Know what the principle difference is between you and your caveman ancestors of thousands of years ago?

Education. that's it. Put him in your cot as a baby and he could be doing your job right now. Put you on his pile of leaves and you'd grunt and look every bit as serious as he did. Although he had the same intelligence, there would be little point in trying to teach him calculus, because he has no concept of any of the required contextual knowledge. This kind of knowledge is built layer upon layer and we take it for granted.

Think about what we need to know just to be able to sit in front of a computer and decide to learn about it. We have a model of the world in our minds that allows us to infer new information based on the knowledge we already have. You may never have used, or even seen a computer before, but I'll bet that you can figure out how to turn it on. And that's not as important as knowing that you have to. You can see that it conforms to the pattern you have in your head called "electrical devices" - it has a power lead and a plug, that means it will probably have a power switch and will do something with the energy it consumes. If it has a device which you can identify as a screen, you'll be looking there for clues as to it's operation...and so on. Regardless of how inexperienced you think you are, you are actually an expert in the modern world and how to get around in it by using your previous experience - context. If you have a mental model for the genus "computers" then you will be looking for clues as to its species - Mac or PC? Server or workstation? And you will have different expectations and past experiences associated with each of these to guide your interaction.

It is for this reason that I get so annoyed by the term "Knowledge Management". Computers don't manage knowledge, since there is an infinite variety of context among users. Everyone interprets information in different ways based upon their own experiences. We even place different associations on the language we use. You can demonstrate this easily enough - ask six colleagues to list their associations for a few different words and phrases. Even if you keep it simple, you'll be hard pressed to find two which are alike. I have tried this test on rooms full of executives and I've never had an exact match. This is a well known phenomenon amongst negotiators, who are trained to be wary of the connotations others may attach to their words. Even the US administration conceded to this effect when they renamed their Middle Eastern activities "Infinite Justice" to "Enduring Freedom". It seems that to the Arab world, the former implies aggression, while the latter is more defensive. To make matters worse, these associations are dynamic, they are altered by time and events. What do you associate with the term "Desert Storm"? Chances are that your associations now are very different from what they would have been before 1991. Bear all this in mind when you hear of Knowledge Management - Computers manage information while knowledge is managed by people.

Good functional designs build on our existing knowledge base, fitting into one of our existing contextual models, allowing us to "shortcut" to what we want to do, rather than spend too much time learning a new model. These days, most of us (in the developed World anyway) have a mental model for computers based upon our previous experience with them. Anyone who has spent the briefest of time at a computer will be familiar with the concept of a graphical interface and a mouse (and how the two interact). More and more specialisation means more complex functionality - but that doesn't mean leaving the comfort zones of our existing contextual knowledge behind.

The reason that we have graphical user interfaces (GUI) at all is that they fit our mental models of how we interact with the world. With the invention of the GUI Files became "objects" that we could think of as occupying a space - even if the space was just an illusion. The direct manipulation of objects is something that we have been doing since the birth of man, it's not a skill which has to be learned, just adapted a little. The illusion became even more realistic when Doug Englebart invented the mouse in 1968 (yes it really was that long ago). Doug invented the concept of direct manipulation and as a result the illusion of workspaces that seemed real would soon dominate the computer interface.

Direct manipulation in a graphical environment may have become a standard, but what about the information itself? The problem for systems designers is how to cope with delivering information that is likely to mean different things to different people. Dynamic systems which generate content "on the fly" according to a user profile can present different navigation, branding and even content to each user, but even this won't avoid all of the pitfalls caused by the differing interpretations of users. How many times has your mouse pointer hovered over a web page while you tried to figure out which link takes you where you want to go? This situation arises, because there is almost always one person involved in the design who will make decisions based on their own interpretations, but the opinion of an individual makes for a staggeringly bad brief. Only slightly better is the "user survey" where a group of users are polled for their ideas. The problem with this approach is that users rarely behave how they say they do. It is far better to simply observe users performing a series of tasks in a consistent environment. This approach can generate useful data based on the facts of user behaviour and avoids the subjective opinion which hampers so many interfaces.

Information management and putting green practice have an important point of commonality - you can "play the feedback". If your first putt is a little short, you compensate with your next shot. Monitoring the behaviour of users allows systems designers to play the feedback too, by observing which elements users misinterpret. One of the best indicators is how they navigate and how long they spend reading pages. There's a good chance that if someone spends only a few seconds on a page full of text, it wasn't what they wanted to see. Beware of tripping over your own feet though - the results too, are open to interpretation !

We are in danger now of building too much complexity into the interface. I remember watching Doug Englebart mistakenly clicking the right mouse button instead of the left and remarking "whoever thought of putting two buttons on a mouse!" (his original design had only one) My current model has five and a wheel. The problem is that hardware, unlike software can't easily change it's configuration on request…at least not just yet. Technologists at the Massachusetts Institute of Technology (among others) are working on "soft hardware" which can adapt to a user's requirements. One such marvel is a touch sensitive input device on a display similar to a laptop screen. It can become a keyboard, drawing pad or even a second display at the user's behest - and the user chooses (or even designs) the layout.

Our cavemen ancestors made their own tools for a good reason - they didn't have anyone else to do it for them. But the complexity of modern interfaces necessitates that they are built by specialists and it's the differences in knowledge and experience (context) between the two groups that causes problems. Users are no less skilled than designers, the two simply misinterpret each other's actions.

Good design, whether for the web, or for a VCR means anticipating which mental model your interface fits and conforming to the logical expectations of your users. It's no accident that control button layout on a VCR, audio tape recorder or graphical MP3 player is the same - users expect it and they know how to react to it. If you've invented the World's first time machine, this layout would make a good platform to build upon. Everyone would know how to fast-forward to the future, rewind to the past, or stop. Come to think of it, there are a few events I must go and play back a few times...

The User illusion was first Published in the Journal for Work Process Improvement.

© Ian Richardson 2008.