Sunday November 5, 2006
Some Historical Usability Research
Why We Have the Guidelines We Do
By Colin Lieberman
At geek Valhalla last weekend, I picked up a collection of papers on usability from 1988, Design and Evaluation of Computer/Human Interfaces: Issues for Librarians and Information Scientists, edited by Martin A. Siegel.
One paper from this collection, Systems Interfaces Revisited, by W. David Penniman (then Libraries and Information Systems Director, AT&T Bell Labs), is particularly interesting — it discusses the underlying human needs in interface design, dramatically increasing the value of modern web usability guidelines by answering why modern studies reach the conclusions they do.
(As an aside, older papers like this are also interesting when they show their age: “CD-ROM is another potentially significant technology for information delivery.”)
From Babylon to the Present
Penniman starts with historical context:
In 490 B.C., the fastest way to send a message was through a human messenger running as fast and as far as he could... The data rate for that “system” was well under one word per minute... no really universal breakthrough came until the invention of telegraphy in the 1840s.
He points out that while cuneiform tablets could store one character per cubic inch, and modern technology can store 125 billion characters per cubic inch, humans are still only capable of processing about 300 words or symbols per minutes.
The mathematically inclined reader can compute: a Sumerian of 4000 B.C. could get through a cubic inch of data in one-fifth of a second, but us moderns require several centuries to process all the information that could be tightly packed onto one cubic inch of a hard drive platter.
Nothing groundbreaking here, but interesting for its own sake.
Penniman cites a 1959 checklist for the “ideal information service” by Harry Goodwin of Battelle Memorial Institute. The list, paraphrased by Penniman, is profoundly familiar.
- to get information desired
- at the time it is desired
- in briefest form
- in order of importance
- with auxiliary information
- and indications of reliability
- and authority of the information and its source
- to exert minimum effort
- to be screened from undesired or untimely information, and
- to know negative results are reliable.
Remembering that this is a list from 1959, let’s look at some of its points again, in the context of contemporary web usability best practices:
Users are looking to get information:
- in briefest form (we know users scan text, and guidelines advise us to facilitate this)
- in order of importance (good writing for the web dictates placing the meat of the matter first, and providing details for those interested further on)
- with auxiliary information (alternative presentations, such as graphs, charts, and audio or video content, as well as links for further reading)
This is from 1959. Clearly, the data Jakob Nielsen collects about how people use the web comes from something deeply ingrained in the way humans search for information. Modern usability guidelines are simply recapitulations of what librarians and behavioral scientists have known for decades, specifically tuned to the medium of the web.
So why do we forage for information, and why is it important that we follow usability guidelines?
Pennman cites a 1971 piece of research by McGuire and Stanley in the Journal of Nervous and Mental Disease about the conceptual models people use when approaching information storage and retrieval systems:
- When confronted with a problem, individuals draw on past models (model development).
- New models are built from old models as well as observation (consolidation models).
- Similar experiences among different individuals elicit different models due to the consolidation process.
- After “understanding” (model formation) is achieved, disconfirmation is very difficult.
- Well-developed models are not systematically checked once formed by the user. Assumptions are made early and last long.
As web professionals, there isn’t actually anything new here for us. We know that it’s bad form to deviate from de facto standards for exactly these reasons.
E.g.: Users expect a site search on the top of the page — they’ve developed a model of web sites where that is where a search box can be found. Placing our search box in a different location means that it won’t be seen, because users are not actively questioning their models.
Yet as we also know, and as the point about similar experiences among different users eliciting different models implies, not every standard is a standard among all users. Take the examples of navigation-dominant versus search-dominant users. We know that good design means that needs of both of these models are met.
What Does it Mean?
Penniman concludes that:
- Unique behavior and individual differences define the outer limits of the system and the degree to which it must adapt to a variety of user characteristics.
- Groupings of behaviors define shared models and will lead to the design of tailored components and command structures.
- Group behavior differences define the varieties of models that will be necessary to adequately serve all users.
The short version: apply guidelines and standards that have correctly identified trends among majorities and subgroups of web users. (But you already knew that.)
Back to our ancient Sumerian with his clay tablet. The fundamental problem with the web is that there is more information available than any person could be expected to process in a lifetime. If the sites you work on have more than a few thousand pages, your corner of the web alone is beyond what most folks will ever be able to perceive.
By following published usability and accessibility guidelines, and by understanding the psychology behind them, we are able to produce work that is findable, scannable, navigable, and usable. We are able to tackle this problem of too much information with good odds of success.
Foolish Consistency and Hobgoblins
More importantly however, is the value of enriching our understanding of the guidelines and studies we read. Many writers (with whom I don't really agree, but I see where they’re coming from) have identified an unpleasant trend in accessibility and usability - not seeing the forest for the trees and slavishly adhering to guidelines such that content suffers.
By recognizing the basic psychology behind the guidelines, we can make better decisions, without needing a pocket rule book, and we're better equipped to explain those decisions to our friends, bosses, and colleagues.