Wikipedia Death

19 June 2006

A recent blogosphere controversy was caused by Nicholas Carr's entry claiming the "death of wikipedia" (yes I know my response is slow, but I didn't have the time to write while on the road). His initial post struck me as rather odd, saying that wikipedia was dying because 0.01% of articles had a rather mild protection. It's like saying democracy is over when a town hires a policeman.

His follow-on article helped explain where he was coming from. Carr's view was that wikipedia had a myth of complete openness that was never really true, and that we need to get beyond this myth to really discuss its impact.

I'm finding this discussion interesting (and a tad amusing I'll admit) because, like many of my readers, I've been familiar with wikis for a long time. They've come out of the software development community. Many teams that I run into, inside and outside Thoughtworks, use them to communicate and build up collective notes.

I remember many years ago when Ward came up to me and invited me to take part in his experiment of a collaborative space to discuss patterns - the original wiki. I was an active reader during the late 90's as Ward's wiki became the central point for articulating Extreme Programming.

Ward's wiki was, and I gather still is, a completely open wiki. Anyone edited any page at any time. There wasn't even logging and registering - the only trace you left was your IP address. The early wiki had no version control - which really did make it a whiteboard. Nobody took on maintaining or editing anything. Ward told me he did the occasional spot of 'gardening', but only with his famously light touch.

The result was a complete mish-mash. There were some good pages summarizing stuff, most pages had long discussions, which often went nowhere. Still the result was way more interesting, and useful than I thought it would be when Ward first explained the concept to me.If nothing else it introduced Extreme Programming to many early adopters who took it beyond its initial roots.

So in contrast to Carr's wikipedia myth, my interest in wikipedia lies in how much structure and control it has in contrast to Ward's wiki. I see pages separated into article and discussion, version control, user registration, editors with watch lists, arbitration committees - it's all very different to Ward's 486.

Yet this structure has worked. Wikipedia is a far more useful resource than Ward's wiki. Keeping the discussion separate to the article page seems to work really well. Editors care about keeping coherent articles. All in all the wikipedia has proved a useful resource - comparing it to things I know about it's reasonably accurate.

But despite it's closedness compared the Oregonian original, wikipedia is still distinctively different - even open - compared to much else that's come before it. Anyone can edit 99.99% of the pages - but doesn't mean edits stick. I'm sure anyone can easily fix a typo on pretty much any page. But make a substantial change to a page and an editor will be asking you why. The interesting thing is that this is all after the fact control. There's no permission you have to get to change a page before you begin, instead you have to justify what you've done afterward. That's a big shift to many people, particularly in the corporate world.

Yet of course we as a society live off after the fact controls. There's little that really stops me driving down the wrong side of I93 in the morning, yet there would be consequences should I do such a thing. Certainly there are places where it's necessary to prevent bad things from being possible, but in many cases it's more efficient to be much looser but know there's a consequence to any breach. We try to follow that principle at Thoughtworks, leaving people with lots more latitude than in most organizations - but then following up with those who abuse the system.

Much of the agile software movement is about loosening controls, shifting more to asking forgiveness rather than permission. Yet loosening controls isn't the same as anarchy and no control - a misrepresentation that's commonly thrown at agilists too. It's about asking how we can use a minimum of controls, so that that we don't suffocate the good in our desire to protect ourselves from the bad.