Jef Raskin on Cooper
Design isn't always "self-evident"

Jef Raskin wrote:

When crashing around the Web, I ran into the 22 Feb 2000 interview with Alan Cooper. Nodding my head in agreement as I went along, I came to an abrupt stop when this icon of design said, "We believe that good design is self-evident."

If you believe that, then you are stuck in a rut, because the value of deep improvements are rarely self-evident, and even when a better design -- if unfamiliar -- is shown to developers or experienced users, they tend to reject it. Often, it takes careful explanation and having them gain experience with it before the improvement is understood. Those of us who have made major improvements in the way people use computers know that this dictum if false. If you are making small changes, however, it can be a useful rule of thumb.

Later, Cooper says that "Technology had firm limits dictated by physics." and says that by contrast, "In the Information Age, in the Digital Age, the limitations are not imposed by the fundamental characteristics or the physics of our devices."

He leaves out that what we are limited by is the fundamental characteristics of humans, what we learn when we study ergonomics and cognetics. It is human capabilities that must be understood deeply, and these are as controlling in interface design as physics is in mechanical and electronic design.

I was completely agreeing with his statement that "The no.1 thing that you can do to create a better product is to bring the Interaction Designer in early. What you'll find is that Visual Designers brought in too early will freak out." I understand that, because they have nothing to design visually until the product is defined. Then he says, "An HCI Professional brought in sufficiently early will freak out because they have nothing to do, nothing to measure. They don't know how to design software which actually solves problems. A real Interaction Designer will rub their hands with glee and say, "Oh Boy! We have an opportunity to do something really good here" "

I am an HCI professional (I thought he was, too, unless he’s not making money at it) and I say the exact same thing, bring me in at the beginning and I rub my hands with the very same glee. That’s when a knowledge of HCI can have its most beneficial impact. HCI means Human-Computer Interaction (with the term "computer" interpreted broadly; it should be Human-Artifact Interaction, but that’s pretty abstruse sounding). HCI professionals design interaction. What is this distinction, then, that he makes about being an interaction designer? Is it someone who, by contrast with someone who designs interaction, designs interaction? I fail to see the difference.

There is, in this interview and in his books, a bias against using mathematical and formal research methods, which is a serious mistake. Because of this bias, he misses many opportunities to refine interfaces before they are tested, and to reduce reliance on argument and belief by real engineering. To over-rely on those tools is also a mistake, but Cooper doesn’t even mention them in his books, and sometimes denies that they can possibly work (for example, he said that you cannot measure the efficiency of an interface widget; this is simply incorrect).

I’ve seen some nearly useless people in HCI, and I’m sure that if the term Interaction Designer gets used widely, there will be some dumb Interaction Designers too. The name change is no guarantee of ability. From reading his stuff, I think that Cooper is likely to do a pretty good job of designing an interactive system, so long as he stays away from using straitjacket tools like Visual BASIC that encourage you to use out-dated and brain-dead widgets and crude methods of interaction. We agree that tools should never be chosen in advance of understanding the task the user wants to accomplish.

Regards,
Jef Raskin

Jonathan Korman wrote:

As an interaction designer at Alan Cooper's design studio, as well as a bit of a fan of Jef Raskin's work, I read with great interest Raskin's comments on your recent interview with Cooper. It seemed to me that Raskin mostly took issue with Cooper over some misunderstandings, which I found unfortunate since I thought this distracted from distinctions between their points of view which could lead to a fruitful discussion.

Raskin says that, in comparing the physical limitations of mechanical technologies with the limits of digital devices, Cooper "leaves out that what we are limited by is the fundamental characteristics of humans." Perhaps Cooper did not stress this enough in the interview, but uses this very point as a major theme of his second book. I believe that Raskin and Cooper completely agree about this.

The interesting distinction comes where Raskin refers to Cooper's "bias against using mathematical and formal research methods." Cooper does indeed believe that the field has focused too much on quantitative methods, most notably usability testing, as he discusses at some length in the interview.

Cooper's attempt to contrast "HCI" and "interaction design" refers to this. He intends to associate the former with a heavy dependence on quantitative techniques, and so in that sense, no, he would not call himself an "HCI professional." I sympathize with Raskin's confusion on this point, as I suspect that neither term has a crisp enough definition to support Cooper's hairsplitting.

Cooper's greater attention to qualitative methods also motivates his assertion that "good design is self-evident." Raskin should not feel so alarmed, as Cooper would certainly agree with him that developers and users do not always recognize better design at first blush. But Cooper strongly believes that an interaction designer should recognize good design when she sees it, without having to always turn to formal research. He made this comment to contrast with practitioners who have become so focused on testing that they do not trust their own judgment. He means these people when he refers to professionals who, when "brought in sufficiently early ... freak out."

This question of the differences between quantitative and qualitative methods merits exploration. Raskin asserts that Cooper's "bias against using mathematical and formal research methods" constitutes a "serious mistake" because of two opportunities that Cooper misses. I would like to take a close look at these.

First, Raskin asserts that Cooper's approach will not enable him to "refine interfaces before they are tested."

In fact, Cooper focuses on qualitative methods in part because he has seen other design organizations reach for testing before they have refined their designs to a sufficient degree.

Having worked for Cooper's studio for more than three years, I can say that our design work undergoes a process of successive refinement. We do not use any methods I would call mathematical, but we do have a rigorous and effective process, so I wonder whether it would qualify as "formal research" for Raskin: we do ethnographic research of potential users, we perform a methodical series of design exercises, we work in small design teams in order to avoid the twin pitfalls of idiosyncratic design and design-by-committee, and so on.

Second, Raskin asserts that quantitative methods can "reduce reliance on argument and belief by real engineering."

I wonder what Raskin means by "real engineering." Good interaction design demands creative methods very different from those appropriate for recognized engineering disciplines. I have seen a number of "interface engineering" efforts, but they either do violence to the proper meaning of engineering or result in unsatisfying design.

Quantitative methods can work very well to create efficient task-level interactions, and I do not want to dismiss the value of this important work. Cooper wants to critique the singular reliance on quantitative methods, not to eliminate them altogether. He talks so much about qualitative methods because they do a much better job of identifying users' needs and motivations, and therefore can lead to great breakthroughs in thinking about deeper product concepts. This can create great benefits by eliminating unnecessary tasks altogether, by significantly changing the nature of a task, by identifying unmet needs.

Consequently, I feel uncomfortable with Raskin's suggestion that we need to "reduce reliance on argument and belief." I suspect that this may reflect the puzzling unwillingness in our field to recognize the creative work of interaction designers. Many of the leading figures in interaction design--people like Bruce Tognazzini, Ben Schneiderman, and Jef Raskin himself--have done great work while seeming to credit their research rather than their thinking. I look at their work and see not the fruits of careful research so much as "self-evident" good design that springs from a vision of how we can better serve people's needs.

We often bemoan the lack of respect for interaction design in the development process. Like Cooper, I think that we could start to change that by relying less on quantitative "formal research" to justify our work, and instead should accept more responsibility for our judgment.

Jonathan Korman
Supervising Designer
Cooper Interaction Design

The role of self-evident justification vs. justification by quantitative methods in creative activities like design is interesting to me. I'm enjoying the discussion in this forum.

As an advocate of Cooper's methodology, I think he has solved a critical question in UI design methodology that was not adequately addressed previously, namely, how do we take all of that important information that we've gathered about our users and feed it appropriately into the design of the user interface?

It's not news that a great design comes about only when we know our users and meet their needs. But Cooper's concept of the user profile/archetype/persona gives us an organizing principle that helps us design for our users. And it is very powerful to bring the persona to the design team and use it to make design decisions. Should we put this feature in and how should it be presented? What should be the logical flow for using this feature? The answer comes by referring to the persona--by knowing whether the persona needs the feature and how he or she needs it to be presented.

Carla

I appreciate Korman's insight that Cooper and Raskin's fundamental disagreement is with respect to the utility of quantitative methods in interaction design. However, I think Korman misses Raskin's point about the value of an engineering approach.

I acknowledge that Raskin's use of the phrase "real engineering" may have been unfortunate if it alienated any non-quantitative readers. However, I don't think he meant to imply that a non-engineering approach to interface design is un-real or invalid. Instead he was trying to distinguish engineering from other disciplines that don't use numbers to support their conclusions.

Raskin is not suggesting that quantitative methods are better than qualitative ones, but rather that there is an appropriate place for each approach in the argument for one interface design over another.

I am all for the value of the designer's judgement. I do not agree that good design is self-evident; I am certain that good design requires good designers. However, I have also learned that good design is irrelevant unless it turns into good products and for that I need the buy-in of my business and technical partners. These people, it turns out, often appreciate a quantitative argument.

I believe the main reason why this works is that by creating a quantitative expression of my design intuition I have translated that intuition into a language common to me and my partners. I have taken it clearly out of the realm of opinion, jargon and design magic and turned it into a nullifiable hypothesis. Not only does this effort show good faith on my part, but it means that in the ensuing discussion we are all truly talking about the same thing.

Tim Sheiner
Director of Design,
Electronic Brokerage Charles Schwab