A fix for fixation? Rerepresenting and abstracting as creative processes in the design of information systems

What I read

In this article, the authors investigate the circumstances that may lead to designers fixating on a certain type of design solution and therefore limiting their ability to think broadly. They conclude that encouraging designers to first abstract a problem, and to then make it again specific, encourages unique idea generation and novelty of design solutions.

The authors first describe the associative nature of thinking and problem solving, where a path leads from one idea to another. In the context of a problem, this moves from a problem towards a solution; the natural tendency is for this movement to occur in a single path that may lead to a rut, but those paths are not clearly obvious. They indicate that “one challenge for creativity is encouraging thought to travel in many directions” and “a second challenge to creativity, then, is encouraging thought in viable directions.” They note that “One way to be creative is to first generate a broad range of ideas, then check to see if they conform to the constraints of the problem”, as this is how the mind naturally works. These new designs come to life through adaptation of previous designs, and they refer to this as transfer.

They question, “what is the best way to increase the right kinds of thoughts?” and indicate that “when designers are advised to perceptually reorganize sketches, they produce more interpretations of the sketches,” and that is related to the abstract or specific nature of the problem formation, and the ability for a designer to transfer from one domain to another.

They briefly describe their study, and indicate that “to measure the creativity of designers’ solutions, we adapted previously proposed models of creativity, where creativity is measured by… originality or novelty, and the second scale evaluations practicality, usefulness, or correctness.” They indicate that they will “measure creativity by counting the number of unique ideas generated by a designer.”

The authors indicate that, in practice, designers are given concrete situations and fixate on solutions, but a way around that fixation may be to “rerepresent” the solution—to abstract a situation, and then recast it, again, specifically. They use the term to “mean a process of sequentially representing an idea in different mediums, different levels of abstraction, or both.”

Next, the authors describe their study. Students were provided with problem topics; some were written in specific, while others were generalized. They were asked to create a new situation, but that was “structurally similar.” The results were then rated by two evaluators; originality was rated on a scale of 1-9, from extremely unoriginal to extremely original. Matrices were created, reliability was calculated, and the solutions were judged for “correctness,” where if the solution was structurally matching, it was analogous, but if it was not, it was insufficient.

They then present their results; they ground their results in a previous study that indicated that “creativity is greater for those groups that have higher degree of fluency and flexibility.” They conduct factor analysis and perform further quantification of the results, which I will skip here.

A second experiment was conducted to test if the results of the first could be replicated using a different measurement of creativity, defined as fluency. In this experiment, students were asked to come up with as many solutions as possible to a problem; one group was given an abstract problem, while the others were given a concrete problem. They similarly analyze these results.

The authors then discuss their findings, and the implications of these results. They found that in both experiments, “abstractness promoted original ideas,” and participants also created more ideas (and ideas that were more original) when presented with the abstract problem statement. However, “creating more solutions does have a cost. A smaller proportion of them actually fit the problem constraints…” They note that “a good and quite general procedure for increasing design creativity is to instruct designers to first generate an abstract solution and then generate concrete solutions.” This is likely transfer at work, and it’s hypothesized that the abstraction step creates a deeper mental model of a situation. It also likely “frees up” associational processes.

They also propose a practical implication in the design of software—that “concrete requirements may fixate designers… adding a design step in which requirements are abstracted may be useful,” as it encourages different associations. A “fading procedure” might be used, where requirements (or portions of a diagram) might be selectively removed, until “the domain associations have been reduced.”

One of the larger implications of the study is that “rerepresenting generated a greater diversity of solutions than an initial abstract scenario,” but this may lead to “negative correlations between originality and fit.” They conclude that “creating solutions that expand the number and originality of design solutions has costs, in that many ideas must be rejected. Nevertheless, the payoff is probably worth it.”

What I learned and what I think

There’s so much going on here.

First, this recasts problem framing for me, and explains it in a way that I’ve never thought of before (woof, reframing framing.) When we look at a situation in a new way, it makes a great deal of sense that we’re first making it abstract, and then making it concrete again, but in a new way. That the concrete to abstract to concrete is effective is because abstract is still appropriate; it’s not the lateral move to nothing or the “how is a table like a duck?”, and isn’t really a lateral move at all. Abstraction with a concrete structure is an easing of requirements, or finding the essence of spirit of them. It’s sort of first-principles or root-cause (“Why are you asking me to do this?”) and maybe one of the reasons requirements are so obnoxious to designers is that they are often presented as concrete, but with no cultural permission to abstract them for just a little while.

I also wonder if there’s something in there about speed; what does a little while look like? If it takes you a week to understand the initial articulation of the problem, and another week to abstract it, and another week to put it back together, that’s crazy. What makes some designers able to do this quickly?

And, there’s something in here about the Goldilocks that they mention (I love that the Cog Psych community has rallied around that one…), but not in the framing as much as in the abstraction. You learn how far back you can go, and in consulting, it’s really based on the appetite of the client, and their ability to see relevance in a broad abstraction. Research generates the “fixed requirements”—the “research said this, so we must do it”—and the abstraction comes through the concept diagrams we like to use; those stand up the scaffold of a real solution space. I can do that in a day, although I know if I gave it two or three or four it would get better, and probably people who do give it that long do it “better,” but there’s no time.

Another idea that was new to me is the “fading” idea, that an approach is to just outright ignoring selective pieces, slowly, until they have forced the abstraction to happen. I think they mean it in a literal diagram (maybe an entity relationship visualization), but I have to believe that the same approach can be used with a bulleted list in a Jira ticket.

I also appreciate the idea that divergence can be valuable not in coming up with many ideas, but in coming up with many ways to interpret ideas. There's something valuable there...

I’m wondering; how do you learn (or teach) to abstract, without abstracting “too far”? That’s definitely worth some time.

I’m back to the language of assessing creativity, and I can’t help myself; I’ll try to force a thoughtful critique rather than just an emotional response.

If we’re going to study design and creativity through an objective lens (which is a bad idea), and try to somehow formalize findings into principles (which is a good idea), then we really do need something to measure. The challenge here is the relationship between coming up with a solution, and assessing if it’s a good one, because that’s really the end-game for a lot of the ways people think about creativity. It’s certainly how it’s judged in a professional context. So you ultimately get to, what is a solution, what is a good solution, and what is the assessment process to get there. I’m not giving an inch on rejecting the more or new equals good, and I think there’s only room for a real principle if it has real applicability, at least in a practiced vocation like design. So “good” in practice means that it makes it through the machine and into production, bonus points for people internally being happy about it, bonus points for users using or liking it, and bonus points for it hitting some sort of KPIs. I guess that’s probably the most effective way of judging it; assigning key performance indicators that can be measured, producing the design, and seeing if it hits those.

The problem with all of this is the need to extract a creative behavior from the context of where it happens, so you can isolate different parts of what’s going on. And it’s impossible. Even if I do give on the “count the ideas”, they need to be counted in the context of where they are made. Even if I do give on the “uniqueness of the idea,” they need to be judged for uniqueness in a real context.

So, there’s the study, practicality aside. Two groups, both with the same design problems in the same company and context, working towards the same KPIs (I can’t even write it without getting annoyed), and then judging the output against those. This is probably a fundamental to other aspects of understanding humans and human behaviors; can you (should you?) measure something that is always contextual, out of context?

Also, the end is great; "it is probably worth it" is probably right.