There was a very interesting post over at Slashdot where a reader asked the gathered populace how to get users to read error messages.

The question, to some, might sound trite or like the standard complaining from technical support staff – and yet, this is a real problem and is wrestled with by technical staff and by usability engineers as well.

There were a number of interesting responses (and thoughts) to this question:

  • Force users to read it. In one case, the support staff put a code within the text of the error message, and required entry of the code to continue. Users who called saying they were stuck were told to read the error message to the support staff – at which time, the user typically said “Oh, never mind…” Alternately, another story told of a site that stopped people cold at an error message, requiring them to call technical support. If they tried to bypass the message with a reboot, the offending application would not work for 15 minutes.
  • Start using error messages appropriately. This requires more work by developers, but the idea is that users are so used to error messages that are meaningless that they ignore them entirely.
  • People will filter out things that are unexpected or don’t fit the model. This is backed up by research (apparently) of fighter pilots who, during a simulation, overwhelmingly blotted out the fact that there were huge items in their path on the runway. In another famous case, people were given something to watch (a conversation I think it was) and no one reported the huge gorilla (yes!) that wandered through the picture.
  • Pictures are more easily noted and remembered. Some suggested using pictures – one or two said they had done this with excellent results.

What do you think about error messages? How do you get users to read them?