How do we create new and interesting pages on this wiki? And why do we do it in the first place? Can we guide this process?
Part of the answer lies in self-selection: All those community that do not manage to create new and interesting pages fade and disappear eventually. There’s a ruthless weeding out going on in the attention economy.
Another element is that it is easy for newcomers to join, it is easy for them to learn through observation, never having to speak up until they feel confident to do so. It is also easy for the existing contributors to integrate newcomers because we can watch what they are doing, and anybody can undo mistakes. We also use specific techniques to build a sense of community.
But there’s more, specific to wikis aiming for a pattern language: We start using page titles as replacements for whole ideas, allowing readers to “drill down” to the core of all the important concepts, building a LinkLanguage as we build the site.
And finally, the specific lessons learnt on particular wikis, mostly on CommunityWiki itself.
The original wiki was the PortlandPatternRepository. Its purpose was clear – at least initially: Allow people to collaborately extend a computer science PatternLanguage. But then people decided that process was important, too. And the pattern language got extended. And then they felt that people themselves were important. And Zen. And religion.
It was amazing. These days, the PortlandPatternRepository is the WikiWikiWeb, and its mission is an unsolved riddle. Other sites appeared, however, tackling various areas: Meatball was about people, people, computers, and people (MeatBall:MissionStatement), CommunityWiki is about similar stuff, EmacsWiki is about using and programming Emacs, etc. What is going on, here?
Sites appeared that scratched an itch: The original founder has an idea and starts a wiki. The original pages must be “interesting” enough for others to join. WikiFounding is a slow and laborious process. Many wikis probably die and nobody ever knows about them. The big WikiFarm””s don’t mention how many wikis never take off. Right now, Oddwiki has 395 wikis, 182 of them have not been edited in half a year – and wikis not edited in a year are being deleted automatically, so they’re no longer counted. Thus, the surviving wikis must be offering “something” – some kind of entertainment or help or knowledge that others find interesting. Interesting enough so that some start contributing.
In general terms, these wikis do “problem solving”. For there to be problem solving, there needs to be a way to add new problems, edit them, link them to related problems, a way to build trust in the solutions suggested, which also means building trust in the wiki contributors, which means learning to know them, which means learning about OffTopic issues. And slowly, a community forms.
The architecture of the wiki also encourages PeerReview and social process instead of strict workflows encoded in the software. See HowWikiWorks.
These observations give us the basic structure of most wikis:
A PatternLanguage is a set of named entities (”patterns”), each containing a problem description, and a “best practice” solution to the problem. It’s a very nice way of collecting and structuring useful, pragmatic information, without much formallism and without academic rigor.
The social dynamics on a wiki as described above, however, modify the result. As people come together and strive to create a page per pattern, collecting information, the realize that they also need pages for themselves, the need pages to define the prerequisites on their pattern pages, they need pages to discuss unintended consequences, they create pages for tangential information, and so the entire system starts to grow beyond a simple pattern language.
Some of the new pages without a pattern are still useful in other contexts, and thus the pagename can be used like a pattern: It has a catchy name, and it links to more information. The only difference is that the new page is not a pattern description. Thus, where as some pages on the wiki form a PatternLanguage, a larger superset of these pages forms a LinkLanguage.
We can derive some more consequences from this observation:
On this wiki, we have tried various experiments adding new features to the wiki software, trying to gauge the effect it will have on our own community. Here’s a selection of these experiments:
DenotingAuthor: This was largely a success. We used alternating indentation to denote different authors, we tried alternating font slant (normal vs. italic) to denote different authors, we tried so-called macros that would expand
-- [[AlexSchroeder]] or
[[AlexSchroeder]] 2006-03-13 20:03 UTC and similar sequences into signatures, but none of them was as popular as the use of real pictures (”portraits”) and an alternating background hue. As soon as we started to use those, all the other features fell into disuse.
Editable SideBar: This was largely a success. We use it for announcements, to list members, linking to related communities, and indicating TelePresence. More discussion to follow.
Blog within the wiki using DatePage””s: Largely a success. More discussion to follow.
Switchable CSS: Not used.
CommunityMaintainedCss: Success! Collective experimentation with new features happened on CommunityWiki, but as soon as the SideBar looked better, nothing else was done. On Oddwiki, however, the editable CSS has proven to be very popular.
LangueFrançaise: French Translations. Partially successful. Not sure how much reach they have in the French wikisphere; not sure how much this enabled things like the Oddwiki:FêteInternet/.
FrontPage: Indeterminate effect. We like it, but does anybody else?
MultilingualExperiment: Multilingual pages seemed to be quite popular for a while, but multilingual contributors mainly switched to English or left the community.
WhoIsWatchingUs: Failure. We liked it, but lost interested. We’re not using the data to create trails from page to page, creating new short-cuts and pointing out interesting links to go on to.
What can we learn?
A major problem for multilingual wikis is motivation of translators. The following is based on observation alone; no formal interviews were conducted.
On CommunityWiki a single individual, ChristopheDucamp, has been very interested in translating, because he is trying to bridge the gap between Community Wiki (predominantly English) and CRAO Wiki (French). He has also managed to make the French wikisphere/blogosphere a paying job, so getting new input from other languages is of immediate interest to him.
Similar explanations apply to the various translators on the Oddmuse site: They all felt at some time or another that they were helping their language community connect with the rest of the Internet, but dropped their translation efforts after a while. One possible explanation could be that the motivation of bridging communities only motivates as far as helping non-English speakers to get started. As soon as they have their own Oddmuse site with the basics in place, no further translation is necessary, as many hackers seem to believe that technical aptitude and knowledge of English go hand in hand. Thus, non-English translations only have to cover the basics for the non-technically minded users.
The MultilingualExperiment on Community Wiki seems to indicate that some people are interested in novel approaches to multilingual sites (eg. interlacing various languages on a single page), but the sub-community was not motivated enough to start a large-scale translation effort.
Given my doubt in an inherent motivation of translators, an entirely different approach for multilingual sites seems more appropriate. Instead of defining timely translations as goals, here’s a different list of success criteria:
If the wiki is supposed to be maintained by volunteers, it must take motivation into account – it’s a top priority.
There are various ways of tweaking the social system of a wiki – these are the “tools” of the information architect trying to influence the wiki process:
Each of these choices is called a “convention”. The sum total of these conventions is called a WikiProcess. A WikiProcess is roughly analogous to a structure of governance for a wiki, but not quite; a WikiProcess encompasses style guidelines and CommunityExpectations, whereas the structure of a constitutional government is restricted to DecisionMakingProcesses and absolute laws.
Every wiki has its own WikiProcess, but one can still roughly categorize them. Most wikis today work off a consensus-based model called TheWikiWay.
While this page was hosted on CommunityWiki, LionKimbro noted that he used the CommunityMaintainedCss all the time. He loved it and was disappointed when it was swiped out. The context was important to figure things out.
He also claimed that people use and link to the FrontPage.
And he frequently checks WhoIsWatchingUs.
MattisManzel also felt that it was I who had turned down the multilingual experiment. He also linked to NoLeader when he said it.