Hypothetical

It is important not to confuse a precise user taxonomy with a real person. Real people are of great interest as raw data, but they are frequently useless and often detrimental to the design process. A fine wine helps a successful dinner; raw Cabernet Sauvignon grapes tiny, tough-skinned, and seed-filled would ruin it. Many scientists, with a reverence for the empirical, confuse real users with imaginary but more valuable design personas.

The other major problem with real users is that, being real, they have funny quirks and behavioral anomalies that interfere with the design process. These idiosyncrasies are not extensible across a population. Just because one user has a distaste for direct manipulation doesn't mean that all or even a plurality of users do. The same works in reverse, too. Our real user might be fully capable of getting over some cognitive bump in the interaction road, whereas the majority of other users cannot. The temptation to attribute such capabilities to all users because one very real human exhibits them is strong but must be avoided.

In particular, we see this from company presidents. For example, one president we have worked with hates typing and wants to do all of his work without a keyboard. He has issued a directive that all of his company's software will be controlled only from the mouse. It is reasonable to want to use just the mouse to control the software, but it is not reasonable to shut out all those users who are more comfortable with the keyboard. The president is not a very representative persona.