So... the longer I think about this Wiki thing, the more it seems like a really good idea - a missing piece falling into place.
Here's my proposal, which I turn over to this, the larger community that suggested the Wiki in the first place:
The Wiki should consist mainly of short concept introductions plus links to longer posts, rather than original writing. Original writing goes in a post on Less Wrong, which may get voted up and down, or commented on; and this post should reference previous work by linking to the Wiki rather than other posts, to the extent that the concepts referred to can be given short summaries. The intent is to set up a resonance that bounces back and forth between the Wiki (short concept summaries that can be read standalone, and links to more info for in-depth exploration) and the posts (which make the actual arguments and do the actual analyses).
My role model here is TV Tropes, which manages to be, shall we say, really explorable, because of the resonance between the tropes, and the shows/events in which those tropes occur, and the other tropes that occur in those shows/events. And furthermore, you know that the trope explanation itself will be a short bite of joy, and that reading the further references is optional.
There would be exceptions to the "no original research" rule for projects that were multi-editor and not easily prosecuted through comments - for example, a project to make a list of all posts with one-sentence summaries in chronological order.
There are also obvious exceptions to the "link to the wiki" rule, such as for any case where it really was futile to reference anything except the complete argument; or where you wanted to talk about part of the argument, rather than the general concept argued; or when you wanted to talk about a conversational event that happened in a particular post.
I would suggest that the general format of a Wiki entry be a short summary / definition (that can maybe gloss or list some of the arguments if there's room, or even give the argument if it can be put really briefly), with the body of this being no more than a screenful as a general rule. Then links to posts, with descriptions (for each post) of why that post is relevant or what it has to say - probably one or two sentences, as a rule. Then references to outside posts on the same topic - although if the best reference is an outside discussion, that could come first.
Summaries of whole sequences could also go on the Wiki - since it seems more like static descriptive content, rather than debatable analysis and argument, which is how the wiki/blog dichotomy is starting to shape up in my mind.
Given unlimited development resources we'd want to integrate the two userbases, have a karma requirement to edit the Wiki, and such things, but we don't have much development resources (whines for Python volunteers again). But I would still like to see a list of recent edits and/or active pages in the Less Wrong blog sidebar, and a list of recent blog posts and recent comments in the Wiki sidebar. Of course the first priority is getting the Wiki set up on Less Wrong at all, rather than the current foreign host - I'm told this is in In Progress.
Once my old posts are imported from Overcoming Bias, it would be nice if someone went through them and changed the links to posts (that reference concepts per se, rather than conversational events or parts of arguments) to links to the Wiki, creating appropriate pages and concept summaries as necessary. That is, it would be nice if I didn't have to do this myself. Anyone interested in volunteering, leave a comment - it'd be nice if you had some comments to your name, by which to judge your writing skills, and perhaps some karma. This is a large job - though 10 posts a day would get it done in two months - so don't step up if you don't have the time. It does seem like the sort of thing that should definitely get done by someone who is not me.
I've already seen at least one person call Overcoming Bias "a bigger, and far more productive, time sink than Wikipedia or even TV Tropes". Done right, it really could be more addictive than TV Tropes, for the intellectually curious, because it would seem (and be!) more productive; when you browse TV Tropes it feels like you're wasting time.
I suppose I should feel slightly nervous about this, but it still seems like something that ought to be done if feasible, even though it sounds a bit scary - and I hope I'm not just saying that because I'm tempted to break out into mad scientist laughter.
Thanks for this reference. This concept is what I was going at at the IRC meetup. The main disagreement with Eliezer's model seems to be that he thinks that the blog posts still have to hold the majority of content, with wiki only referencing them with very short introductions, whereas I think that the wiki should grow into a thing in itself over time, converting the content of blog posts into wiki articles. Naturally, articles should be organized in a zoom-in manner, with few-sentences summary, then couple-paragraphs introduction, and only then full-length article. Each of the levels of detail are self-sufficient, and this is also a model implemented in Wikipedia.
The difference between Wikipedia and this CIA model you describe is that for Wikipedia, the original research is done anywhere, by the rest of the world, while for a thematic wiki, the original research is communicated and discussed on the associated blog. I'd like LW to embrace this model.
That penultimate line should say
Shouldn't it?