Thursday, 5 November 2009

Thoughts on wikis

Over the last few years I have, like everyone else, put more and more information into wikis.

I've used three main platforms: MediaWiki (Choral Public Domain Library,, PBwiki (e.g. and TracWiki.

Those who have known me for a while will be aware of my tendency to bang on about information as an organisational asset, how difficult it is to get people to share information at all, and the consequent need to provide information repositories with capture mechanisms that involve minimal time and effort to use.

Information is more or less useless if it's inaccurate, if it's unclear, if you don't belong to the audience for which it was written, or if you can't retrieve it in an effective and timely manner.

So I was, and remain, delighted with the concept of a wiki, which allows easy capture, sharing, review and correction of information, which can give it a structure that permits effective retrieval and avoids duplication, and which encourages accuracy and clarity.

However, the devil has turned out to be in the detail.

When people put information into written form, it may be well structured, or it may be a brain dump. In my experience, any structure comes from the author's head and is internal to the document under preparation, unless the individual's day-to-day work specifically involves structured documentation. Therefore, in my view, if we are to get information out of people's heads and readily usable by others, we have to start with the content, and accept that structure may need to be added later.

If you ask someone to document something for you, they will typically create a Word document. It may have diagrams and complex tables. There may be careful formatting aimed at making the information easier to understand.

All wiki platforms I have used appear to make the following assumptions:
1) Formatting, beyond a very basic level, doesn't matter
2) People are prepared to put time and effort into learning and using the platform and/or its WYSIWYG editor
3) All content fits cleanly into a predefined structure
4) Once the information is in the wiki, people don't need to get it out again
5) Users are IT literate and have a conceptual interest in the wiki as a platform

The first alarm bells started to ring for me when I realised that not all wiki platforms used the same markup language - surely, if anything needed an open standard, this does.

I then started to wonder (and I still do) why wiki markup languages existed at all except as a structure-imposing superset of normal HTML. Do all wiki developers really want to waste their time writing the world's umpteenth not-very-good WYSIWYG editor?

I tried and failed to rename wiki pages (come on, chaps, even the most basic HTML editing tools let you change page names and automatically update all the relevant internal links for you). Structure is surely not a once-and-for-all thing.

I searched in vain for ways of adding links that did not involve two browser windows plus cut and paste (the honourable exception is PBwiki).

I pasted Word information into wiki WYSIWYG editors, even Word information that had been saved as filtered HTML and then put on the clipboard from IE, and most of the formatting disappeared (except, again, on PBwiki).

I attempted to export a set of wiki pages from one wiki and put them into another, or even to generate a PDF with all the linked information in it ... Surely I'm not the only one who wants to use the wiki for capture, maintenance, and day-to-day information retrieval, but still to have the option of publishing the results in final form? And surely I'm not the only one who gets fed up with one wiki platform and decides to try another?

To sum up, to say that I am disappointed with what has happened to a fantastically simple and brilliant idea would be a major understatement.