Clearing up misconceptions about interface design.
There are no magic numbers.
We use heuristics and guidelines to create good designs. Heuristics are imperfect by definition. And we sometimes get guidelines wrong. This blog takes a closer look at the curious case of the ‘magic number 7 +/-2’, and why some still apply it to restrict interface designs.
The problem statement.
Let me catch you up on the problem, in case you are not familiar with it already.
When designing an interface with multiple visual elements, we should limit the number of elements to 7+/-2. True or false?
A) True, because Miller’s law states that human short-term memory is limited to the magic number 7 +/-2.
B) True, to avoid excess, and poor visual organisation, which would lead to a poor user experience.
C) False, a more recent study found that Miller overestimated capacity, and we should aim for 3–5 elements (‘more is less’ heuristic).
D) False, because visual search is preferable to memory-based recall. Therefore, we should display as many elements as possible.
None of the above choices make sense. All are partially true, but none is an acceptable answer to the question. I am going to summarise very briefly what Miller’s law says, and why it doesn’t apply to interface design. Then, I try to reflect on why people still keep applying Miller’s law to design, despite the readily available and wide-spread intelligence against doing so.
Miller’s law and why it has nothing to do with designing interfaces.
Miller’s law states that the average person can hold 7 +/- 2 items in their working memory. ‘Holding items in one’s memory’ refers to a situation in which you have to memorise, and subsequently recall arbitrary information. Think symbols flashing up on a screen, tones that only differ in loudness, or someone telling you their phone number. The information is available for a moment, then it disappears.
Miller’s work was pure psychological research in the 1950s. Participants’ job was to encode fairly arbitrary items. This has nothing to do with the tasks we encounter in using websites: reading, searching for content, navigating menus, retrieving complex information. Your website is not a unidimensional stimulus that disappears after a few seconds. The viewer can come back to it again and again whenever they like. Sure, too many elements are overwhelming but that’s nothing to do with Miller’s law. That’s just clutter.
Then why do people still reference Miller’s law in interface design?
Type ‘Miller’s law design’ into the Medium search function for a demonstration of this diehard misapplication. Advocates of Miller’s law in design, have received between 3 and 48K claps, and 150 comments by readers. Most comments applaud the blogs, some calling it ‘eye-opening’ and ‘life changing’.
A few of the commentators challenge the authors. One conscientious reader Luke Madeira explains the misapplication, referencing the Nielsen Norman Group, an entire website dedicated to busting UX myths, and pundit Edward Tufte himself. His expertise perishes between the prevailing gratitude of misled readers who vow to apply the magic 7+-2 in their design work from now on.
In the problem statement, I offered four nonsensical answers to the multiple choice question. Some answers may have sounded reasonable enough bragging with terms like ‘short-term memory’. If it sounds smart enough, we may be swayed to believe its false authority. Plus, if we have heard of Miller’s law, or the magic 7 +/-2 before, we fall prey to thinking: I heard of it before, and it sounds legit, so it’s probably that.
It may be this sciolism [amateurish half-baked knowledge] that keeps the wrongful application of Miller’s law going and going like a merry-go-round. If it sounds familiar and somewhat trustworthy, it’s probably true? The limitations of the human visual system and our working memory imply that we can’t process a cluttered interface with too many elements. Memory capacity is relevant in Miller’s work, too. But in a different context. Applying Miller’s law to visual image search is a strange thing to do.
The ease with which you can memorise a phone number has nothing to do with how many elements you can handle on a website. The common feature may be ‘there are too many elements’. But the scenarios describe different tasks with different performance indicators. One is whether you can remember and recite a number, the other is reading, and interpreting meaning, or finding tabs in a menu.
There is a diversity of tasks afforded by websites. None of them resemble the experimental tasks that Miller tested. His study participants repeated arbitrary sets of digits, for example, or categorised tones that differ only in loudness. In web design, the performance indicator may be how quickly you can navigate the website. If you insist on having laws, Fitts’ law and Hick’s law may be relevant in this case. Fitts’ law describes that the time needed to navigate to a target, is a function of the distance between starting point and target, and the size of the target area. Hick’s law predicts that the time we need to make a decision increases logarithmically as a function of the number of possible choices offered.
Another aspect from Miller’s work that people transfer to visual design is chunking. It refers to the grouping of elements in memory tasks. We commonly chunk telephone numbers, or bank account numbers, into groups of 3 or 4 digits to reduce transcription errors. Chunking items to memorise them is not the same as cleverly arranging visual displays, just because limited cognitive capacities feature in both explanations. One is Miller, the other is Gestalt laws. Clutter makes for poor user experience, but there is no magic number. That would be both too easy, and weirdly prescriptive.
The state of knowledge dissemination, and critical thinking.
Blogging is a fast and easy way to share information. Formal reviews and editorial processes put a break on the dissemination of knowledge. But professionals publishing without peer-approval is a two edged-sword, clearly.
I know a lot of capable people who hesitate to blog. They are aware of all they don’t know, and they fear to expose themselves to a critical readership ready to poke holes in their work. Yet this is the beauty of fast and interactive publications. We need not only to share, but also to discuss the intelligence we think we have. And we need to be careful not to over confidently blog about ‘laws’ if we haven’t triple-checked that we understand them.
Don’t passively accept what I’m writing. Question and challenge me if you find something is wrong, only partially correct, or misleading. My colleague Camellia has written a blog post about critical reading, a vital skill that seems to be in decline. Of course everyone should do their homework before publishing an article. Yet, the reader has to critically assess who is writing, what their expertise is, whose interests they represent, and so on. It is our responsibility to read and learn enough, before we confidently assert that we understand a topic.
To avoid half-baked truth and wicked recommendations, Vizzuality works with psychology researchers at the University of Cambridge, UK. This blog post has been written by HCI researcher Melanie Herrmann, and internally reviewed by Dr. Will Skylark, and Vizzuality’s Lead Writer Camellia Williams.
