Documentation is incredibly important, however many people seem to only regard it as backup in case someone leaves the company. That's only the tip of the iceberg. Good documentation can save vast amounts of time (a short note about an error could save hours of headscratching when the same problem strikes again). We've all said at one point "oh, I've seen it before, I just can't remember what the solution was".

Documentation does need maintaining (updating, removing duplicate or obsolete information). The most important information to pay attention to are procedures. Tried and trusted procedures prevent mistakes - particularly in crisis situations where people may not be thinking clearly. They also allow you to spread knowledge around - instead of having one expert in an area, more people can become familiar with a system or product. This is really experience written down - such important information shouldn't be left in people's heads!

Finally documentation saves time - including the author's. I generally think of things in this order (worst first):

• do the job myself
• tell someone how to do it
  • this takes longer, but if it's repeated it saves effort in the long run
• write a document on how to do it and send it to someone
• send them a link to a pre-written document
• do nothing. This happens if someone else has found the documentation and done it themselves

Most pieces of work fall into the latter two categories now unless they are new pieces of work.

Back to the subject of this entry - how to solve documentation (since "everyone hates writing it"). Firstly document as you go along - cutting and pasting commands can be a little tedious, however it is far faster than trying to recall what you did after the event. More importantly it is also less error prone (particularly with regard to omissions) and if you capture error messages in full, they are now searchable. However the biggest advantage of documenting as you go along is that it is done. Too often documentation is never written as people move onto "more important" or "more interesting" projects.

Lastly, wikis (or some other system with very simple editing) are essential in my opinion. Some of our most valuable wiki edits have been minor updates done by a subsequent user "if you see error 12345, you need to do this". That two minute edit can subsequently save hours of time.