Miller's Magic Seven
October 31, 2004 at 12:10 PM
Digital Web Magazine contributing author, Bryan Eisenberg, has published an article entitled Debunking Miller's Magic 7. The article, of course, goes in to details about disproving the theory that George A. Miller expressed in his 1956 research paper. Both Keith and I have written about magic numbers and the rule of seven before. I think it's long since been determined that Miller's theory is only best used with a grain of salt. It's a good metric to try to stick within the ballpark of, but certainly not law. It's a Breadth vs. Depth issue (thanks Henrik) that can only help narrow the margin for err. Ultimately the best way to go about finding the right metric to us is to practice Informed Design.
I'm constantly perplexed by the need of usability practitioners to "debunk" Miller's paper. What needs debunking is all of the myths that have sprouted up about this paper as it's true meaning and significance have been lost in the mysts of time. I sometimes wonder if people who propogate these myths have even bothered to read the paper. For example: "Another common Web design rule is the Rule of Seven. In the 1950's, George A. Miller put forth a general rule of thumb that the span of immediate memory is around seven +/- 2 items. Many Web designers still follow this rule today. Again, there is not usability data to back this rule up when it comes to the Web and how people use it. This rule could fit, or it might not - it depends on the goals of the site and the sites users." Miller did not present 7 +/- 2 as a "rule of thumb". It was presented in the context of "limits on our capacity to process information". Miller has pointed out that 7 +/- 2, as it's applicable to the span of immediate memory, is based on unidimensional data and nonsense, single syllable words, neither of which has anything to do with our ability to read or comprehend text. Further, guidelines based on the span of immediate memory (or, as it's now known, short-term or working memory) are not applicable to web design. Unless, of course, the designer is showing the list of items, and then making it disappear after one or two seconds. So, in point of fact, it's not even a good metric "to stick with in the ballpark". Practitioners who understand Miller's paper understand that if there is a design guideline to be drawn from the paper, it's that computer interfaces and documents should be designed to accomodate the limits of working memory. This means understanding the expertise and prior knowledge of the users, making apropriate use of "human factors" to faciliate visual scanning, and not going out of your way to overload the users' cognitive processes.