See also DebateTool.

This idea has been in the ether awhile and we may have even discussed it already (please direct me to the appropriate page if so), but I haven't seen it implemented yet so may as well put it into words:

Some sort of online hypermedia thing containing contentious propositions, such as political ones ("church and state should be separate", "the US should work towards to ouster of Arafat", "welfare is a good idea") (maybe framed more precisely, but you get the idea).

Each proposition is a node in the network.

If you feel strongly, you register to agree or disagree with the proposition. You then cite your reasons/arguments.

The arguments are themselves propositions and are hence hyperlinks to other pages.

Users can register various ways to say who they agree with; the opinions they see can be sorted by this (mechanisms may also be included to let you see things from totally different points of view).

You also may certify whether you feel there is a "consensus" as to the veracity of falsity of a proposition; and which group has the consensus. For example, a layman like myself might look up "value-added tax is good" and find that, say, there is a consensus among economists that this is true or false, something which i didn't know. Or look up "global warming is happening", and find that there is a consensus among atmospheric scientists that this is true.

Finally, like "LiquidDemocracy", you should be able to give a proxy to others on certain issues. for example, I would give a proxy to Prof. Daphne Koller on propositions about probabalistic A.I. Then if you were looking up something on A.I. and you had noted me as someone whose opinions you are interested in, and I hadn't written an opinion on whatever you were looking up but she had, her opinion would show up on your screen (i.e. would be selected for viewing by the moderation system).

The point is, for example, you can look up an issue, such as "our defense budget is too big", and see pros and cons. Then you can look up the pros and cons and see arguments supporting or knocking down these arguments. you can go as deep as you like; at some level you will probably go beyond your expertise, in which case you may still be able to reach a decision (if one is needed) by trusting an expert who has an opinion at that level. More importantly, you will maybe be able to find out when certain (authoritative to you) communities have a consensus for or against something that you thought was undecided.

It would also be eventually desirable to authenticate experts as to their expertise and as to their correspondence with the real world person whom they claim to be. this other level of complexity could be added later.

why not just a wiki?

I guess a wiki could do most of this, but it couldn't do the moderation/filtering of commentators to show only those you trust because it has no concept of individual comments on a page. the moderation is absolutely key here, though, because the user wants to be able to tell at a glance if an argument holds water, even if they don't have time to evaluate that argument themself (i.e. the "existence of global warming" node; you may have arrived at that node from a larger issue, say "pollution controls are good/bad", and maybe you are not interested in the technical details of global warming, you may just want to know very quickly whether this contingency is at all credible. So the sort of thing you are looking for is a very short something along the lines of "there is a consensus among scientists that global warming is occuring, there is no consensus among the american public, Prof. X whom your friend likes thinks there is global warming".

Also, a wiki page may end up listing ALL the pros and cons on a proposition, while what you are interested in is only the important and credible ones.


A method needs to be introduced to allow wiki-like consolidation of similar commentary. A proposal:

1) Each page has a top section which is a wiki page (publically writable). This section explains succinctly the argument or proposition expressed by the page. Most people voting agree or disagree will not need to write any personal comment, they will just indicate links to their reasons for agreeing and disagreeing (and maybe the relative weights of their reasons). Theoretically, there will be no reason to write any separate commentary; anything you could want to say could be put into terms of "i agree because", or "i disagree because", and then the stuff after the "because" is contained on the node dealing with the appropriate subargument. (you may also agree with a subargument but disagree that it is relevant for some reason; this is sort of a complicated sort of subargument that is attached not to the main argument or the subargument, but to the link between them.)

2) People will still be allowed to post their own comments, however, indicating various unforseen comments that don't fit within the system for whatever unforseeable reasons. people will be able to "synth", or synthesize, multiple redundant comments into one, in which case a note will be sent to the authors of the comments that were synthed; with their permission, the old redundant comments will be deleted and redirected to the synthed version.


Lemme know if this has been done.

See also DebateTool.

PropositionNet sounds like some technical A.I. logic thing. DebateNet? sounds too specific to formal debate. ArgumentNet? sounds too contentious. What is a good name for this?

:How about DecisionNet?? ::I think this framework could ultimately be used for decisions if that were desired, but i think it would start off being used only to gather information and summarize the points of view that were out there, rather than deciding between them. Calling it DecisionNet? might make people think otherwise (i.e. that the system was for deciding between varying points of view) – BayleShanks

(… in response to another post …)

Your concerns are valid. that part of it is sort of organizational, but I think it'll dictate the tone of the whole thing. I am envisioning each page being rather short and concise, with the whole thing being a highly modularized structure. most written expositions of one side or another would be broken down into lots and lots of tiny blocks. Many arguments would be little more than a sentence along with links to supporting subarguments (ie. the "pollution limits are good" proposition might not say anything besides referencing definitions of those terms and referencing subarguments like "pollution limits help prevent global warming". Beneath "pollution limits help prevent global warming" might be such supporting propositions as "global warming is harmful" and "global warming is a real threat". "global warming is a harmful" may have little text but be mostly links to things like "global warming will destroy ecosystems", "global warming will hurt the economy", etc.

This is not the standard way that arguments are presented but I think it would be useful to have them in that form. I want to minimize the amount of time the reader has to spend if they want to be quick. For instance, say I believe global warming is occuring/is a threat, and I am reading an article on pollution limits. It is inefficient to have to trudge through the stuff about how global warming is a threat, because I already believe that. Similarly, sometimes article make arguments based on assumptions that I really do not believe. In this case I try to disregard the part of their argument resting on that assumption, but I still have to read over it.

A more important reason for putting stuff in such a modular form is to allow evidence to be precisely attached where it belongs. Footnotes work for this but only to a point; ideally, new subarguments should be able to appear wherever supporting references appear.

I agree that reading stuff only in this form would actually make it harder to get a summary of arguments, though. But there's a pretty easy solution; allow the user to browse with "lookahead", i.e. if you are looking at "pollution limits are good" with a 3 level lookahead, you see "pollution limits help prevent global warming", and indented beneath that is "global warming is harmful" and beneath that, indented, appears "global warming will destroy ecosystems". Eventually you'd want your lookahead to be filtered, i.e. if most commentators who you trust say some argument is "absolulte nonsense" (i.e. maybe it is not an argument at all but spam advertising or something), then maybe it won't be displayed at all, or maybe it will be displayed but its children won't.

I bet a few more additions would make this even more readable. for instance, if you want a quick overview, you don't really want to see clutter like "global warming is harmful" between "pollution limits are good" and "global warming will destroy ecosystems". Such arguments could be marked as "hide for summary" or whatever by commentators and would be hidden by default (i.e. the computer invisibly, recursively replaces these with their children in displays; I would turn summary mode on by default).

Also, some individual arguments may take a little while to properly explain (i.e. stuff about fossil records and evolution), in which case you could have a short summary that would appear in lookaheads, along with a link to a longer, clearer explanation (for example, "scientists have found fossils that show similar creatures that appear to be evolving").

So in the network, every argument would be chopped up into real tiny pieces, but you'd be able to display a lot of these pieces at once in order to create a summary, with the help of tags on the various propositions that help tell the computer how and when to display them.

This sort of capability is another reason that a Wiki wouldn't be able to do the job.

Btw, what sort of data structure is this? it is not exactly a tree; for example, the truth of the proposition "global warming is harmful" does not exactly support the truth "pollution limits help prevent global warming"; rather, it helps support the validity of the link between the proposition "pollution limits are good", and "pollution limits help prevent global warming". the general shape of this sort of structure is clear, but does it have a canonical name? — BayleShanks

Hmm, this seems fairly close to TypedThreadedDiscussion - at least technology-wise.

Maybe it'd be worthwhile to try to split up the ideas in here ?

  • Making a computer-readable map of the contentious propositions ( → MappingArguments)
  • Allowing users to formally express their agreement with propositions
  • Allowing users to note if there is a 'consensus' (from specialists) on that issue
  • Building some kind of WebOfTrust - akin to LiquidDemocracy

Some of these sub-goals are valuable in themselves, and may be more reachable.

I suspect the biggest problem might be convincing people to use and trust this system.

CategoryInformationManagement CategoryDecisionMaking

Define external redirect: ArgumentNet DebateNet DecisionNet

EditNearLinks: WebOfTrust


The same page elsewhere: