I was re-reading the Cluetrain Manifesto today. The following item hit a nerve.
"3. Conversations among human beings sound human. They are conducted in a human voice."
Maybe that's why user assistance sometimes increases frustration. Help isn't conversational.
Trying to teach people how computers communicate isn't very helpful. Sure, people can learn help conventions and adapt to our help interface. But should we expect them to work at it at exactly the moment when their frustration level is highest?
I don't learn well when I'm frustrated.
H-E-L-P is an emotionally-packed word.
Real human conversation is sensitive to frustration and other emotions. The voice of help should be that of the hostage negotiator or the crisis hotline expert. Help should say "Hey, you look really frustrated, but I think I can help. What's bothering you?"
Perhaps help should even let users vent a bit.
"Use the field below to blow off steam. Then, tell us what the problem is. We'll try to offer some advice, and adjust the brevity and directness of the response based on the number of offensive words in your query."
In human conversation, sometimes the other person simply doesn't have the answers we seek. But there's something about talking to a human being in an emotionally-expressive way that calms us down and puts us in a better state of mind for solving the problem.
Often, we can figure out the answer for ourselves, as soon as we calm down and think it through.
Help lacks that sympathetic voice. It doesn't know how to calm us down.
Related: Technical writing and consumer emotion.