Representation in collective design: Are there differences between expert designers and the crowd? 1
Darin Phare, 2Ning Gu, 1Michael Ostwald
The University of Newcastle, NSW, Australia
[email protected] [email protected] The University of South Australia, SA, Australia
[email protected]
Abstract. This paper tests if a web-based crowd would, in comparison with an expert benchmark group, exhibit observable differences and similarities when they interact with varying forms of representation. The study uses an adapted online environment to provide the necessary decentralised and open conditions to support collective activity. The methodology uses semiotics to comparatively describe the processes both qualitatively and quantitatively. This paper presents the general findings of an analysis using data collected from a permanently open two-week design session. Comparisons with an expert benchmark group reveal how crowds engage with representational imagery to communicate design information in an openly shared and decentralised web based collective design context. Keywords: Design • representation • crowds • collective intelligence • design
1
Introduction
Due to the explosion of web-based technologies that enable mass communication, it is now necessary to revisit more diverse conceptual definitions of mass participation in design. Web-based technologies are increasingly using rich media content to allow large groups or crowds of motivated online individuals to contribute toward solving complex problems [1, 2]. The study of collective human intelligence in design is gaining traction through leveraging web-based outsourcing systems, known as crowdsourcing [3, 4]. The drawback with this approach is that crowdsourcing engages an online „crowd‟ in which members work in isolation to a heavily mediated structure of the design process. Such structures require participants to work according to a set of stages that artificially model the design process alone and submit work, often in isolation. Simply stated, crowd members who engage in crowdsourced design function independently of one another and without real-time communication and as a result, the online crowdsourcing structure neglects the premise that design is often characterised as much by its collaboratively social activity [5] as it is by the process of design itself.
adfa, p. 1, 2011. © Springer-Verlag Berlin Heidelberg 2011
Collective intelligence is recognised as a universally distributed intelligence; constantly enhanced and coordinated in real time, resulting in the effective mobilisation of skills [6]. Crowdsourcing rarely provides a platform for members to coordinate their activity in real time. As a result it denies the crowd the social opportunity to freely express design meaning independently of the mediated context. Thus, any result of the distributed intelligence is „collected‟ and not „collective‟ intelligence [4]. Levy [2] argued that one of the critical criteria for successful collective intelligence is real time coordination. Real time coordination is fundamental for social activity where the freedom to express and communicate enables the distributed intelligence to constantly enhance itself. These three criteria alone offer striking parallels to the social activity of design [5]. Through the work of Maher, Paulini and Murty [4] we have a conceptual framework that describes three criteria for design in this collective context; motivation, communication and representation. Motivation and communication have been extensively explored [7, 8], but representation has garnered less attention. Representation is an important vehicle for generating and communicating design meaning [9,10,11,12,13,14,]. What is not so well known is how a universally distributed intelligence would coordinate its use of representation to express and enhance design related knowledge under open, real time conditions.
2
Background
The representation, in design, is often described as a graphic notational object. By exploiting the informational potential of different types of representation the designer has to hand a means to create a vast visual record of ideas and concepts [15]. Fluency with representational artefacts enables designers to effectively communicate design at all stages of the process with other design professionals. Such graphic notational objects are generated by using analogue media (graphite and ink), digital media (3D CAD modeling) or a combination of both [15, 16]. By definition, it is expected that the majority of non-designers do not possess a similar fluency with design notations. However, the ability to exploit the informational potential of various notational signs, including models, maps and pictures, is recognised, designer or not, as a universally human skill. This skill is summarised as pictorial competence [17], which is a signbased ability that allows us to understand the representational content of “pictures, ranging from the straightforward perception and recognition of simple pictures to the most sophisticated understanding of specialised conventions” [17]. With pictorial competency being a sign based ability, the informational potential of a group of signs can be well accounted for by leveraging semiotic principles. Design and semiotics share several procedures that are directly related to the function of design representations; they both rely on descriptively graphic notation systems to provide functional and generative content, often in simultaneous combination [18]. At its core, semiotic theory is a framework in which three types of sign can be categorised, depending on how they allow for comprehension. These categories include
Peirce‟s semiotic triad (CP 1.369)1; Icons, Indexes, and Symbols to identify the leading quality and characteristic of the image itself. The Icon, Index and Symbol provide a coordinated way of talking about how meaning is expressed via the relationship between Representamen (the form a sign takes), Object (the entity to which the sign points), and Interpretant (the qualities expressed by the Representamen) [19, 20, 21, 22, 23]. The use of semiotics in the interpretation of a crowd engaged when in design activity within a web environment is an excellent qualitative approach for dissecting, explaining, and evaluating design meaning. Accounting for imagery in the online design environment by using the Icon, Index and Symbol is the formulation for the analysis presented in this paper.
3
Research design and data collection
There is no universally agreed statistic that defines a „crowd‟, but Surowiecki [24] stated that the crowd is best defined by the diversity of its constituent members, which is the adopted approach for this study. Our simulated crowd group consisted of a globally dispersed group of 18 participants. The range of personalities, gender (62% male) and diverse range of occupations was sufficient to simulate a crowd in a laboratory environment. The expert benchmark group consisted of four highly qualified design participants. The online environment in this study was a shared presentation tool called Prezi, which was selected on the basis of its ability to provide participants with an openly shared online space. The brief was open-ended and required both groups (separately, yet under the same online conditions) to generate ideas for an environmentally friendly approach to modular housing. The session remained open for 14 days. We aimed to report on the experiential differences between the expert and crowd groups and expected that the results of expert participants‟ design activity within Prezi would provide a baseline dataset sufficient for comparative purposes with the crowd group. The data collected from the experiment was coded using a coding scheme specifically developed for this framework. Coding scheme. To review the role of representation in our study, it was necessary to capture the activity involving the representation from the moment it was introduced from an external source into the online design environment. By employing Peirce‟s semiotic triad of the Icon, the Index, or the Symbol (CP 1.369), it was possible to categorise the initial semiotic value of the image according to its leading semiotic characteristics. This was undertaken for all images used by both groups for the purpose of providing an original general semiotic (coded as Sg) context. The general semiotic context (Sg) 1
All references for the work of C.S Peirce come from the collected works published by: Peirce, Charles Sanders (1931-58): Collected Writings (8 Vols.). (Ed. Charles Hartshorne, Paul Weiss & Arthur W Burks). Cambridge, MA: Harvard University Press. Any direct reference of the collected papers is written as: CP (collected papers), Vol. number (1 through to 8). This is followed lastly by the entry number (E.g. CP 3:340).
provided a starting point from which it was possible to observe and code any subsequent modifications in the semiotic quality beyond the original (Sg) state. Having been introduced into the experimental environment to convey design meaning, the semiotic value became bound to a new design-related context (Semiotics in the design context = Sd(n)). The movement between and within these contexts was numerically categorised as transitions (Tr(n)). Such changes in meaning can be identified according to the type of semiotic combination they transition (Tr (n)) from and to. For example: General to design context Design to design context
= Sg(General semiotic)→Sd
(1) (Semiotic value)
(1) (Semiotic value)
= Sd
→Sd(2)(Semiotic
value)
When meaning in any given image is changed through contextualisation, a shift occurs from what the Icon, Index or Symbol originally signified to a new or additional signified meaning. To understand how representations were used to generate design meaning in our web-based crowd context, it was important to be able to categorise the design-related content of the image. Larkin and Simon [25] and Suwa and Tversky [14] suggested that the pictorial devices for expressing meanings and concepts in design consist of depicted elements, such as objects, spaces or icons, and their spatial arrangements. These summaries of design-related information, conveyed visually, are the distilled result of extensive protocol studies of designers sketching in action, and design theories. In their 1997 study of designers and their sketches, Suwa and Tversky [14] outlined four major informational categories, each containing a number of subclasses of information. These were depicted elements, spatial relations, abstract relations and background knowledge. In summary, our coding scheme captures the changing semiotic and the informational values according to two contextual shifts. The first is coded as a general to design context and written as:
Sg(General semiotic)→Sd(1)(Semiotic value)+(Design Information value) The second contextual shift is the movement that occurs within a design related context. It is written as:
Sd(1)(Semiotic value)+(Design Information value)→Sd(2)(Semiotic value)+(Design Information value) Example: Sg(icon) Sg - Icon
(Properties)
PHOTO Properties/Materials - Tyres
→Sd (1)(index)
(Properties)
Sd(1) - Index PHOTO Properties /Materials -Tyre as building element
4
Results
To make the following comparison more effective, it was determined that only the first three transitions presented enough data (Sg through to Sd (2)) to be reliable. From Sd(2) through to Sd(4) there was not enough data in either group to be able to reliably compare and generalise the activity. As such, the data for transitions Sd(2) through to Sd(4) are omitted. Furthermore, there was a threshold imposed on any data below 5%. Data falling below this range across the transitions was merged to its nearest semiotic counterpart. The remaining data considered viable for a comparative analysis was reduced to three main transitions (Sg→Sd(1)→Sd(2)). 4.1
Comparison of representational distribution
The following analysis is achieved using statistical information and cumulative information drawn from the coded semiotic and informational activity of both groups. From this, the aim was to develop an understanding of the range and scale of the data. To cumulatively describe the construction and movement of design meaning by both groups in a customised online design environment, we employed a combined quantitative, comparative analysis of the semiotic and informational values of the representations, including the combination of semiotic and informational changes over time. The crowd group uploaded 232 general images from external sources and the expert group uploaded 81 images. In the crowd group, six images of the original 232 images were copied, and subsequently re-used within a new circle by one participant; therefore, the total final image count was 238 images (232 with six duplicates). The expert group did not reuse any images and so their total contribution remained at 81 (Table 1). The number of images uploaded was proportionately different between both groups, with the experts using an average of 6.5 more images per participant than the average crowd member who used 13.7 images per participant (Table 1). Table 1. Contributions average by group. Expert Crowd
Participants
Images
Average per participant
4 18
81 232 Diff
20.2 13.7 6.5
Of the 232 images initially introduced by the crowd, 188 were icons and 44 were symbols. No indexical images were provided. Of the 81 images introduced by the expert group, 46 were iconic, one was indexical, and 34 were symbolic (Table 2). The initial use of imagery in the expert group was heavily based on importing the icon, or symbol. In total the expert group introduced 57% icons and 42% symbols with 1% indexes over 14 days. Of these, five were initially then interacted with, followed by another two interactions, with one final interaction based on the same imagery. The initial use of imagery in the crowd group was also heavily based on importing the icon, or symbol.
Table 2. Distribution of semiotics types as introduced. Sg – Starting point Crowd Expert
Count
%
Count
%
187 0 45 232
81 0 19 100
46 1 34 81
57 1 42 100
Icon Index Symbol
4.2
Comparison of informational distributions
To determine the intended design meaning being engendered into the representation, we adopted the first of Suwa and Tversky‟s [14] two categories of design-related information (Major categories and Subclass). The major category is divided into: Properties, Spatial, Functional, Technical and Background Knowledge. All images were identified and categorised in association to their closest related major category. Table 3 presents the comparative descriptive statistics for how the images were used to describe design content according to Suwa and Tversky‟s five main types of design-related information classes [14]. Having applied the developed coding scheme that enables the identification of the static and shifting semiotic qualities, it was then necessary to further clarify the intended design meaning from Sd(1) onwards. Table 3. Proportions of the major categories of design information.
% CROWD Properties Spatial Functional Technical Background Knowledge
Sd(1)
Sd(2)
44 4 8 23 21
15 48 22 15
% EXPERT Properties Spatial Functional Technical Background Knowledge
Sd(1)
Sd(2)
28 72
20 80
Table 4 comparatively shows the proportional distribution of the representations‟ semiotic values within the major categories of design information of both groups. A chi-square test showed that the proportion of images assigned to the five categories (Properties, Spatial, Functional, Technical and Background Knowledge) differed significantly between the expert group and the crowd group, (4) = 90.3, p< .001. Table 4 shows that the majority of the expert group‟s images are in the Background Knowledge category (72.2%) and the Technical category (26.7%), whereas the majority of the crowd images are more widely spread across the Properties category (33.6%), the Background Knowledge category (23.9%) and the Technical category (22.8%).
Table 4. Combined design information and semiotic distribution
%CROWD Icon Index Symbol Icon + Index Icon Spatial Index Symbol Icon + Index + Symbol Icon Functional Index Symbol Icon Technical Index Symbol Index + Symbol B / knowledge Icon Index Symbol Icon + Index Index + Symbol Properties
5
Sd(1)
Sd(2)
6 29 1 1 3 1 2 6 4 1 8 14 3 18 2 1
11 4 48 11 11 15 -
%EXPERT Icon Index Symbol Icon + Index Icon Index Symbol Icon + Index + Symbol Icon Index Symbol Icon Index Symbol Index + Symbol Icon Index Symbol Icon + Index Index + Symbol
Sd(1) Sd(2) 2 23 2 1 40 26 4 1
20 60 20 -
Discussion
Both groups imported iconic imagery that was either scanned (such as book pages), or symbolic drawings personally created by using a variety of traditional and digital media (scanned hand sketches and Adobe Illustrator drawings). Both groups used similar formats such as Graphics Interchange Format (*.gif) or Joint Photographic Experts Group (*.jpeg) images. With these images, both groups exercised pictorial competence in intuitively or intentionally borrowing existing features of iconic imagery to express abstract meaning indexically, such as ideas and concepts. In both groups, the use of the representation therefore was the one commonly shared act in which both groups would engage in order to communicate design meaning within the online design environment. Peirce (CP 5.171) referred to this borrowing of existing characteristics to describe non-existent things or ideas, as abduction; a reasoning process allowing for the visual description of something that does not exist, such as in design. Abduction is crucial for the creative process because it enables the individual to reason upon elements embedded within existing iconic imagery in order for them to be isolated and borrowed or combined to communicate new concepts. The abductive process of borrowing of qualities was a shared practice in both groups, but it is within each group‟s abductive processes that there were observable differences in the range of information and the levels of abstraction by which that information was conveyed. Figure 1 is an example of an image used by an expert. The black and white image of
exposed concrete steps was used with the main intention of conveying an abstracted consideration based on the receptive qualities of interconnected modular components and construction methods. The crowd member introduced an image of a refurbished shipping container (Figure 2). The intention was to express a component suitable for modularity; furthermore, the image contained information regarding how such a unit could be furnished.
Figure 1. Expert representational use.
Figure 2. Crowd representational use.
For the expert group there was a much higher level of abstraction within a narrow informational framework, in contrast to a much lower level of abstraction and a much wider informational framework in the crowd group. The difference in semiotic distributions and the levels of abstractness by which these informational values were conveyed suggests differences in the experts‟ vs. novices‟ thinking within the collective design context of this study. Having generated the information (via the Sg→Sd (1) transition), certain participants began to add new interpretations (at varying levels of abstraction) indexically for the sole purpose of contributing new informational variables to the existing design-related meaning of the image. These interactions occurred on an ad hoc basis over different timeframes. All interactions that took place only occurred through indirect means in both groups. This revealed that the movement of design meaning did not occur through normal collaborative processes but by the incremental addition of „small and discreet chunks‟ of information applied to existing imagery. This type of interaction is common to collective and web-based systems and is understood as stigmergic collaboration [26]. By comparing the semiotic data of the expert group and the crowd group in the ODE, similarities were that the movement of meaning occurred in both groups. Therefore, we can infer that in an openly shared web-based context, such as the ODE provided in this study, experts and a crowd alike make use of the representation to generate and express additional design thinking based on existing and already designcontextualised imagery (Sd(n)).
6
Conclusion
Emerging from the comparative statistics were two key characteristics: both groups similarly used the iconic image to express index and symbol based content (called Abduction), and informational movement emerged in both groups, as revealed by the semiotic and informational movement of design meaning (transition). The presence of this initial activity was promising because the accumulating representational contributions vindicated the collective laboratory conditions. Moreover, the presence of activity centered on the representation was an important finding because it implies that in employing the representation to communicate design meanings, the image is a pivotal tool for the crowd and the expert alike. This suggests that there is little difference between the crowd and the expert when it comes to the uptake of imagery to express design information. Each participant in both groups demonstrated a capacity for intuitively using icons to creatively build indexical analogies for expressing ideas; this indicated that abductive processes were evident in both groups. However, the difference in semiotic distributions and the levels of abstractness by which these informational values were conveyed suggests differences in the experts‟ vs. novices‟ thinking within the collective design context of this study. In addition, the crowd‟s individual contributions of imagery were prolific, which were consistently added to and interacted with, creating a consistent stigmergic movement of design meaning. This was in contrast to the expert group, whose members traditionally do not design under collective conditions, which might explain the participatory differences between the crowd‟s consistent activity and the long period of inactivity in the expert group until the last two days of the test period. In summary, both the crowd and the expert group naturally adopted the image as the main carrier of meaning under the provided conditions. However, the engendered meaning was different between the groups.
7
References
1. Yu, L., & Nickerson, J.: Cooks or Cobblers? Crowd creativity through combination. Proceedings of the 2011 annual conference on Human factors in computing systems, NY, 1393-1402 (2011) 2. Lévy, P.: Collective Intelligence: mankind‟s emerging world in cyberspace. Perseus Books, Cambridge, Massachusetts (1997) 3. Paulini, M., Maher, M, L., Murty, P.: Understanding collective design communication in open innovation communities. Journal of CoCreation in Design and Arts (2011) 4. Maher, M. L., Paulini, M., & Murty, P.: Scaling up from Individual Design to Collaborative Design to Collective Design. In J. S. Gero (Ed.) Design Computing and Cognition (DCC) 2010, Springer, 581-600 (2010) 5. Cross, N., A.C. Cross.: Observations of teamwork and social processes in design in Design Studies, 16 (2) 143–170 (1995) 6. Ratti, C., & Claudel, M.: Open Source Architecture, Thames & Hudson (2015) 7. Paulini, M., Maher, M, L., & Murty, P.: Motivating participation in online innovation communities, International Journal of Web Based Communities, 10 (1), 94-114 (2014)
8. Maher, M. L., Simoff, S. & Gabriel, G. C.: Participatory design and communication in virtual environments, In T. Cherkasky, J. Greenbaum, P. Mambrey and J. Kabber Pors (Eds.), PDC2000 Proceedings of the Participatory Conference, 127-134 (2000) 9. Goldschmidt, G.: The Dialectics of Sketching. In Creativity Research 4, 2, 123-143 (1991) 10. Goldschmidt. G., Casakin, H.: Reasoning by visual analogy in design problem-solving: the role of guidance. Environment and Planning B: Planning and Design 2000, 27, 105-119 (1999) 11. Goldschmidt, G., Klevitsky, E.: Graphic Representation as reconstructive memory: Stirling‟s German Museum Projects. In G. Goldschmidt and W. L. Porter. (Eds.) Design representation, Springer-Varlag, London, 105-126 (2004) 12. Schön, D. A.: The reflective Practitioner: How Professionals think in action. Perseus Books (1983) 13. Asimov, M.: Introduction to Design, Englewood Cliffs, New Jersey: Prentice-Hall (1962) 14. Suwa, M., Tversky, B.: What Do Architects and Students Perceive In Their Design Sketches? A Protocol Analysis, Design Studies, 18, 385-403 (1997) 15. Ari Akın, Ö.: “Simon Says” Design is representation, Arredamento Mimarlık, 82-85 (2001) 16. as, E., Eden, H. & Fischer, G.: Enhancing Communication, Facilitating Shared Understanding, and Creating Better Artefacts by Integrating Physical and Computational Media for Design. In Proceedings of Designing Interactive Systems: Processes, Practices, Methods, and Techniques, Amsterdam, August 1997, ACM, 1-12 (1997) 17. DeLoache, J. S., Pierroutsakos, S. L., Uttal, D. H.: The origins of pictorial competence. Current Directions in Psychological Science, 12,114–118 (2003) 18. Ashwin, C.: Drawing, Design and Semiotics, Design Issues 1(2), 42-52 (1984) 19. Nadin, M.: Design and Semiotics, Semiotics in the Individual Sciences, Vol. II (W.A. Koch, Ed.). Bochum: Brockmeyer, 418-436 (1990) 20. Chandler, D.: Semiotics: The Basics, 2ndEd, Routledge, London (2002) 21. Everaert - Desmedt, N.: Peirce's Semiotics, In Louis, H. (dir.), Signo. Retrieved from: http://www.signosemio.com/peirce/semiotics.asp (2011) 22. Chapman, M., Ostwald, M. J., Tucker, C.: Semiotics, interpretation and political resistance in architecture, Contexts of architecture, The 38th lnternational Conference of Architectural Science Association ANZAScA 2004, Launceston, Tasmania. (2004) 23. DeGrassi, M., Giretti, A., Ansuini, R.: Models of Design Activities: Towards Effective Design Scaffolding in M, Zambelli., A. H. Janowiak, H. Neuckermans (Eds.) in Browsing Architecture - Metadata and Beyond, 50 – 65 (2008) 24. Surowiecki, J.: The Wisdom of Crowds: Why the many are smarter than the few and how collective wisdom shapes business, economies, societies and Nations. Doubleday Anchor, New York (2004) 25. Larkin, J. & H. A. Simon.: Why a diagram is (sometimes) worth 10,000 words. In Cognitive Science, 11, 65-99 (1987) 26. Elliot, M.: Stigmergic Collaboration: The Evolution of Group Work, M/C Journal 9, (2) (2006)