Skip to main content
Tufte QuoteAs a grad student, I had always heard about Edward Tufte, an authority on information visualization (‘infovis’) whose teachings had became foundational in the field.  The New York Times calls him the “Leonardo da Vinci of data.” You can imagine my excitement (as an information architect here at Acsys) when I heard about- and attended- his one-day course in Boston last week.The Information IS the InterfaceGreat design is about the relationship between the user and content. It is not about visual embellishments or ‘over produced styling’, but about helping users process and experience content. Think of the Apple iPhone UI and all the elements that went away: the scrollbar, the cursor, the start button and more. Removing these elements created a closer interaction between user and content by eliminating middle steps (moving a mouse, tracing the cursor to the target, etc). To achieve this goal, the designer must be endlessly self-effacing. Superfluous elements should be reduced to the point where the information is the interface. A good metaphor is a map: in maps, there are no boxes around city names. The only elements included are those that denote geographical locations and spatial relationships. But does that mean the less content, the better?  No.   Think of every element of a visual display as either signal or noise: it’s signal if it is meaningful, it’s noise if does not carry meaning.  The goal is to maximize the signal-to-noise ratio by reducing noise.Compare the two graphs below (figures 3 and 4) showing the exact same information. Which is better?Starbucks(That’s right folks, go for the brew.) The only difference between the graphs is that one is in 2D while the other is in 3D.  Notice that in both graphs the colors correspond to a size of a drink; therefore, color is a signal. But if the colors were only there for aesthetic purposes (and did not correspond to a drink size) Tufte would consider that noise. The 3D layer is thus an excellent example of noise, because it introduces additional visual elements which do not convey meaning: shading, height variations, slanted positioning, etc.  In addition to introducing noise, the 3D layer makes the bars harder to compare with one another, and to map onto specific calorie numbers on the Y axis. Conclusion: the 2D graph is better, because it maximizes the signal-to-noise ratio. Thus, you determine if an element is signal or noise by virtue of its relationship with the content. You cannot determine it from its presence or absence alone; a 3D layer could introduce signal if it helps to convey content. Notice how the 3D graph on figure 3 helps to convey spatial orientation.The “information is the interface” ideal speaks to the user-centered design methodology, since the reason people visit a website or app is ultimately to experience the content. I see great value in keeping it in mind when creating visual displays, since it helps the designer be self-effacing and put the needs of the user first. Yet I also wonder if the line between noise and signal isn’t much fuzzier than one might expect, when it comes to interface design. Improved aesthetics cause people to perceive an interface as not just more beautiful, but easier to use as well, even if it is not. So, ‘noise’ sometimes can improve people’s cognitive processing of content. When it does, shouldn’t it be considered signal rather than noise? For all we care, isn’t users’ perception what counts? Something to think about....