Context vs. Consistency

For as long as I can remember, consistency of user interface has been advocated as the Holy Grail of good usability. People have the same menus, the same visual cues. Nobody panics; everything is familiar.

Only recently did I come to realize that context was more important for usability than consistency. A good example of context is Microsoft’s ribbon. It’s not exactly what I have in mind, but it’s half-way there. The ribbon groups related tasks together, and offers a somewhat intelligent context based on what you’re doing at the moment. It’s a good start.

A helpful, meaningful context needs to be derived from user’s past activity to intelligently predict what the user will come looking for next time. The key word here is “predict.” Amazon does it right. If you look at a book, they will plaster their home page with suggestions of what others bought and what may be related. The former is mining existing data, the latter—predicting based on the same data.

Another example is Salesforce. When you log in, you have a sidebar with shortcuts of past activities, which is helpful in its own right. But Salesforce doesn’t try to predict what activity I’m likely to do the next time I log in.

This brings me to machine learning. You can build a meaningful predictive model if you have the right tools. Unfortunately, machine learning is an academic field, with PhDs upon PhDs keeping it to themselves. Consider cryptography: what took mathematicians years to figure out is available to you in three lines of code. Not so with machine learning, unfortunately.

Oscar Wilde said, Consistency is the last refuge of the unimaginative. However, given the state of technology, it seems consistency is an inferior, yet easier alternative for the foreseeable future.