why? because, again, you can buy a politician, it is harder to buy a whole population. SO they would not understand those nuances, simply because many of them would not hold for them.
Third point, understanding those nuances would lead to better proposals. Maybe. This claim is just unproven.
And finally you say that asking for different categorisations of vote would not lead to people to think the issue more deeply. First I challenge this, having to answer, if you are going to be active or not respect to an issue, does help people clarify their position to themself. But what is more important, it clarifies the position to everybody else.
Richard, having just spent the best part of 1 hour in answering this massage, after having yesterday evening spent a long time in reading your message, I am full of frustration. I wish the discussion here to be an encounter of people who have the same aim, and who try as hard as they can to make themselves understood. This was always the feeling I had with Ed, with Manuel, and with all the other that joined the mailing list. I don't have this sensation this time. SO could I please ask you to be more clear, and more precise in your messages? We are all here investing precious time in writing this wiki, working in the program (or the formula), time that if not would be spent with our family, or on vacation. Many thanks,--Pietro 08:48, 2 September 2008 (UTC)
I'll get the ad hominem out of the way first. I am not trying to be cryptic, imprecise, disruptive or uncooperative with my messages, and I am sincerely sorry that you have perceived them to be so. I understand that you are feeling frustrated, but I do not find it very civil to respond to a proposal with a "you, and specifically you, are wasting my time". As this sandbox grows, many people will challenge your ideas - while emotions run high sometimes - I believe progress will be better made keeping the discussion at an intellectual, not emotive level. To borrow a line - we can disagree without being disagreeable?
"Our aim is not to understand why politicians are corrupt, our aim is to have a politics which is not corrupt." - I'd argue that much which we perceive as corruption, while fitting a broader narrative which may include some amount of actual corruption, could also be objectively viewed as honest compromise. There are some politicians whom I wish would never step foot in office again, and while partisan sources parse every action of theirs in the worst possible light, I find myself still having to give these otherwise despicable politicians credit where it is due. I labour this point because you use this absolutism to argue that 'yes/no' could never, not ever, possibly work. I believe that you are conflating "a system that gives people a voice" with "a system that gives voice to every person" -- the former is achievable, the latter is akin to voting not upon clear enactable categories, but upon the multitude of opinion-flavoured posts that appear in a discussion forum.
Second point - "the truth is that the more you force people to work, the less people you will find willing to follow you" - agreed in principle, but if there has to be a bar, the question instead becomes how high should it be? I would be perfectly happy to see those who are content to think an issue through in some detail, be the early adopters. As Metagovernment currently stands, that is actually the case now -- without a software system to work upon, we are crafting some of the most fundamental precepts the hard "old-fashioned" way. We continue to try to lower the barrier of participation, and beyond a certain point - the release of version 1.x of Metascore, it is safely out of our control - and the users can redeclare the precepts as they wish, and as it should be.
A fine counter-point to this might be "the bar should be lowered as much as possible, as quickly as possible". While it is a fine sentiment - you, yourself, in attacking me - argue against it elegantly.
"Third point, understanding those nuances would lead to better proposals. Maybe. This claim is just unproven." -- please forgive me, but maybe you could explain why it is not an obvious assertion? Also, how can I be expected to empirically prove an assertion about a hypothetical system?
"having to answer, if you are going to be active or not respect to an issue, does help people clarify their position to themself. But what is more important, it clarifies the position to everybody else." -- counterexample proposal - "should we lower income tax by X% to Y%?" - agreed that activity levels are useful metrics in some cases, but you haven't answered my suggestion that perhaps they would be better placed as secondary (optional) voting positions for those proposals which would benefit from them.
A proposal passing with a thousand caveats is the antithesis to consensus. — Richard Franks 15:10, 2 September 2008 (UTC)

Sectioning resolutions

How about all policy submissions are broken down into sections (much like they are now in the U.S.) based on the concepts they are proposing. Each section has a for/against check-box next to it. at the bottom of the submission, there would be a large for/against. If one simply clicked "for" at the bottom of the submission, all sections would automatically be marked "for" as would happen with "against". however a voter can click "for" have all sections be checked "for" then go and vote "against" on a specific section. If enough people vote for a policy, but against 1 specific section, the policy will be law pending a re-write and re-vote of the section in question. basically a policy could pass but not be instituted as law until it was perfect and it would protect against unrelated "attachments" (such as the business reform act which contained the entire text of the methamphetamine act).Symetrist 17:11, 3 September 2008 (UTC)

While the price of liberty is deemed to be eternal vigilance, or something similar, the structure you propose would _require_ vigilance -- it would be harder for "attachments" to pass, but it would still be possible. Assuming that time permits, perhaps individual sections could be voted upon individually, and then bundled together to create another for/against resolution which references those sub-sections which have already received individual attention. Technically, there wouldn't need to be much if any distinction between sections and resolutions, as each could reference and be referenced.
Actually, I think that might be quite close to the synthesis voting idea? — Richard Franks
We had a similar discussion early-on in the list server and Manuel thought it would be burdensome to implement. Also, it seems to me that resolutions can end up having strange consequences if they get only partially passed. For example, what if the funding plank gets shot down, but the rest passes?
And yes, the concept of synthesis scoring is an attempt to be a more elegant solution. If someone likes everything about a resolution except for one section, then they make a new resolution with the changed section and invite people to give it a synthesis score. If enough people say that it is a better synthesis than the original, then it should be able to supersede the original. — Ed Pastore 19:24, 5 September 2008 (UTC)
"what if the funding plank gets shot down, but the rest passes?" I was trying to account for that when I said this: " If enough people vote for a policy, but against 1 specific section, the policy will be law pending a re-write and re-vote of the section in question. ". Say for example your scenario happened, a proposal was agreed upon unanimously, save one section regarding where the funds originate. It would be considered law, but wouldn't be officially implemented until a re-write of the section in question was voted on and approved. Basically the people would be saying "we want this to happen but we don't want to divert funds from our schools to pave a new highway" or something like that. Also, every resolution could have comments at the bottom (like a blog post) and people could then explain why they were against something etc. Others in agreement could "digg" what they're saying which could give them the "karma" etc for user ratings. Symetrist 19:54, 5 September 2008 (UTC)
Ah, I see. Yes, that does make sense. However, it seems like it introduces more complexity than the synthesis scoring concept because there still may be many competing resolutions. If there are twenty resolutions, each with fifteen different sections, it would be burdensome for people to consider going through all that detail. If instead there were fourty resolutions, but each was being directly compared to others through synthesis scores, then the user burden is not so intense. Nonetheless, I'd like to see what some others think, since what you're proposing is, imo, otherwise reasonable. — Ed Pastore 20:32, 5 September 2008 (UTC)
My understanding of synthesis voting was that it was in lieu of actual voting. That is - the more people in a community who vote upon an issue, the less effect synthesis voting would have - where more actual votes are likely for controversial issues. If that is correct then one situation which synthesis voting allows for which Symetrists suggestion doesn't, is the case of contextual compromise. E.g. where no-one wants to reduce funding for schools and so wouldn't vote for that section, but is prepared to accept the overall resolution for some badly needed hospitals.
This seems to retain some of Pietro's idea of recording varying levels of support while still achieving consensus. I think this is conceptually different from "attachments" assuming that this is a well discussed issue and the case for the compromise has been well argued. Fundamentally it doesn't stop a better resolution from being tabled, but assumes that one either doesn't exist or hasn't been found. It further assumes that the community will discuss resolutions more heavily which are controversial - e.g. not all sections passing consensus, but overall resolution consensus would be an obvious red-flag. — Richard Franks 21:46, 5 September 2008 (UTC)
What if there were some sort of maximum number of resolutions which could be voted on for a given time period? For example there could be a max of 3 per week (or another agreed upon number and time period). Proposals could be submitted at any point but instead of being immediately "up for vote" they would be queued for vote at a later time. Proposals being queued could be viewed at any time for people to "study up" on the issue which would also give them more time to make an informed decision. Also, by incorporating the user ratings/karma a proposal could "cost" karma. for example say I have 150 units of karma, diggs etc and a resolution costs 100 to post for vote, I would spend my points proposing the legislation and my karma/diggs etc would be reduced to 50. Obviously there are some scenarios where a vote would need to take place immediately (i.e. an asteroid is headed towards earth and they need to decide the best course of action). In scenarios like that...I'm honestly not sure what proper protocol should be. Symetrist 00:21, 6 September 2008 (UTC)
I think as a general rule, we should not be afraid of information overload - assuming that there will always be more information than an individual can consume, the question instead becomes "how do we filter/present to the user such that they get the information most relevant to them?" I'd use that philosophy to argue against any hard (quasi-static numeric) limits on procedural items such as max resolutions in a given time period. Scalability is a requirement, not an option? — Richard Franks 17:53, 6 September 2008 (UTC)