Friday, February 25, 2005

Brian: Avoid Victoria buses

A 59-year-old Victoria resident is facing as many as four separate assault charges after he was arrested trying to steal a lock of hair from a bus passenger last week.

Tuesday, February 22, 2005

More meaningless usability percentages from Nielsen

I am impressed with how Jakob Nielsen has made a career in usability from this simple formula:

  • Come up with some usability guidelines
  • Measure some stuff against those usability guidelines
  • Report on the results, using meaningless compliance measures to make them appear authoritative ("The world's websites will be 72.4% compliant by June 17, 2015!")
  • Revise guidelines
  • Repeat

There is now a new twist: evaluating guidelines from the past as grist for more meaningless numbers. He looks at 19 year-old Air Force guidelines of the mainframe era to see which are still relevant today, and comes up with the 70% as still being relevant. The examples he picks as being no longer relevant puzzle me, which is why I find his "statistics" to be little better than if he just used a random number generator. Can he be serious?

  • "Guideline 4.2.6 said to provide a unique identification for each display in a consistent location at the top of the display frame." If this is no longer relevant, then what's with all the title bars in desktop UI's and banners in websites, and tabs in Firefox, and taskbars in Windows and the Mac OS X dock? Since 1986, the "unique identification" has grown to include icons, and the title bar text for a web browser includes the title of the page being browsed. Nielsen says, "Today, screen identifiers would clutter the screens with irrelevant information. They would not help modern users, who move freely among numerous locations." Today, screens are bigger, not smaller. Title bars are clutter? If you are moving among more numerous screens, don't they need labels even more? Has he ever tried writing a manual or providing support for a software product? It's a lot easier when things you have to write about or support have names.
  • "Guideline said to assign a single function key to any continuously available feature." While Nielsen is correct to say that "function keys are no longer the primary way of operating computers" that doesn't mean that function keys should be abandoned; good desktop UI's assign a range of control key, alt key, and even function key shortcuts to available features. And certainly these should be single key assignments that don't change over time. His main argument against this guideline, "Modern systems attempt to be modeless, so many features have become ubiquitous and accessible from anywhere," doesn't seem relevant to whether the features need key assignments. For that matter, I'm not even sure if it's true that many features are accessible from anywhere.

Overall, he is making the good point that human performance doesn't change much over time, especially compared with technology. But the pseudostatistical nonsense undermines the credibility of the whole profession, as does his odd choice of examples.