GeneralWikiLog

MeatballWiki | RecentChanges | Random Page | Indices | Categories

This is RustyFoster's original description of a "general" WikiLog -- something that can act as a wiki, as a WebLog, or something inbetween. Similar to ViewPoint, it describes in broad terms a possible implementation of a WikiLog system, a vision to guide developers, and a basis for discussion of the various features involved.

Basics

Take, as a starting point, a system like the ScoopEngine. Then imagine that instead of "stories" and "comments" being arbitrarily different, what you have are "posts". A post is any block of text submitted by a user.

Posts have some properties:

Content: The text itself
Author: Of current version, and of previous versions. "Contributor's list" might capture this better.
Categorization: Like topic/section, but could be more complex/flexible. Actually, this is just a part of lineage. Imagine a "topic" as simply a parentless post that has no content, just a "topic" descriptor. If you want to tag a later post as part of this topic, you just need to make sure it inherits from the parent topic post.
Lineage: "Post Y was a response to Post X", or, more generally, "Post X inherits the following properties from Posts A, B, C, and D. Lineage can be multiple-- thing about this in terms of an OO "multiple inheritance" paradigm here.
Revision History: "This is v.15. It was submitted by Joe.", plus all the previous versions and associated metadata
Ratings: Individual user assessments of value. Consider Scoop-style 1-5 rating as a starting point, but this go elsewhere.
Statistical Metrics: Collective stats, gathered from ratings, including average score, high/low, std dev, etc.

Now, consider that a "post" is equivalent to a "page" in a wiki. Each one is editable by anyone (potentially, access control lists could come in here-- it may be that every post is not editable by every user, but for basics, assume they are). So, a WikiLog would be equivalent to a WikiWiki in the case where each page contains one and only one post.

However, a post and a page are not necessarily equivalent. Posts are more or less free-floating in the system. A post can be a thread-style "reply" to another post, or it could be less a reply and more a derived concept, which deserves to be associated strongly with the parent. Posts can also derive from multiple parents, so one post could show up as a reply to more than one other post. So, in the other degenerate case, a WikiLog is equivalent to a weblog if everyone posts their input as a reply to a previous post, and no one ever edits anything.

Postulated, then, is a system which can act exactly like either a Wiki or a WebLog, depending on how you use it. Having established those endpoints, what gets really interesting is what happens in the middle.

The System

The way I envision this system working is somewhere between the major endpoints. Any user could create a "section", or parentless post that just confers the value of it's topicality, for sorting and categorizing purposes. There would be a "New" page, where anyone could post a brand new post, that only inherits from one or more topic posts. People could choose to read all the new posts, and rate them as they come in ("submission queue", but not segregated from the site).

Contributors can also spin off their responses as replies to a post. These are posts in their own right, they just inherit a "related" property from the parent. In this way, a familiar threaded discussion could take place.

Now you have to think about how to present this. The possibilities are almost endless, but here's a few ideas. You could have display modes that approximate a weblog page, with a top-level post at the top, and it's children below, threaded or nested, etc.

You could create a "Front Page" that winnows the best posts out of the current crop, by various metrics. I would probably consider average rating, number and quality of responses, and some time-decay. You could provide for posts to stay on the front page as long as they continue to get revised and gather new responses. You can also create similar front pages that derive from a particular "topic" post.

You could even create a "Front Page" for a content post. Taking the parent as a starting point, you could apply any of the normal metrics to all posts that derive from it, and treat those like "stories" in the display.

Since there is multiple inheritance, you could tag more than one top-level post, and find children of one or more of them, and make that a display. Imagine that you're interested in online communities, women, and sports. You could select "Show me everything that derives from the topic post "Internet" and/or the topic post "Community", and that has this story about female rugby teams as a related post.

I'm still clarifying this in my own head, so the above may or may not be clear. Questions welcome, so I know where to expand.

Upshot

Don't forget that anyone (provisionally) can edit any post. This ability includes the ability to link one post to another ("derives from", "related to", possibly even "contains") as well as the ability to revise and improve the content of a post. I see the result of all this being a web of interrelated knowledge, which, crucially, knows it's interrelated. Each post knows where it comes from, and what links back to it. This is an incredibly powerful thing, because you will end up with "clusters" of related information, which can intersect and overlap each other without losing coherence (for either cluster).

Machine intelligence could play in too, with the system assigning "strength" to links between data, and crediting more-traveled links with higher relevance. This would be throwing EverythingTwoAtSlashdot in there too, just for the hell of it. :-)

I think we're all groping toward the same concept. This may be a step further in that direction.

Practicalities:

Do I have time to do this? Hell no! But I may eventually, and I think it's worth starting a dialogue on the idea. What do you think?

Contributors: RustyFoster

High Volume

Given the high volume of different news events, on a larger site some sort of moderation or at least categorization is necessary to decide which new pages/notes/posts/stories to read each day; WebLogDigests. One way to look at a WebLogDigest is as a subset of RecentChanges (RatingGroups would be really neat). However, quicker than manually compiling a digest would be to use automated WikiVoting and have the server compile it (RatingGroups-moderated RecentChanges). Perhaps an even better solution would be to create an InformationDerivativeMarket that rewards those who correctly guess which stories others in their RatingGroup will find interesting (in this case, one would buy and trade "stocks" of various stories of the day, stocks which payoff depending on how many people liked them after reading them; see also GiftWiki).

This idea seems to be similar to ViewPoint.

It would also be cool to substitute voting for "i agree" "i disagree" posts within a WikiWebLog discussion (RatingAsContent); if there were automated WikiVoting, it would be easier to do this. I'd like to note that neither of these uses of voting implies the use of voting for community decisions or for moderating individual comments. -- BayleShanks

Comments

"What do you think?" +1, Insightful. (Oops, wrong site. ;-)

Wow. That's quite an essay there--it took a few readings to get all the ideas out of it.

I've been heading in a vaguely similar direction with ViewPoint. One of the better features of a WebLog is how easy it is to add your individual opinion--just hit reply and type. Wikis often get a bit more complex, since you need to make "editorial" decisions about how to insert your ideas. I've considered a "submission" idea to make this easier (one just submits one's text and other editors insert it), but submissions lack the immediate feedback of seeing your posting in a discussion thread.

Another possibility I considered was allowing wiki pages to contain a discussion thread (displayed below the main text). The main idea was that people could either edit the main page text directly (like wiki pages now), or reply in the discussion area (for back-and-forth or peripheral discussion). I'm not sure if it's worth the work to integrate the wiki and discussion-thread software, however.

A few weeks ago I had another idea for integrating wikis with weblogs--specifically integrating with KuroShin. (I was going to wait until after the 1.0 release of UseModWiki, but I guess now is an appropriate time.) Weblogs are very good at discussion, but not very good for long-term organization. Wikis can be very powerful for organization, but are difficult to use for immediate discussions (especially with a large number of people). The main idea is to add a wiki structure adjacent to the weblog for longer-term organization. The wiki should have some special markup to make it easier to link to stories, comments, and users. Optionally the weblog could add markup (like a <wiki> tag) for easy linking to the wiki.

One of the biggest benefits could be a usable long-term index of all the past stories. For instance, suppose I vaguely remember a K5 story about whether or not college is worthwhile. If the story was more than a month old, you probably can't find it in the individual sections. The K5 search utility is excellent, but most people aren't willing to do the extra "work" for a decent search. A wiki could be used to organize the old pages in hierarchical topics similar to the [Google Web Directory]. Additionally, people could make individual lists of stories, like "best of K5" lists.

Another possible use could be giving short summaries of long stories, and/or giving an overview of the comments with links to the individual stories. For instance, on a recent story the summary might look like:

: [I'm Not Dead Yet] Genetic research on fruit flies found a gene treatment which significantly extended their lifespan. This gene is also found in humans, leading to speculations about a possible treatment. Story discusses possible risks, and possibility that testing may take too long to be beneficial to people today. Notable comments include:

Discussion on genetic and other factors of aging (telomerases, free radicals): [1]
Testing methods and possible side-effects: [2]
Multiple comments on attitudes toward death, including a thread at [3].
Potential changes to population [4] and retirement [5].

There are lots of possible uses of such a wiki. I'm not certain how well the fully-open wiki idea would scale, but it would make an interesting experiment. Now we just need someone to bell the cat... --CliffordAdams

Each contribution of a Wiki can be seen as a Wiki:AddressedAndSignedMessage, where the lists of addressees and signers can be empty or nonempty. Editing, i.e. reusing a message is a new message. The involved participants can be human or not human. Thank you for keeping alive this meme. -- FridemarPache

Of course, that being said, I'm going to have to defend wikis with the Giant Flaming Clue Stick of Death.

: My interest is in online community-building. This, I think, sits right in the middle of wiki and weblog,

The software is the least interesting part of an OnlineCommunity. The community are the people. The software is their medium for exchange. Certainly the medium shapes the flow of the community, but it is not the community. Thus your implication that wikis or WebLogs aren't communities is somewhat striking. I think the community here, on WikiWiki, or HaskellWiki would disagree as well as the communities on KuroShin, SlashDot, or HalfEmpty?. Heck, ByteMagazine?'s comment section had a strong community last time I checked.

: Wikis are possibly even worse at encouraging community than weblogs, because they lack a basic discursive nature. People like to go back and forth on things, and there isn't a feeling of discussion on a wiki, but one of editing and revising what others have said. Yes, you can "discuss" on a wiki, but it goes against the grain of the system.

Wikis are extremely rich and complicated. Their stength and their weakness are the same: little limiting structure, and a lot of empowering structure.

I would seriously doubt the people on WhyClublet or [TrikiWiki] would agree with your assessment that discourse runs against the grain of a wiki. And the thousands of Wiki:ThreadMode examples bely this assertion. Look at the Wiki:ExtremeProgramming pages, or even better, the Wiki:DesignPatterns ring of pages to see the kinds of excitement that can bubble up from discussion. On the other hand, I agree wikis like CLiki without RecentChanges and good diff tools don't have a discussion feel. Also, small wikis like the defunct PolitizenWiki didn't have any discussion because there weren't many community members. If you look at the early days of MeatballWiki, this is true.

That being said, yes, wikis are worse than WebLogs at encouraging communities. First, they are user hostile, which doesn't help. Second, they are deeply bizarre. WebLogs look like newspapers, wikis don't look like anything (at first; the analogies are interesting). Third, most of the learning curve isn't explicit like the TextFormattingRules, but implicit like the CommunityExpectations; after all, most of the work of a wiki is done via CommunitySolutions. Wikis are too much work and too confusing.

Also, I think the collectivism of a wiki grates against the individualist North Americans that populate the internet. Let's not forget that wikis grew out of Smalltalk culture and consequently rely heavily on (pseudo-)Eastern philosophy. Indeed, until recently, one of the strongest metaphors on WikiWiki was Wiki:WikiMaster, alluding to Zen mastery. I think the SoapBox metaphor that WebLogs wrap around fit more into the deeper social roots of the internet than wikis. On the other hand, from what I can tell, the Japanese MoinMoins are doing really well.

: I think you've hit the nail on the head. Wikis are bad at creating individualistic community. I want more creative ways to connect people to wiki, without diluting the wikiness. Hence (for example) my interest in having WebLogEntry?s' names be in the same NameSpace as the rest of the wiki (see http://pikie.darktech.org/cgi/pikie?WebLog). -- JohnAbbe

: The basic gist is, marry a wiki's ability to create complex nests of knowledge with a weblog's encouragement of dialogue, and I think you'd have a very powerful space for community.

I agree with this. I always thought it would be superior to have a wiki backend for a web log to dump continuing discussion. But then again, since wikis are so fun, maybe it would kill the web log? ;)

JoiIto is doing something like this, with a linked [wiki] and [blog]. People seem to mostly ignore the link between the two, commenting in the section where the material is originally posted. OTOH, the software isn't as seamless as it could be, which could be a barrier for some people who might otherwise contribute.

: Ratings: Individual user assessments of value. Consider Scoop-style 1-5 rating as a starting point, but this go elsewhere.

You don't rate if you can edit. If you don't like something, fix it. You can't have both. It makes no sense. Like what would you be rating?

There are different levels of involvement with a post: rate it, comment on it, or change it. Not every post is going to inspire the same degree of intensity. Also, people are generally more comfortable commenting than changing. Remember, the wiki approach is still pretty radical.

: What do you think?