What's so bad about orphaned wiki pages?
This question came into sharp relief in a conversation today with Jerry Michalski. The exchange was triggered by a harmless (I thought) edit I made to a wiki for a private community to which I belong. I added this
to the wiki's homepage. Within an hour, Jerry emailed me to express concern. I was surprised and delighted. I always learn from exchanges with Jerry, and this was to be no exception.
Jerry felt the button encouraged community members to create orphaned pages, pages not linked to from any other page.
Jerry's right, but what's so bad about that? Jerry dislikes orphans because they don't play nicely with the broader corpus of knowledge. Jerry likes to see a wiki grow contiguously. He likes his wikis to creep, not hop.
For some activities (e.g., documenting a unified process), I agree with Jerry on the value of contiguity. But wikis can be used for all sorts of activities where hopping makes more sense than creeping. If a community is using a wiki to post questions, offer suggestions, file trip reports, or share random ideas, it seems artificial to ask users to link into a structure. The contributions are disconnected, so the wiki pages should be as well.
So how do we avoid the informational lawlessness that comes from a bunch of unlinked pages? Answer: tag them. An orphan with tags doesn't scare me at all.
Michael - great post, and something I'd been thinking about recently as well. When I'd first started poking at "enterprise wikis" about 4-5 years ago, I received quite a bit of grief from my fellow content management analysts at the "user hostile" modes of interaction with wikis. Creating pages by creating a link to a non-existent page? Completely bizarre to most people. It's like going to the start menu to shutdown... whoops! ;) Not to mention CamelCase, lack of WYSIWYG at all (at the time), essentially no security/permission-model, etc..
You bring up an interesting point though - so how, for example, do allow people to create their own departmental or project-based workspace? Make it as easy possible to set themselves up, and have at it! If it takes a "new page" button, so be it, even if the same functionality is buried in there (non-obvious to most) already. "Purity" is over-rated!
This is one reason why, IMHO, blogs are much more understandable to people, as an enterprise tool. Dirt simple to create a new post, or respond/comment to an existing post. Not so obvious (quite) when doing a trackback between blogs, but slightly more obvious than creating a new page via [thisIsANewPage].
But on the wiki front, obviously much has changed since the early days of "enterprise wikis" and with "gardening tools" that help to identify orphaned pages, and as you say, tags, that help to link to higher-level concepts, it's not as problematic as one might think.
Bottom line: even though "hypertext" as a concept has been around, via Vannevar Bush and his 1945 "As We May Think" essay, we still need to learn to live and fully think in a hypertext world. So, until then, we need to hack the system, as we always will...
Great talking to you again today. Always useful to bounce around ideas with a KM and wiki practitioner!
Posted by: Dan Keldsen | January 02, 2008 at 11:35 AM
As I read this I began to wonder about google-like search functionality within the context of the wiki. Of course there is the straight word search, but what about assessing links to a page and other algorithms. I do agree that tagging is preferable to hierarchical structure, but I find in practice that it is very easy (and common) to forget to add the tag. I think there is still work to do to make pages findable.
Posted by: Lee White | January 07, 2008 at 10:05 AM
Lee and Dan,
You both make great points about usability and findability. The "New Page" button is great at inviting first-timers to get into the wiki thing. (Dan, I think the point about blogs v. wikis is spot-on.) But Lee is right that this creates a findability problem. Tagging is yet another foreign behavior, especially to first-timers. It's not an established behavior for most people, and we in the wiki world are still working on the right way to integrate it into the user experience of our products.
My preferred solution, for now, is a "New Page" button that automatically places tags based on where in the wiki the button appears. Its applicability is limited to certain kinds of wikis and business use cases, but I expect our degrees of freedom to expand as wiki design and user familiarity develop.
Posted by: Michael Idinopulos | January 15, 2008 at 09:51 AM
Michael,
When I first set up a wiki which came with a "New Page" button on each page, I fielded a few emails about "What happened to my page?" The inquirer wanted their page to be accessible, to be linked to the "whole." I gardened accordingly. So when I saw your thoughts about letting orphans be orphans and tagging them, it made me think.
Like most everyone, I've spent a decade clicking links. In addition, structuring information is fascinating to me (how exactly does x relate to y?), so I questioned your idea of simply tagging an orphaned page.
Then I thought about Wikipedia and realized I had no idea how pages were/were not linked, nor was it relevant. In fact I could conceptualize Wikipedia as a huge body of orphaned pages in which a page was found by searching on keywords. The difference, of course, between Wikipedia and most wikis was that I know Wikipedia's subject matter: everything. I could search on anything and expect to find it. But I would be unsure of the parameters of most other wikis. I wanted breadcrumbs to lead me. Better yet, I wanted a table of contents with links on the home page so I knew what I might find inside.
Of course another version of a table of contents is a tag cloud.
So how about: an orphan with tags and a noticeable tag cloud isn't scary.
I was surprised to find when I looked just now that according to a "Pew Internet & American Life Project" survey of December 2006 (www.pewinternet.org/pdfs/PIP_Tagging.pdf), 28% of online Americans have tagged, though the percentage falls with age (32% of 18-29 year olds and 23% at 50-64). That was way more than I expected.
Your orphan tag idea (with tag cloud) now seems squarely on target.
I'm less certain about the automatic tag suggestion in your comment. I have random thoughts; I associate: the new page I add may be irrelevant to the content surrounding the "Add Page" button. Perhaps as I search through an unfamiliar wiki landscape I use the first "Add Page" button I happen on. Would the user be alerted: "So this page can be found by others, these tags will be added to it_______ or change them to________" Don't you think a page tagged "a" when it was about "z" could easily make people look askance at the wiki provider and taint the wiki overall?
(Of course, even tags or keywords we come up with ourselves are not necessarily the ones others might use. I just googled "structure of wikipedia." The second paragraph of the first hit was "Wikipedia's present power structure is a mix of anarchic, despotic, democratic, republican, meritocratic, plutocratic, and technocratic elements.")
Posted by: Julie Spriggs | February 26, 2008 at 11:02 PM
Great post Michael. We have always tried to avoid the use of orphaned pages in our online projects simply because they may or may not cause problems from an seo standpoint on public projects.
Posted by: barney party supplies | December 15, 2009 at 12:24 PM