Rediscovering GOMS

Tantek Çelik offers three hypotheses of human interface design and asks for prior art. Here are the three hypotheses.

These claims relating the interactions of various aspects of an interface to cognitive load and ultimately to usability are not new. This is essentially what GOMS (Goals, Operators, Methods, and Selection) analyses do.

GOMS dates back to the early 80s. It is a means to assess human-computer interaction in terms of fundamental steps. Pulling part of the Wikipedia entry for one flavor of GOMS, we see that a GOMS analysis

…eliminates the goals, methods, and selection rules, leaving only primitive operators. Only six operations are provided by the base theory: 1) pressing a key, 2) moving the pointing device to a specific location, 3) pointer drag movements, 4) mental preparation, 5) moving hands to appropriate locations, and 6) waiting for the computer to execute a command. The times for each of the six operations have been empirically determined. The operations for a compete task are arranged into a serial stream, and total task execution time is a simple calculation.

Implicit in GOMS is that more operations increases cognitive load, so I’d argue that Tantek’s first two hypotheses are neatly captured within a larger GOMS framework. One thing I like about Tantek’s second claim is that he’s highlighting latency as a key factor. It’s one of the main drivers behind AJAX, so it’s important enough to pull out on its own.

Tantek’s third point is interesting in that he’s claiming a geometric relationship between cognitive load and usability. (That a relationship exists at all is a fundamental assumption of GOMS.) I’m not sure the shape of the curve matters that much, but the problem I have with a geometric relationship is that it implies that UI efficiencies that reduce moderate amounts of cognitive load aren’t going to yield big changes in usability. I think Twitter is actually a counterexample to his claim. The kinds of interfaces Twitter replaces aren’t things I would place at the high end of the cognitive load spectrum. Yet Twitter’s appeal is in large part due to its stripped down interface.

GOMS doesn’t have near the visibility of Fitt’s Law, so it’s not surprising to see the current generation of designers “rediscovering” it. I think any time designers try to codify design knowledge as hypotheses, laws, design patterns, or design principles, the field wins, so I’m glad to see Tantek raising these ideas.

It’s worth noting that GOMS analyses may be useful, but are certainly not a complete approach to UI analysis and design. (It they were, GOMS would have the visiblity of Fitt’s Law.) GOMS is pretty low level and doesn’t always consider the broader task in which the user is engaged. (Frankly, in grad school we pretty much blew off GOMS analyses precisely because they felt too low level. There’s no point investing in a formal GOMS analysis to a first-stage prototype when the kinds of changes you’re likely considering are wholesale changes based on larger questions like whether the user even understands what to do in the interface.)

Finally, since Tantek mentions Fitt’s Law, let’s throw Hick’s Law into the mix too, since it covers some of the same usability territory.

Hick’s law, or the Hick-Hyman law, is a human-computer interaction model that describes the time it takes for a user to make a decision as a function of the possible choices he or she has.

This is another case where Twitter shines (and Twitterific really shines): it’s not just that Twitter has a simpler interface, but that it really only gives you one choice as to what to do. Where Fitt’s Law is useful for graphic UI design, Hick’s Law really helps at the conceptual design stage.

posted March 13, 2007 by eric

Add comment:

Remember personal info?

Comment:Emoticons /
Comment spam is an unfortunate problem on our site. Making you answer an Obvious Question is one way to defeat the spambots.
Obvious Question: How many states are in the United States (50)?

Small print: All html tags except <b> and <i> will be removed from your comment. You can make links by just typing the url or mail-address.