Proposal for a structured agreement tool

How do you plan on profiting enough to pay for hosting and maintaining the site?

For reference, hosting a lightly-used site is really freakin' cheap.

Thanks for the comments. I will try to address them.

I am less concerned about hosting/maintenance than I am about development costs. If it is not successful, then it won't be large, or need large bandwidth - costs will be low. If it is successful, then some sort of freemium, ad-supported model may be possible. Of course, the territory between these two extremes might be awkward. Joel Spolsky [http://en.wikipedia.org/wiki/Joel_Spolsky] has taken his Stack_Overflow engine [http://stackoverflow.com/] and gotten VC funding for an attempt to roll it out as a social network platform [http://stackexchange.com/].

All sorts of propositions are acceptable, as I tried to make clear - although I do think that it will be important to seed the site with 'good' models - perhaps some of the 'sequences' here at LW, for instance. It doesn't matter to me if the system is used by a cult who think that John Lennon was the second coming of the Messiah.

On the 'nit-picking' point, there is a characteristic of the model offered by 'A Pattern Language' which might help - the Conclusion section of the proposition is stated deliberately in open terms, rather than attempting to be rigourously prescriptive. So, instead of 'Thou shalt not kill', one might conclude that killing people is almost never justifiable, and offer links to other propositions on, say 'self-defence', and 'assisted suicide'. One need not aim for a structure that would satisfy a formal logic analysis (although there is no reason not to attempt such). My own propositions would be a set which describe my own web of beliefs, in a way that is cloudy enough to admit nuanced interpretation, but clear enough to be useful. I would hope that this would make it possible;1 for others to register interest in them, and 2 for me to accept refinements suggested by others. I think that many of EY's posts conclude in the same way - offering tools and strategies, rather than hard-and-fast rules.

I think you are right about forking. However, I would contrast this with the status quo. I have been trying to locate my own 'community of interest' on the web for over a decade - I believed that the internet would allow like-minded people - particularly ones with minority interests that could be adequately communicated in words - to find each other. Sites like LW are the poor best that I have found. Why poor best? Well, while there is much of interest here, I am already concerned that certain ideas I hold dear may prove to be anathema here. EY set the site up, and it has a clear and strong thrust. It would be foolish/rude/ineffectual to push ideas which are antithetical to that thrust here. So, sites like this are too rigid - forking is not allowed. More general forums are too loose for the concept of forking to even have a meaning. You see, I would be happy to work with even 5 people at a very high level of agreement. I imagine that there would be many sets of groups which would share broad foundations of shared propositions, but grow increasingly fragmented as the propositions began to operate at more detailed levels. Perhaps it will be important to allow associations of groups,

Precisely these sort of considerations are what I hoped for. Even if this remains a thought experiment, I will find it immensely valuable.

While I am certainly not against the idea of a tool that can be used to create formal arguments, the proposal has a subtle but radical difference.

DISCLAIMER: I am not a mathematician, and do not fully understand the concepts I attempt to explain in the following.

In his work published as 'Notes on the Synthesis of Form', Chris. Alexander developed an algorithm for converting a matrix of relationship strengths between analysed sub-elements of a design problem into a 'tree-like' structure. In other words, a hierarchical diagram in which each node can have one connection only, to a higher status node. The number of nodes in each level decreases as one moves upwards, culminating in a single 'master' or 'root' node.

Following the success the publication of 'Notes...' brought, Alexander was employed to work on the development of the metro rail system in San Francisco (the BART), and put his method to work. As a rationalist, he was concerned to find that the results of his work appeared to be failing to fully address the realities of the design problems involved.

His conclusion was that the necessary function of his transformative algorithm which selected the least significant relationship linkages to be broken in order to derive the tree-like diagram was the cause of the problem; some identified real-world relationships were being ignored. And even though these might be ranked lowly, omitting them altogether was destructive.

The essay which captures this understanding is published as 'A City is not a Tree' - read it here: http://www.rudi.net/pages/8755.

In it, Alexander contrasts the tree-like diagram with another; the semi-lattice diagram, which, although still hierarchical, allows for connections across branches, as it were, so that overlapping sets of relationships are legal. Semi-lattices, I believe, are not susceptible to formal logical analysis, but nevertheless can be better mapping tools for complex, real-world systems.

My proposal would deliberately allow for semi-lattice linkages. This would allow, to come up with a quick example, a proposition that called for more cycling to link both to a proposition for less carbon emissions and a proposition for congestion reducing transport initiatives.

Tree diagrams are fairly useless in addressing real-world conditions, as these are usually formally complex, with elements occurring in overlapping sets more often than not. As a result, policy documents are not structured like tree diagrams, and do adduce all sorts of linkages, but do this in a totally unstructured manner, and are often functionally weak, while appearing to address everything. As EY says (everywhere); "A theory that can explain everything, prohibits nothing, and so gives us no advice about what to expect."

My hope for the proposal is that it could bring structured, coherent agreement on sets of principles without the need for total agreement on every aspect of every point.

This looks very interesting - I will have a more thorough look and report back.

It's good to have positive responses!

Recommendations: I have a feeling that this will turn out to be relatively trivial - as I see it, it is essentially a parallel problem to the one that google solved, but with a data set many orders of magnitude smaller. The system surveys the relationships between propositions that have been supplied by users (I don't think it need/should infer connections for itself), and determines a relevance score.

Moderators: I am sure you are right, and that this could easily be a fraught area. Ideally, the system would be flexible enough to allow for groups to self-identify and self-manage in a variety of ways - unlike wikipedia, where a single ecosystem is the be-all and end-all, there is space here for a wide range of ecosystems - groups would certainly rise and fall, fail, feud, coalesce; but no propositions will disappear when a group associated with it collapses - the propositions remain for other users to find and work with. The system should be set up as an evolutionary playground, with mechanisms that are rigged to marginally favour groups that coalesce more, flame less.some sort of ranking system would allow neglected propositions/networks to fall to the bottom.

Groups: I believe that as humans are a social species, human communities are essential structures and the necessary (but not sufficient) locus for creative responses to problems (even if the responses are associated with individuals, those individuals will generally be more effective within a supportive community). However, the communities of the past generally used restrictive and non-rational codes as social glue (variously, religion/tribal culture/geographic isolation/economic dependence/political domination etc) People who wish to use their consciousness to improve their lives have often had a hard time in relation to these codes, and at an accelerating pace in the last century, have abandoned traditional communities - leaving these dominated by the happily or cynically non-rational. Secular cities with weak communities, fundamentalist rural areas with strong but misguided communities. I am interested in developing tools and mechanisms which can allow people to form communities on the basis of conscious assent. I am enough of a darwinist not to want to attempt to say what I think those communities should aspire to.

I would use such a tool to attempt to set out my own web of beliefs and interests, in the hope of connecting with like minded individuals. My hope is, that with a large subset of the 'ground rules' clearly communicated and the relationships between them identified, that highly effective development of new/difficult/ill defined areas could be worked on. I would also use such a platform to make proposals like the one we are discussing here. I would expect certain groups to turn up quickly; cults; obsessives and the like - but I fondly hope that these will either remain tiny or will disintegrate or, most likely, find that the structural characteristics of the tool tends to expose inconsistencies in their thinking, and leave of their own accord. I would expect that political groups might follow the same trajectory, albeit more slowly and with less fireworks. I would hope that purposeful individuals with positive and constructive intentions will find it a more congenial arena that wider forums or more narrow blogs.

A concrete example: as a founder of a small school, I attempted to establish the principal policies of the school as a network of 'patterns', ranging from fundamentals (Support the developing child; Learning is part of living; Self-governance; Family involvement is fundamental; Ethical consciousness and action) to intermediates (Relationship with the state) to specifics ( In the absence of a tool like the one proposed, collaboration with other people involved was impractical, and we retreated to the standard solution of sets of prose policies, which no-one ever reads, and which are inaccessible and becoming less relevant to actual practice with every passing week.

As to what groups might accomplish, I can say this very succinctly; they might well succeed in expanding the range of metaphysics available to humanity. By which I mean, expanding the number of ways in which we can usefully communicate about phenomena. As this seems to me to be the only reliable yardstick by which to measure 'progress', I modestly suggest that this might be a Good Thing.

Business Plans: given some sort of 'freemium' model, than yes, users/groups wanting privacy / security would be able to pay for various levels of same, or presumably for implementation of stand-alone installations on their own servers. Open-source software offers successful models for this (Drupal is one example).

"basically, it's an interactive prose design document" - I hope not. The structured and templated nature of the Pattern Language approach has a higher order than prose, and should increase both the information content and the utility of the resulting collection of propositions. The book 'A Pattern Language' , which was published in 1977, and conceived years previously, works exactly like a hypertext document, with directed but essentially free-form navigation the implied mode of use. The requirement that all content is directed towards the resolution of the particular problem/proposition at hand is rather powerful in practice, and the requirement that each proposition should be assigned larger scale propositions which it helps to refine, and needs smaller scale propositions to flesh it out in turn imposes a need for clarity of thought about the outcome of the system as a whole.

Engagement with the project: at this stage, comment and discussion is what is needed.