Sam Marshall has an interesting post about people's preferences for seeking information.

People are lazy. Rooting through a paper filing cabinet is much harder than asking Nellie (as the archetypical colleague). Searching Google, now it exists, may be easier than asking Nellie. Older generations may have an ingrained expectation that databases rarely have the right answer; teenagers seem to assume Google can answer everything. If you wanted to know film times at the local cinema, would you ask Nellie first or use the net?

He also quotes Dennis Pearce from Lexmark who said:

[In 1977, Tom] Allen also found that the engineers he surveyed got their knowledge from the sources that were easiest and most familiar to them, not the ones that were most reliable or accurate, even when they themselves were aware of this. In other words, they knowingly sacrificed accuracy for expediency.

Both Sam and Dennis are right on the money, but an additional aspect to consider is what type of information is being sought. Easily identifiable data such as cinema listings are more easily retrieved from the net than friends or colleagues, who we realise will, like us, be unable to provide accurate and up-to-date information. Fuzzier information which may require some additional explanation of context is harder to find online, e.g. user queries about software are often best dealt with by someone who knows the software and can talk us through our problem. Online sources frequently let us down because they can't refine advice for individual situations in the way another human being can.

This latter sort of enquiry I have seen made by friends via instant messenger or IRC before they have even looked online. The assumption they are making is that if they ask friends, they will be able to specify more clearly what the problem is and hopefully therefore find a more pertinent answer than if they searched the help pages on an official website. There's also the desire to sidestep Google, which can return a huge number of useless results when searched for generic or common terms. If you don't have the information required to refine your search, you can waste a lot of time ploughing through hundreds of results and, knowing that this is likely to happen, it's an obvious solution to skip Google and go for an easier route - asking friends.

The ideal knowledge sharing application will deal easily and quickly with different sorts of information and allow for both retrieval of discrete data and consultation of friends/colleagues for the exchange of fuzzy data. This latter type of knowledge sharing, particularly when someone knows that they need to know something but doesn't know quite what they need to know, is why KM software needs to be social, rather than just a document management system.

Social tools, particularly blogs, allow people to build an online presence, 'advertise' their areas of expertise and make it easier both for others to tap them for information, fuzzy or discrete, and for them to ask for help themselves. Often pertinent information comes out of extended conversation, rather than a single, simple enquiry, and allowing for this sort of collaboration is core to successful knowledge sharing applications.