In order to offset the increased monthly costs of my new phone, I recently decided to migrate one of my two dedicated servers to a VPS. After mulling over bargain basement offerings and some major ones, I decided on linode. I figured that most of the services I run on my weaker dedicated server (mail & webmail, DNS, etc) would live as happily on a $20/month VPS as they would on a $65/month dedicated server, and although I've had a long time to start the switch, I commenced last night (with 10 days remaining) by moving my wiki and my dev site.
I noticed that, although my wiki contains a few very often referenced pages on it, most of what is there suffers from pretty severe bitrot. I'm guessing that this phenomenon is well known: the low barrier to entry for a new wiki page is conducive to spur of the moment creations that can be either genius and well useful or completely unnecessary and instantly outmoded. I initially wrote it off as a product of me being the primary contributor to the wiki, and me generally being more interested in actually doing something than writing about it (although this blog is an emphatic statement against that premise).
But then I thought about the wiki we keep at work, and how most pages are so dilapidated that when something is updated it is rarely trusted as current or correct. I'm not so sure that this is a cultural issue; I suspect that in some ways the looseness of the wiki is itself a barrier to entry, and that without structure the problem of "how do I convey this information" can be a wall too high for many would-be contributors. I imagine that many mid-sized wikis suffer from this.
This low barrier to entry also allows for the quick creation of documentation for systems that are not used often. The thought is that the setup of this system required extensive research from the canonical documentation and perhaps newsgroups and mailing list sources, and it'd be best to explain the system's detail as it exists in deployment. This can lead to other problems. When, at some usually distant future date, the system eventually fails for some reason, the state of the deployed software is several generations behind the current documentation and chatter on the subject. The nature of software's rapid change usually leads the wiki document to describe a version so outdated as to be nearly alien to what is con temporarily available.
A few months ago, I made a joke in an IRC channel: "Wiki is a polynesian word for LIES." For the most part, if you google any term with wiki, you get a wikipedia page about the remainder of your search term. This makes it difficult to perform quick wet-finger-in-the-wind style "research" as to the general perceptions of wikis and their suitability for documentation. In my experience, the culture around a wiki has to have near rabid commitment and have a hive mentality in line with its leadership to stand a chance at being anything resembling proper documentation.