Linkography

Creative breakthroughs vary in impact. Margaret Boden distinguished between psychological creativity (P-Creativity) and historical creativity (H-Creativity), defining them as follows:

A valuable idea is P-creative if the person in whose mind it arises could not have had it before; it does not matter how many times before other people have already had the same idea. By contrast, a valuable idea is H-creative if it is P-creative and no one else, in all human history, has ever had it before. (1994, 76)

Obviously there are many more instances of P-creativity than of H- creativity, although society is more concerned with instances of H-creativity that affect the lives of most of us. Radical breakthroughs under H-creativity include great works of art, important scientific discoveries and technological inventions, and influential social reforms with long-lasting effects. However, P-creativity is important to all of us in our daily lives; we value it and try to augment it in various ways. The discussion of design creativity in this chapter has P-creativity in mind.

Since creativity as an abstract notion is elusive, most researchers prefer to study creative phenomena. One distinction that many researchers make was suggested by Howard Gardner (1988), who referred to the creative product, the creative person, and the creative process. Some researchers—among them Teresa Amabile (1983)—add the creative environment, by which they mean a social environment that fosters creativity.

A creative product, tangible or not, is usually acknowledged as a creative product only if it is both novel (original) and functional (useful), and if it is accepted as creative by the community, which may be quite small in the case of P-creativity and very extensive in the case of H-creativity. The most common method of assessing the creativity of products is judgment by experts. Most published research on design creativity treats creative products, often artifacts with enormous social, technological, and/or cultural influence.

The creative person has been studied mostly by means of case studies of individuals with extraordinary achievements and in rare cases also by attempting to profile creativity in terms of personality traits. Donald MacKinnon (1962) tried to find personality correlates of creativity among American architects by subjecting three groups of 40 architects each to a battery of personality tests. One group consisted of “most creative” architects, one of “creative” architects, and one of “regular” architects. Individual architects were assigned to these groups on the basis of a number of criteria, some of which may not be considered appropriate today. On page 3 of his 1978 book In Search of Human Effectiveness, MacKinnon wrote:

To summarize, what is most generally characteristic of the creative architect . . . is his high level of effective intelligence, his openness to experience, his freedom from petty restraints and impoverishing inhibitions, his aesthetic sensitivity, his cognitive flexibility, his independence in thought and action, his high level of energy, his unquestioning commitment to creative endeavor and his unceasing striving for creative solutions to the ever more difficult architectural problems which he constantly sets for himself.

The creative process is difficult to capture and difficult to study. There are a lot of assumptions regarding the properties of creative processes that led to the development of methods and techniques aimed at enhancing creativity in individuals and teams (e.g., brainstorming and TRIZ).

The subfield of design for which the most creativity-enhancing methods have been developed is engineering design. Behavioral studies are conducted in the laboratory and usually investigate design behavior—mostly problem solving; the problem and the experimental design usually are designed to test hypotheses, which are inferred from models of creative thinking. For example, analogical reasoning is assumed to be an important cognitive strategy in support of creativity, as it builds on abstracting the problem and thereby makes it easier to push the boundaries of the solution search space. In one study of the use of visual analogy in design, participants were asked to solve design problems under different conditions (with and without visual stimuli, and with and without instructions to use analogies based on such stimuli). Their designs were scored for creativity by naive expert judges; the scores were then assessed in light of hypotheses regarding the conditions that would yield the most creative designs (Casakin and Goldschmidt 2000). Protocol analysis is often used in such studies. In this chapter we shall look at creativity in the design process rather than at creativity in the design product or in the designer’s personality.

Biographers and researchers have attempted to explain the extraordinary achievements of extremely creative individuals by tracing the development of their great ideas. Howard Gruber was a prominent creativity researcher, known especially for his Evolving Systems approach to creative thinking (see, e.g., Gruber 1980b; Gruber and Davies 1988). He delved into the lives of very creative individuals (including Charles Darwin and Jean Piaget) and described their breakthroughs as emanating from lifelong engagement with creative endeavors. Gruber’s work was based on detailed case studies, in many of which archival materials were used. Gardner (1993) followed suit, inspecting several cases of creative individuals. His studies are less detailed than those in Gruber’s monumental work, but taken together they are highly instructive. In the design literature we do not have many comprehensive case studies, but some short accounts exist (e.g., Roy 1993). Biographies and autobiographies of designers do not differ in principle from monographs on other creative and accomplished individuals in these genres; see, for example, Leibowitz 1989, in which writings by the architect Louis Sullivan are discussed. Case studies, biographies, and historical accounts are most suitable to the study of instances of H-creativity.

In addition, there have been a number of attempts to reconstruct creative processes by prominent personalities. For example, Hans König (1992) proposed a series of actions that according to him reconstruct the process of Juan Gris in the production of a famous still-life painting. Subrata Dasgupta (1994) presented a reconstruction of Maurice Wilkes’ invention of microprogramming, which may be seen as an extremely creative case of design. Dasgupta’s analysis, though computational in essence, is a rare attempt to see the case through a cognitive lens.

The psychometric approach to creativity was pioneered by J. Paul Guilford, who suggested that creative achievement is commensurate with mental abilities that can be differentiated and measured, much as in studies of intelligence (see, e.g., Guilford 1956). Guilford talked mainly about productive thinking, as had the Gestalt psychologists several decades earlier. He claimed that creativity depends largely on divergent thinking, which included fluency, flexibility, originality, and elaboration. In addition, he claimed that transformations of thought, reinterpretations, and freedom from functional fixedness in deriving new solutions were essential to creativity. Guilford’s Structure of the Intellect approach led to the development of creativity tests, of which there are many variants; the best-known ones were developed by E. Paul Torrance. Torrance’s Tests of Creative Thinking (1974), which are still in wide use today, score abilities related to what Torrance called “skills that seem to be important in producing creative responses” (1988, 66). Criticisms of such tests question their predictive power and point out their apparent susceptibility to administration, scoring, and training effects (Plucker and Renzulli 1999).

The only specific psychometric studies of designers of which I am aware are in MacKinnon’s 1962 study of architects’ personalities, which used psychometric tools.

The experimental approach is not radically different from the psychometric approach. However, it usually tends to separate specific problem-solving cognitive capabilities and study them in isolation. An example may be the role of visual imagery, or the role of analogy, in creative thinking and problem solving. This type of research is most appropriate for the study of the creative process.

Neurocognitive studies of creativity center on memory activation in the brain, which displays different patterns in creative thinking than in “ordinary” thinking. The first notable trait of memory is that it is distributed (Kanerva 1988). This means that the storage of memory items is distributed across many memory locations (neurons) in a restricted region in the brain. The second important trait is that memory is content addressable (Gabora 2010). This suggests that the content of a memory item corresponds to the location (neurons) in which it is stored and from which it can be retrieved. As a result, items with related meaning are stored in overlapping or close locations.

Attention to a stimulus causes activation. The pattern of activation may be flat or spiky, depending on the type of attention paid to the stimulus and the kind of thinking it evokes. A flat pattern corresponds to a high level of activation that results from de-focused attention to a stimulus or stimuli. In this case, more overlapping memory locations are activated in the relevant region. De-focused attention is directed at the overall contemplated image and many of its details. In contrast, focused attention, wherein attention is paid to certain details of a stimulus, results in a distribution of memory locations that are farther from one another, and therefore create a spiky pattern. The spiky pattern corresponds to analytic, convergent thought, whereas the flat pattern of activation is more related to associative, divergent thought. Creativity requires variable focus, i.e., shifts between the two modes of attention. This is one aspect of memory activation that is controllable, and the ability to spontaneously widen and shrink the scope of attention and therefore the activation function is indispensable in creative thinking (Gabora 2010; Martindale 1999). Since designing involves rigorous visual thinking and representation (internal and external) and intensive use of visual stimuli of all kinds, these finding are of particular relevance to the study of design thinking and design creativity, although the explorations reported in this book are obviously not conducted in the realm of neurocognitive research. We recognize that, among other differences, the time scale of neurological activity is completely different from that of cognitive operations, and therefore no direct application of neurological data to the type of research discussed here, which is largely behavioral, is suggested. This line of research is mentioned here because the findings so far appear to support outcomes obtained in behavioral cognitive research and because the two fields are moving toward one another. This appears to offer many new research opportunities for the study of design thinking and design creativity.

Creativity can be studied by conducting regular behavioral experiments of the kind used to explore cognitive activity, and the linkographic experiments are seen as belonging in this category, although in most cases no hypotheses are put forth to be confirmed or refuted.

In chapter 3 it was suggested that linking moves forward and backward reflects shifts between divergent and convergent thinking. It was also stated that the creativity literature concerns itself primarily with divergent thinking, and that psychometric tests are geared to measure divergent and not convergent thinking or shifts between the two. This is also true of design research as exemplified by the recent work of Toshiharu Taura and Yukari Nagai (2013). Taura and Nagai maintain that creative design is the function of designers’ ability to expand their thought space (the equivalent of the more usual term “design space”) during the preliminary concept-generation phase. Taura and Nagai recognize the importance of going back and forth in thought, but they still concentrate on divergent thinking and have little to say about convergent thinking.

I take the position, supported by neurocognitive research, that creative thinking involves both divergent and convergent thinking. Support for this view is expressed by several cognition researchers. For example, David Perkins (1981) proposed that critical responses in creative thought involve an intuitive process and an analytical process, which are roughly the same as divergent and convergent processes. Perkins went on to say that these processes are usually discussed separately, but that on his view “the two strategies could occur mixed in behavior” (ibid., 105). Elsewhere he stated explicitly that “typical inventive thinking appears to demand a generous helping of both [divergent and convergent thinking],” and that “inventive people are mode shifters” (1992, 249). In a later text, Jonathan Plucker and Joseph Renzulli wrote:

Divergent thinking tests historically occupy nearly the entire creative process spotlight. Since the ability to generate ideas is only one aspect of the creative process (see, e.g., Runco and Okuda, 1988, in their discussion of the componential theory of creativity), its predominance devalues the integral role of creativity in the solving of problems. . . . Runco (1991) observes, “The evaluative component of the creative process has received very little attention. . . . This is surprising because it is a vital constituent of the creative process, and is required whenever an individual selects or expresses a preference for an idea or set of ideas. (Runco 1991, 312; Plucker and Renzulli 1999, 41)

Several methodologists have talked about divergent and convergent design phases in a prescriptive model of designing (see, e.g., Cross 1994; Fricke 1996; Pugh 1991; Roozenburg and Eekels 1995). In reaction to such talk, Ying-Chieh Liu, Thomas Blight and Amaresh Chakrabarti (2003) challenged the notion that conceptual design consists of broad phases of divergence, and then similarly extensive periods of convergence. They proposed an alternative view, “multiple divergence-convergence,” that they claimed was more accurate. That approach, Liu et al. recommended, “is to carry out divergent and convergent activities in each level of solution abstraction” (ibid., 346). Regarding spoken language in design processes involving more than one person, Andy Dong wrote: “Designers routinely draw from a wide body of knowledge and experiences, personal and otherwise to establish the meaning that is assembled into the designed work. . . . Our explanation is that their language is aggregating toward a coherent design concept through cycles of convergence and divergence.” (2007, 11) None of these studies reported empirical investigations at the cognitive level. Since the cognitive level is the main focus of this discourse, let us look at design creativity in terms of forelinks and backlinks of critical moves.

In chapter 3 above, it was proposed that forelinks represent divergent thinking and that backlinks stand for convergent thinking. In this chapter, special attention will be paid to forelinking critical moves. This is not to say that backlinking critical moves will be neglected or considered secondary.

In his studies of the lives of very creative individuals, Howard Gruber did not conduct experiments, did not study short “live” episodes, and did not look at small-scale cognitive processes. But some of his insights into long-term creative endeavors also apply to short-term processes, just as smaller-scale neurological patterns are relevant to explorations at the cognitive scale. Gruber, who avoided overemphasizing single inspirational moments, proclaimed that “emphasis is withdrawn from the supposed single great stroke of insight and transposed to the many moments of insight that occur in the course of a creative effort” (Gruber and Davies 1988, 244). Elsewhere, Gruber stated that “interesting creative processes almost never result from single steps, but rather from concatenations and articulation of a complex set of interrelated moves” (1980a, 177).

If Gruber is right, we may conclude that the networks of links in linkographs may be useful in detecting creativity, in this case design creativity. Linkographs do not necessarily tell us everything there is to know about the creative process; referring back to the protocols may yield additional noteworthy information. For example, a protocol may reveal that a well- structured and highly interconnected linkograph describes a process involving analogical reasoning or imagistic reasoning—something that cannot be inferred from the linkograph alone. It is reasonable to assume that, at the cognitive scale at which design thinking is being dealt with in this book, design processes differ from one another in the level of creativity they exhibit. Here linkography will be used to point out these differences.

For an example, let us return to the Delft team protocol, documenting a bicycle rack design. A breakthrough was clearly identifiable when the designer John made a move in which he responded to the need to do something about the backpack’s straps and suggested that the team see the bicycle rack it was designing for the backpack as a tray (unit 32, move 30):

Move 30 (John) So it’s either a bag, or maybe it’s like a little vacuum-formed tray kind of for it to sit in.

The tray idea was soon accepted by the other members of the team, who recognized that it addressed two aspects of the problem that were not resolved by seeing the rack as a flat plastic surface: how to contain the backpack’s straps and how to protect the backpack from mud. The team immediately developed the tray idea, recognizing that it addressed all their concerns (later the production method was changed from vacuum forming to injection molding). Figure 6.1 shows the linkograph of unit 32 of the team protocol, in which all of this occurred.

Figure 6.1

Linkograph of unit 32, Delft team protocol.

Between moves 28 and 54 there is a heavily interlinked group of moves, including three webs (moves 30–37, moves 41–46, and moves 45–51) and several critical moves, thirteen of which are at the threshold level of six links. Seven of these moves are CM⁶s>, five are <CM⁶s, and one is a <CM⁶>. Let us look at these critical moves—or this “critical path,” as John Habraken would call the sequence:

Move 28> (Ivan) We’ll just call it that for now, er bag, put it in a bag, we’re gonna need some sort of thing to do something with those straps

Move 30> (John) So it’s either a bag, or maybe it’s like a little vacuum-formed tray kind of for it to sit in

Move 32> (John) ’Cause it would be nice I think, I mean just from a positioning standpoint, if we’ve got this frame outline and we know that they’re [the clients are] gonna stick with that, you can vacuum form a a tray or a

Move <35 (John) Maybe the tray could have plastic snap features in it, so you just like kkkkkk snap your backpack down in it

Move 41> (Ivan) Uh uh what if your bag were big er, what if you’re, you’re on er, if this tray were not plastic but like a big net you just sorta like pulled it around and zipped there I dunno

Move <42> (John) Maybe it could be part, maybe it could be a tray with a, with a net and a drawstring on the top of it, I like that

Move 44> (Ivan) A tray with sort of just hanging down net, you can pull it around and and zip it closed

Move 45> (Kerry) It could be like a a a window shade so you can kinda, it sinks back in, so it just

Move <50 (John) So what we’re doing right now, though is, we’re coming up with like again classifications of solutions of kind of all, they’re all “either or” things, I mean like we wouldn’t do the net and the shade and the snap in with the tray either or any one of those will probably

Move <51 (Kerry) A net can be combined with a shade, I mean you could have a retractable net that, that’s how I thought of it

Move <54 (John) I think tray is sorta a new one on the list, it’s not a subset of bag, it’s a kind of er yeah, but oh yeah yeah oh I see, shade straps is how do you dress the straps on the back

This is a very interesting group of moves, all generated in a little less than three minutes (between moves 28 and 54). In the first move the bag concept was seen by Ivan as a fait accompli that still required a solution to the problem of the straps. As soon as the tray concept was raised, just a few seconds later, it was contemplated in conjunction with concepts raised earlier (bag, net, shade), and attempts were made to combine all of these concepts. One could even interpret the verbalizations by Ivan and Kerry as a careful and subtle struggle to preserve ideas that they had already accepted and that they were not easily willing to discard in favor of the new idea. John, the initiator of the tray concept, insisted that the tray was a stand-alone idea that was not a subset of any of the previous concepts. All the same, as a veteran team player, he displayed openness to the possibility of a combination of ideas—hence the double-speared move 42, in which John conceded that the solution could be a tray with a net and a drawstring.

Several of the participants in the Delft protocol workshop pointed to this passage as a turning point in the design process, and this was evident in one way or another in the notations they used in their analyses of the protocol (Cross 1997). But linkography, Nigel Cross said, was “the analysis that came closest to both tracing the history of the emergence of the tray concept and indicating its important role” (ibid., 314).

For present purposes, this provides an excellent example of the birth of a new and creative idea that was not completely detached from what had happened earlier in the process, including the flat plastic rack in earlier units of the protocol. While relating to previous deliberations, this idea also affected subsequent moves in the design process. The linkograph shows that moves 60–68 are heavily linked to moves 41 and 44, and to a lesser degree also to move 42. Moves in posterior units also related to the tray concept, but this cannot be shown in the linkograph of unit 32. This short episode emphasizes the network nature of design thinking in general and creative design thinking in particular. It also shows that inventive ideas, if they are to be successful, must be developed further and must be rigorously assessed against previous work, including design decisions already made and design criteria already agreed upon (e.g., containment of straps). Therefore, we get both CMs> and <CMs in this passage—that is, both divergent and convergent thinking take place, with frequent shifts between the two modes. The high percentage of CMs in this unit is also indicative of its highly productive nature: at the level of CM⁶, 27 percent of the moves are critical; of those, 58 percent are critical because of forelinks. At the level of CM⁷, 19 percent of the moves are critical and of those, 62 percent are critical because of forelinks. It may also be worth noting that most of the CMs are strategically located at beginnings and endings of discernible structural components of unit 32 (see chapter 5).

We can also scrutinize the threshold level of CM⁵ and compare the values found in unit 32 with those of units 37–39 in this protocol (see table 5.3). The proportion of CM⁵s is 36 percent in unit 32, 24 percent in unit 37, 23 percent in unit 39, and 10 percent in unit 38. Fifty-six percent of the CM⁵s in unit 32 are critical because of forelinks. We see that, despite the creative leap, the proportion of CMs> at all criticality thresholds is not overwhelming, between 56 percent and 62 percent. This means that a considerable number of critical moves are devoted primarily to assessment and evaluation, thus concurring with the theory that calls for divergent and convergent thinking in creative feats. The proportions above demonstrate that we see an increase in critical moves in a passage recognized independently as creative in the sense that it includes a breakthrough. These critical moves are rather well balanced between forelinking and backlinking, that is, divergent and convergent thinking, with frequent shifts between them.

As was discussed in chapter 5 and shown in tables 5.6 and 5.7, critical moves are in fact bidirectional. That is, although they are critical owing to their links either forward or backward, they also “dedicate” 20 percent or more of the total number of links they generate to the opposite direction. This reinforces Perkins’ (1981) view that the two modes of thinking “can occur mixed in behavior.” Here we ought to refer once again to Gruber, who wrote: “Insofar as divergent thinking plays some role in creative work, the system within which it occurs must regulate it, prevent it from running amok, require it to generate a well-chosen alternative.” (1980b, 287–288) The required actions to which Gruber pointed can be interpreted as convergent thinking embedded in backlinking critical moves and backlinks of critical moves with mostly forelinks.

Let us return to a study, described in chapter 5, wherein Dan Tatsa looked at ideas generated by architecture students in a semester-long studio project (see Goldschmidt and Tatsa 2005). After the end of the semester the students themselves assessed their projects’ ideas in terms of criticality, which meant contribution to the final outcome. In addition to a connection that was found between the level of self-assessed criticality of ideas and link indices derived from linkographs (as reported in chapter 5), the same study also looked at correlations between ideas and the final grades assigned to the students. This is of interest here because the criteria for awarding grades emphasize creativity. First Tatsa calculated the correlation between grade and the number of all ideas; this correlation was not significant. He then repeated the calculation for critical ideas only, and found a significant positive, even high, correlation between their number and the project’s grade (N = 8; r = 0.699, p < 0.5). This confirms that processes that included a higher number of critical ideas, as assessed by the students themselves, yielded outcomes that teachers deemed higher in quality and more creative. This finding is in line with results from the Delft protocols research described in this section, namely, that creative episodes display a high proportion of critical moves.

Let us now turn to another example, one derived from studies by Remko van der Lugt (2001, 2003) that do not focus on critical moves but still count links. Van der Lugt (2001) studied differences among structured idea generation meetings and the effect of sketching on the processes in these meetings. To do so he used linkography that facilitated quantitative analyses and comparisons of processes. Linkographs were constructed for design sessions that were conducted using different methods, so that the pattern of links among the ideas raised in them was revealed. A link in this case meant that an idea built on a previous idea generated in that meeting. In a final step, the participants were asked to appraise the ideas raised in the meeting in terms of their creative qualities. Later these appraisals were correlated with their “linking scores” (van der Lugt 2003). Each participant was given a number of colored stickers. The four red stickers were to be used to indicate the four most surprising ideas, then four green stickers to indicate the four most feasible ideas, and the one blue sticker to indicate the most exciting idea (the idea the participant “would be most excited about developing further”). The 192 ideas that had been generated in four design meetings were then scored by the number of stickers applied to them according to the formula

Score = green stickers + red stickers + 2(green and red stickers) + 4(blue stickers).

Using the scores, van der Lugt divided the ideas into two groups: ideas with a score of 7 or more and ideas with a score of less than 7. Then t-tests were used to check the difference between the groups in terms of the number of links among ideas as notated in the linkographs. The difference proved significant (p < 0.05), with 3.08 links for the high-scoring ideas and 2.13 for the low-scoring ideas. A detailed analysis of the differences between ideas that received colored stickers in any of the categories and ideas that did not receive any stickers also showed that the former had higher rates of linking. Van der Lugt concluded that there was a strong connection between “the perceived quality of an idea” and the number of links it generated (ideas it built on, and ideas that built on it). For van der Lugt “idea quality” is unequivocally related to creativity and the criteria for awarding stickers reflect this stance. However, he proposed that further research should use quality assessments by independent judges, rather than self-appraisals. Even so, this study lends further support to the notion that intense interlinking among design ideas is a prerequisite for creativity.

Let us now return to the MIT branch library design study, in which the participants included “ordinary” architects and studio teachers, one novice student, and one prominent architect (Martin, who at the time of the experiment was internationally very famous). Martin handled the library design task easily, explaining the principles of his design theory in length and then producing a comprehensive solution within minutes. His design was by far the most complete and most creative of the designs obtained in this exercise.² This is not surprising—we know that experts, and certainly masters, “immediately produce higher quality options” (Björklund et al. 2013, 139) and “seem to choose a preferred option rather early on” (ibid., 140), with no need to consider many options. Figure 6.2 is a linkograph of the 22 moves in which Martin completed most of his design solution.

Figure 6.2

Linkograph of episode from Martin’s design process, branch library design task.

When the linkograph in figure 6.2 is compared with those of the processes of Gilbert, Gideon, and Glenda, who worked on the same task (see figures 5.6–5.8), it is immediately apparent that Martin’s linkograph is denser in links and contains a higher proportion of critical moves. Although direct comparison is not possible, because this linkograph is considerably shorter than the other three, it appears that Martin (whose process had two distinct portions) was able to interconnect his design acts in a dense mesh—even more so than Glenda, who stood out among the three linkographs portrayed in chapter 5.

When he first saw the library’s footprint, Martin complained that it lacked organization. He proceeded to explain and sketch how spatial organization might be achieved. After Martin stated that he was not able to work without any context, the researcher drew a rectangle representing a site with the library in its midst, added some trees (scattered about the site), and indicated north. Martin postponed dealing with the articulation of spaces within the library until later in the process. Let us look at Martin’s critical moves at a threshold of four links:

Move 2> It’s important enough because there [outdoor area shaded by trees, across from the library’s entrance] one starts creating a space.

Move 4> I would make the entry—one can come through the cars [parking] here; it’s the case of the other side as well.

Move <9> We start creating a hierarchy. . . . The large trees, the parking lots, the pedestrians, an entry axis.

Move 10> I would then look for a direct relationship between entrance and exterior, because here, the real edge is not this [edge of building], for me it’s that [boundaries of the lot].

Move 17> So this is a way of letting elements of nature into the project. I don’t like isolated buildings.

Move 18> The building needs to reach from here to here [boundaries of the lot]. I would try through this organization to make space reach until there, so it has tension, it has a prolongation. It isn’t a small shed, it isn’t a pavilion on the interior. There. This is a different approach, but already, these elements there have found a different organization, and beyond functions, beyond, I would say, the interior organization.

Move <20 So I will make the organization of the parking lots give force, with the trees, to this idea. I always want to take the entire, all of the land.

Move <21 I think that if there are two roads, there is already something artificial . . . drawn by man. I find it justified to master the whole. Not let the pavilion swim in the middle. So you see how the discourse becomes immediately one of exterior.

In the moves listed above, Martin did the following, while sketching continuously: First, he noticed that the trees, which cast shadows, enclosed an outdoor area adjacent to the library wall. He established an entry point into the library’s footprint across from the wall facing the trees, and designated parking lots on either side of the entrance. He then reflected on the relationship between the parking lots, the entry, and the outdoor space enclosed by the trees in the “back” of the building. He declared that a hierarchy had been established among these elements through a (visual) axis that ran from the entry to the trees in the back, through a glazed back wall. He went on to emphasize the relationship between inside and outside spaces, adding that for him they were inseparable since the edges of the building were really outside, at the boundaries of the lot, and included outdoor spaces. For Martin, the hierarchy and the axis connecting the main elements of the building and its surroundings and governing the relationships among them, were the way to achieve organization, which was important to him. An examination of Martin’s built works and their documentation in professional publications shows that indeed motifs of hierarchy and visual axes play a very important role in his approach to designing buildings. The emphasis on the relation of the indoor spaces with the exterior, which came up again and again in the protocol, is also a hallmark of Martin’s built work (see also chapter 3).

In the linkograph of this short vignette we see that even at the threshold of five links there are three critical moves, or 14 percent of all moves. When we look at CM⁴s, we find eight critical moves, or 36 percent of all moves—a high percentage, especially in comparison with the 18 percent in Glenda’s protocol and the much lower percentages in the linkographs of Gilbert and Gideon. Such high rates are not typical of most episodes, with the possible exception of short passages in which a significant breakthrough is achieved, as we have seen in the case of unit 32 of the Delft team protocol. In Martin’s process in this task we cannot claim a breakthrough; instead we know that a “good fit” among elements, which was quickly established, turned the ill-defined problem into an easily solvable one. Interestingly, the CM⁴s include five CMs>, two <CMs, and one <CM>. The two-headed critical move, quite a rare phenomenon, is move 9, in which the all-important principles of hierarchy and axis appear.

We can compare Martin’s linkograph to those of Gilbert, Gideon, and Glenda in another criterion: that of the proportion of unidirectional and bidirectional moves. In Martin’s entire linkograph we find only one move that has no backlinks (move 4) and only four moves that have no forelinks (moves 7, 14, 16, and 21). This amounts to 25 percent unidirectional moves, and therefore 75 percent bidirectional moves. The percentages of bidirectional moves for the three other library designers are 67 percent for Glenda, 53 percent for Gideon, and 67 percent for Gilbert.³ Similar values are encountered in other linkographs; for example, the proportion of bidirectional moves in the ashtray linkograph (figure 4.3), which is of similar length, is 64 percent. Bidirectionality is related to fast shifts between divergent and convergent thinking, which is typical of creative processes. Therefore, this short episode exemplifies Martin’s almost simultaneous divergent-convergent thinking in a highly creative instance of designing.

Much can be learned from a comparison between Martin’s high-end process and the very low-end process of Alice, another participant in the library design exercise. Figure 6.3 shows the linkograph of a portion of Alice’s process. Alice, a first-year graduate student of architecture, had no design background (her undergraduate degree was in chemistry), and in the exercise she groped to make sense of the library’s footprint and the consequences of entering it from each of the optional entry points. Her great discovery was that penetrating the building near the center is advantageous in terms of people’s circulation in the building. In the debriefing after the exercise, when asked about possible differences between her process and the presumed process of an experienced architect, she replied as follows:

Figure 6.3

Linkograph of episode from Alice’s design process, branch library design task.

I don’t know if he would look at this [footprint] and immediately see these [implications of entry options] and wouldn’t even have to draw it for himself because he can see it. I felt I had to stop and go through all of them [entry options] and actually draw out for myself what I’m saying in words. He might just say “this, this and this. I know what I’d prefer, or what my client would prefer.” . . . He could just look at it and immediately see a lot of things I couldn’t.

Alice got it right: Martin worked very rapidly, as if he could see the complete solution in his head before sketching it, much as Mozart was said to be able to hear a whole symphony in his “mind’s ear.”

The linkographs of the processes of Martin and Alice are similar in length (22 and 27 moves, respectively), but that is about all they have in common. Alice generated a small number of links, and no move in this stretch has more than two links either backward or forward. Therefore, there are no critical moves at all. The proportion of bidirectional moves is 40 percent, similar to Gilbert’s. We also do not discern a visible structure in this process, whereas Martin’s process is highly structured. Alice’s process was not creative in this episode—not necessarily because she is not a creative person, but perhaps because she lacked the skills needed to handle the task and her approach was that of a layperson. Her thinking pattern was not a divergent-convergent one, and she was content with just understanding how entering at one or another point affects movement in the library. To accentuate the difference between the processes of Alice and Martin, sketches by both of them, exploring the implications of entering the library at entry point 1, are presented in figure 6.4 (for all the optional entry points, see the appendix).

Figure 6.4

Explorative sketches of implications of entry 1, marked by arrows at the top. (a) Martin’s. (b) Alice’s. In these sketches the library’s footprint is turned 90º relative to the original drawing (see appendix).

Only a very experienced and creative designer with a solid design theory, and a well-developed kit of tools that allow him or her to mold any design situation so as to be manageable with those tools and according to that theory, can operate the way Martin did here. This entails moving back and forth, thinking divergently and convergently with very frequent shifts in a structured process, while generating moves that are highly interlinked, for the most part in both directions. Alice, a novice with very little domain-specific knowledge, could not possibly match such a performance.

Needless to say, there is more to design creativity than links among moves. Other important features of creative thinking—for example, prioritizing ideas, identifying opportunities, being able to use imagery or think analogically, and discerning useful information in stimuli—have not been discussed in this chapter. We have focused on the ability to connect and synthesize information, something that expert designers do well. According to Tua Björklund (2013, 152), product-development experts “accommodate for a higher degree of interconnections both within the mental representations and between the problem representations and outside knowledge.” Björklund has shown empirically that Alexander’s (1964) “good fit” concept, according to which a successful design synthesis depends on a good fit between the solutions to subproblems into which a complex problem must be broken, is as valid as ever. The aim of this chapter, and that of the introductory comments in chapter 3, was to show that connectivity, or good fit, or a particularly successful synthesis, is a product of rapid shifting between small divergent and convergent thinking acts. The emphasis is on small: not only does creative design encompass divergent and convergent thinking; these modes of reasoning are so intertwined that it is hard to separate them.

Linkography proves to be capable of disentangling instances of thinking as reflected in protocols, and of showing the close relationship between these two modes of thinking and reasoning. Thus, it is not enough to ask designers to generate more ideas, or even to generate innovative ideas. A designer has to acquire the mode of thinking whereby frequent propositions are immediately followed by evaluative steps that make sure that the design holds together at every moment. It is not enough to have ideas, even many ideas; the ideas have to be good, and to be good they must be perfectly integrated and interlinked.