Stigmergic prototyping WORKING PAPER 2.04 Version sent to ECCO May 25, 2011
Dejonghe W. ; Detand J. ; De Couvreur L.
Industrial Design Center, HOWEST - University College of West-Flanders,
Associated member of University Ghent.
Graaf Karel de Goedelaan 5, B-8500 Kortrijk, Belgium
Designing is an activity that aims to change reality. The most challenging design assignments are wicked: the problem cannot be defined until the solution is found. Stigmergic prototyping is a method for handling wicked aspects in the development of new products, tools and services since it takes into account that not only will happen what was intended by the designers of the prototype but also something different that will emerge (express itself, organise itself) in the chosen context, embodied by the spontaneous behaviour of the interacting agents. The prototype will even make the unpredictable observable, because in the process of mutually adapting states it lets happen also something different of what was expected. The prototype is the changing mediator in the interaction, it is designed using time as a design aspect: as a changing trace of interactions. To achieve this, four mutually exclusive attitudes are distinguished giving rise to the adaptive loop of product development instead of the more traditional waterfall method. Examples are given that this method always results in solutions and exploitations. Moreover, these results could not be expected from the start.
Industrial Design, Stigmergy, Prototyping, Cybernetics, Co-Construction
Stigmergic prototyping originated as an answer of working industrial designers to the following evolutions:
Continuously evolving context, deal with wicked aspects. (Rittel & Webber, 1973; Buchanan, 1992)
To design artefacts for use by others requires second-order understanding (Krippendorff, 2007)
The need for co-evolution of environment and products. Design for adaptive capacity, resilience and resourcefulness of systems,…which consist of products. (Sammet, 2011)
Shift from design for... to design with collaborative communities/ design with stakeholders. (Sanders, 2008)
Design as Learning—or “Knowledge Creation”. Setting up strong feedback loops for building design knowledge. (Dubberly, 2011)
Evolution of role of prototyping and his formats within design. (Buxton; Detand, 2010)
Conversation as a designable action..(Jones, 2010)
Stigmergic prototyping is the preferred method used by designers to handle wicked problems, where "the problem cannot be defined until the solution is found" (Rittel & Webber 1973).
Stigmergy is the spontaneous, indirect collaboration made possible and stimulated by a shared medium (Heylighen, 2007a). The concept originated with the study of social insects. In direct stigmergy, as exemplified by the termite-hill building, it is the “work-in-progress” itself that directs subsequent contributions. Indirect stigmergy may be exemplified by the way ants create trails of pheromones that direct other ants to food sources. The trails are left as “side-effects” of the actual work being performed: finding and bringing food to the nest. (Heylighen 2007b). A process is stigmergic if the work (“ergon” in Greek) done by one agent provides a stimulus (“stigma”) that entices other agents to continue the job. (Heylighen, 2007b).
The name prototype originates etymologically from the Greek word “Protypon” (Primitive Form), and consists of two basic terms: “Protos” (First) + “Typos” (Impression). A prototype can be defined as an intermediate representation format (model) of a design, that is used to validate specific features or aspects of the final product (Chua et al., 2009)
Stimergic prototyping is the use of a prototype in a specific context to stimulate interaction that is spontaneously raised in that context, and to learn from it. This learning will have an impact on both the designers and the interacting agents, and this without the need for direct contact or communication. In this sense it is a double-blind method:
The intended influence (by the designer) on the behaviour of the agents (stakeholders) is not made conscious, and certainly not by the presence and influence of the designer. If the intended behaviour does not happen spontaneously, the design of the prototype has failed.
Moreover: the designer will be able to observe not intended uses and (to the designer new) meanings by the interacting agents because traces will be left during the interaction.
Stigmergic prototyping is a method for handling wicked aspects in the development of new products, tools and services since it takes into account that not only will happen what was intended by the designers of the prototype but also something different that will emerge (express itself, organise itself) in the chosen context, embodied by the spontaneous behaviour of the interacting agents. It is a method that even promotes and focuses on the possibility of this specific interaction: the self-organisation of designers and inter-actors using the prototype as medium or mediator.
EXAMPLES
Organizations who have built
their product service and innovation development around stigmergic
prototyping:
Designing intrinsically is an extremely complex activity and thus will benefit from a language developed to create leads to complex behavior. That language is cybernetics.
Cybernetics offers one set of concepts that, by having exact correspondences with each branch of science, can thereby bring them into exact relation with one other. And it can provide the common language by which discoveries in one branch can readily be made use of in the others (Ashby 1957).
The science of cybernetics studies ways of (possibly regulated) behaving, and thus studies (possibly intentionally made) changes in systems. These can be studied independently from the material embodiment of the behaving and thus the laws that are found in the energetic/material world. For example: every transforming system can be studied by relating input to output, up to deterministic precision, without even asking what causes the input to change to output. Cybernetics thus can produce useful predictions by only looking at relations between variables, while ignoring the physical components of the system. Cybernetics studies not so much “what is the actual behaviour of this system”, but “what are all possible behaviours of this system” trying to understand and model the determining or controlling factors that give raise to the actual behaviour. Cybernetics thus could be called the science of the possible and is the preferred language for industrial designers. Designing as a complex activity is at least a second order cybernetic process. Designing could be called the applied science of making the possible experienced.
The result of a “cybernetics of design” is that the aspects which could be important in the product-environment interaction for a preferred behaviour
get identified on an abstract level, ignoring the physical components of the system
get created on a concrete level in different variations or dynamic embodiments using some selected physical components
get partly controlled in the real context, the real behaviour providing a measure (ordering) for the selection of a preferred embodiment
This is enabled by stigmergic prototyping.
The following explains the connection with cybernetics and our affiliation with the ECCO research group
The emerging science of complex systems extends the tradition of general systems theory [von Bertalanffy, 1973; Boulding, 1956], which sought to unify science by uncovering the principles common to the holistic organization of all systems, from molecules and cells to minds and societies. However, the classical systems approach had two major shortcomings: the systems it studied were considered as (1) well-defined, static structures, (2) which are objectively given. These assumptions simply do not work for complex adaptive systems, such as societies, minds, or markets [Holland, 1996; Axelrod & Cohen, 1999]. In these systems, structures tend to be fuzzy, variable and to an important degree subjective [Gershenson & Heylighen, 2004; Heylighen, Cilliers & Gershenson, 2007]: different observers will typically distinguish or emphasize different components, boundaries or relationships. (Heylighen ECCO-paradigm)
The second shortcoming of classical systems theory, its assumption of objectivity, is overcome by noting that knowledge cannot be developed through passive observation of what “objectively” exists, but only through active construction combining a variety of subjective experiences. (Heylighen ECCO-paradigm)
The ECCO research group studies how complex systems evolve. The IDC research group looks how to approach and interact with the various agents (subjectivity of the stakeholders, context as agent) in their interaction and how this should be observed and measured. Designers, with an assignment to reach a particular goal, are agents themselves indeed, influencing the change in a direction coloured by their own preferences.
The following aspects explain the connection with the concept of stigmergy:
There is no interaction possibility without observable aspects. This parallels the action of pheromones: if the pheromones are not perceptible, if they have no meaning for the agent, no spontaneous coordination can result. If the trail is not perceptible using available sensors and interpretations, it will not be followed. If no affordances (Gibson, Norman, …) are available, no actions will follow.
Not all aspects that designers can imagine are equally relevant for the interaction. Some prototypes need functionality in the first place, some prototypes need meaning or appeal in the first place, etc... Not all prototypes are adequate media at any moment in the development process (J. Detand, et al., 2010). A model to qualify and quantify relevance in particular systems will be instrumental to the development of a stigmergic prototyping science able to predict interactions (Dejonghe W., unpublished). The same applies for stigmergy as a scientific concept that can be more than a mere description of behaviour.
Spontaneous behaviour is instrumental in stigmergic prototyping, agents will behave differently when asked for a certain behaviour: awareness of intended goals is not a necessary condition for stigmergic prototyping, the intention to cooperate is not what stigmergic prototyping focuses on. Stigmergic prototypes are effective in the coordination of activities even when the intention to cooperate is not realised. Stigmergic prototypes are made to allow reality to self-organise.
Stigmergic prototypes are adaptive prototypes and should be able to show a variety of behaviours. Stigmergic prototypes and environment are mutually affecting agents. The stakeholders interacting with stigmergic prototypes in a particular environment are being changed by the interaction. This asks for a completely new approach of prototyping, because most of the prototypes known in industrial design and product development are not easily adapted and stakeholders are seen as unchanging interactors. The fitness of stigmergic prototypes is inherently dynamic. (To adapt to a changing environment, the system needs a variety of stable states that is large enough to react to all perturbations but not so large as to make its evolution uncontrollably chaotic Heylighen 2002). Stigmergic prototypes are characterised by an openness “on the edge of chaos”: they don't force the interaction in a certain regime, nor do they provide the affordance for anything random to happen.
In all phases of product development prototypes are used to make happen a new aspect in reality, hopefully as it was intended. We can distinguish four mutually exclusive attitudes giving rise to the adaptive loop of product development. In all four “phases” stigmergic prototyping is instrumental. We distinguish the four attitudes by the four combination possibilities of the binary distinctions:
selecting (obvious meaning) versus “something different than selecting” (this means: “letting something happen” or “letting organise something itself”)
diverging (obvious meaning) versus “something different than diverging” (this means: “converging”)
The four attitudes are given the following names:
|
Name |
Conjunction of distinctions |
|
(Problem/opportunity) Definition |
Selecting AND diverging |
|
Ideation (Idea exploration) |
Let it happen AND diverging |
|
Realisation (Decision) |
Selecting AND converging |
|
Behaviour (Validation, Test) |
Let it happen AND converging |
These four attitudes are distinguished only because these describe the base competences of designers very well. In the very action of designing these four attitudes continuously change and interact.
Problem/opportunity definition. Designers are trained to take into account the demands and wishes of all parties involved in a design (these parties are called stakeholders). A designer must learn the reality of each stakeholder, and this is the reality as it is experienced from the perspective of the stakeholder. This multidimensional skill is the hallmark of industrial designers. Not in all cases this is explicit knowledge. Designers can handle explicit requirements and tacit wishes. Therefore the designer looks for traces of existing behaviour with products and environments to learn what aspects are implicitly relevant for the stakeholder. These are used as benchmarks. Typically in this phase the designer observes that the behaviour of products and contexts are not what he expected based on generally accepted knowledge. The stakeholder/context system frequently did self-organise differently based on the stakeholder's particular awareness in his particular context. Stigmergic prototypes (traces of real behaviour) are thus used by the designer to structure observations, to order idea's, to leave partly interpreted traces of observations that could get a different meaning only later when more information on relevance is available. Designers need those external traces because their (internal) skills proceed differently than what could be expected from a reliable memory.
EXAMPLE
Tandenborstel
Ijsjeshulp
Schommelkuip
Idea exploration. An assignment for a design typically cannot be experienced, it just is an idea, full of potentially conflicting requirements. Thus a priori there cannot be a right or wrong solution to the assignment. The designer will have to find “something different”, other ways to approach that reality and to make the relevant aspects available for the experience of all stakeholders. The ability to come up with various and different embodiments for an idea is called creativity. Creativity is a skill transcending the dichotomy of good and wrong. It creates and explores different possibilities, long before the problem or opportunity is understood, and in the interaction only these could be categorised as fit or not fit to a particular context. Usually the result of creativity is a multitude of fit interactions. More creativity results in more emergent aspects that can re-activate a locked-in situation. Creativity can be trained and has not to wait for the emergence of a random variation in a material realisation to come up with something new. This is instrumental for the conscious generation of new emergent stigmergic systems that could show new aspects provoking a different behaviour in a particular context.
EXAMPLE
Opzettafel
MP3
speler
Ijsjeshulp
Realisation. During product development one cannot avoid to make decisions based on existing information but with a potentially high impact (positive or negative) on the variety of future behaviour. Prototypes are used to communicate/persuade. Going further, some designers deliberately create novel situations that force their stakeholders to change their behaviour and reveal new possibilities or needs to the designer (Rust, C. 2007). The designer has to interact (not only though prototypes but usually also socially) with decision makers to reveal their tacit behaviour that otherwise could not be understood or even not revealed. Focus is laid on expected/desirable features. A lot of features are suppressed. A prototype is unavoidably simplified. Designers are trained to be maximally aware of the presuppositions of theirs and others decisions, and are able to navigate in uncertain contexts. This allows them to revisit the decisions based on newly created evidence. The most powerful documentation of the presuppositions is by leaving material traces (stigmergic prototypes) that could be used later on to reconstruct the path that was followed, but also to (re)construct the nodes where a different direction could now be followed that was not obvious at the first time (this is the so called “design paradox” Ullman D.G. 1992).
EXAMPLE
Badmintonpluim
Rollatorrem
Instructables
health
Behaviour (Validation). Influenced by prototypes, reality will organise itself: something will happen, what implies that something different will be excluded for the experience (it cannot be chosen to experience, it can happen only). This means that experienced time by the stakeholder will get a different shape: what was impossible previously now becomes possible and what was possible previously now becomes impossible. Typically, stigmergic prototypes are made to be tested double-blind in the intended environment, with the intended stakeholders, those only will determine what is relevant in that context. That means that both will influence each other during interaction, and only because of the interaction, and that the traces left by this spontaneous interaction will be used for further development. Doing so the designer, with its limited understanding of reality, does not force reality in his own limited world-view. By his multidimensional skills he provides a leverage for self-organisation in partly predictable and continuously evolving interactions. Prototypes are thus used as objects that interact with agents resulting in expected but also unexpected new aspects.
EXAMPLE
Prismabril
Trilspeelgoed
Nagelknipper
In essence, every stigmergic prototyping process consists of multiple adaptivity loops. Through co-construction the designer mediates/enables interactions. Next to the roles we have 3 main adaptive actions: Design (the artefact) – Implement (within the activity) - Evaluate (with the stakeholder). Preferably, these 3 actions happen in situ, simultaneously.
Stigmergic prototyping asks for a new approach to prototyping. The prototype as an instrument of validation is not the sole focus any more. The prototype is the changing mediator in the interaction, it needs to be designed using time as a design aspect: as a changing trace of interactions. The changing states should be observable and should be observed without influencing the interaction. This very fact requires a new kind of creativity from the designer, not found and required in more traditional research. Traces can be left in the prototyped artefact, traces can also be left in the environment, these traces can be measured (qualitative categorisation, compared with benchmarks, counted if possible). The quality of the prototype will depend on what actions it makes observable.
The prototype should even make the unpredictable observable, because in the process of mutually adapting states also something different of what was expected will happen. The focus is on the dynamics of interactions: the sequence of ephemeral states that can be observed mediated by the prototype, the environments and the stakeholders.
Stigmergic prototypes mutate continuously. Partial solutions and opportunities are isolated and recombined into new prototypes. This asks for a rigourous version management as a first step to document the attractors emerging in the interaction and to treat them as steps to learn the moving equilibrium of states.
The fitness of prototype and stakeholder is ephemeral. Lowest fitness means that the prototype does not play any role in the organisation of the stakeholder's future. The lowest fitness prototype is a meaningless prototype. Highest fitness means that the stakeholder does not want to be disconnected with the prototype any more in the chosen environment: then the prototype is not a “vicarious selector” (Campbell, Heylighen) any more for a limited amount of aspects, it became part of the stakeholder's organisation of its future. The highest fitness prototype is not a prototype any more.
Designing is an activity that aims to change reality. It is not an activity that merely describes something that exists. Designing starts with setting at least one goal: “change reality”. It typically starts with the description of possible realities and to present them so that these can be experienced and judged by stakeholders if these are preferred or not. Rapidly however are added additional goals, by additional stakeholders, like “don't change this aspect”, or “find a balance between those changes” etc.... Rapidly this sets an evolving environment, not so much because the set goals are sometimes in mutual conflict and the conflict should be resolved, but mostly because during the process new ideas and artefacts are made changing the initial conditions (the design paradox Ullman D.G. 1992). The interaction with new ideas and artefacts changes the information content and thus the meaning of the environment so that can be learnt (feedback) from what happens, with the consequence that the goals aimed at during designing will change. Designing has to deal with, and is only possible by wickedness.
The agent that is designing is the evolving team of stakeholders during the design process. Not a single individual is fully in control during the whole process. Because of its wickedness it is impossible to execute a predefined plan of the process. Thus a major skill for a designteam is its ability to navigate in an environment that forces the team to continuously adapt its goals. Moreover the evolving team should show creativity at different levels, should be skilled in creating new ideas (abstract varieties) and artefacts (concrete varieties) to help all stakeholders to interact with reality (the self organisation of reality). Therefore in industrial product development with strict deadlines, the design tasks are picked up by the configuration (of designers, suppliers, specialists of production, distibution, sales, accounting, customers, service, …) that happens to be available and is best suited to make progress in the shortest time. This is the configuration that expects to generate the most results simultaneously at that particular moment (Dejonghe W., unpublished) thus moving forward quicker than its competitors. In this, stigmergic prototyping (“the most suited prototype”) is instrumental (J. Detand, et al., 2010) because it can integrate simultaneously the aspects judged important for each discipline interacting at that moment. This industrial practice started to attract the attention of the cognitive science research community only recently (Heylighen and Vidal 2008).
EXAMPLE
Meta-design
diagram - PJ Stappers - In every order we have a different role
and different peers.
Adaptive
world technology clusters - John Reinfrank. - Lijkt me een
uitdaging om hier alles op te mappen. bit-understanding = WIKI,
bit-doing=instructables, Atom-understanding= experience prototyping,
Atom-doing=high-end prototyping?
Stigmergic prototyping also can be illustrated by ordering known products (product families, product systems) by increasing level of the impact of stigmergy in the total lifetime
Products used as intended by the designers
Products used also differently as what was intended by the designers
Products modified by the stakeholder (hacked in the distribution chain, the use phase, by the service provider...) to show a behaviour that the producer was not able or was not willing to provide
Product family of products showing different features for different markets
Product family of products with personalised components, designed for a specific user
…
Quickly succession of generations of product(family)s to change and learn from the market, as now is an increasingly common practice for the market leaders in small electronic equipment, communication systems and games
…
Custom made and custom serviced products
...
Open source local production (DIY) and continuous adaptation of products for evolving context-dependent interaction
Ashby, R. 1957 Introduction to Cybernetics p4, http://pcp.vub.ac.be/books/IntroCyb.pdf
Buchanan, R. 1992. Wicked Problems in Design Thinking, Design Issues, 8(2), 5-21. http://www.hum.aau.dk/~kanstrup/PS/09022011/Buchanan.pdf
CHUA, C. K., LEONG, K. F. & LIM, C. S. 2009. Rapid Prototyping: Principles and Applications, World Scientific Pub Co. Inc.
Dejonghe W. Unpublished, available at http://designforeveryone.ugent.be/Werkelijkheid/Designing_experienced_simultaneity.html
Detand J., et al., "The role of prototyping in developing assistive devices," in Proceedings of the 4th international PMI conference Gent, 2010.
Detand J., et al., "The role of prototyping in product development," in Proceedings of the 4th international PMI conference Gent, 2010.
Dubberly, 2011 http://www.dubberly.com/articles/design-as-learning.htm
Gibson: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.169.1004&rep=rep1&type=pdf
Heylighen 2002: http://pespmc1.vub.ac.be/Papers/EOLSS-Self-Organiz.pdf
Heylighen 2007a: http://pespmc1.vub.ac.be/Papers/AcceleratingEvolution.pdf
Heylighen 2007b: http://pespmc1.vub.ac.be/papers/opensourcestigmergy.pdf
Heylighen ECCO-paradigm: http://pespmc1.vub.ac.be/papers/ECCO-paradigm.pdf
Heylighen and Vidal 2008: “Getting Things Done: The Science behind Stress-Free Productivity” Long Range Planning Volume 41, Issue 6, December 2008, Pages 585-605 available: http://pespmc1.vub.ac.be/papers/GTD-cognition.pdf
Jones P. 2010, http://www.dubberly.com/articles/language-action-model.html
Kelley, T. "Prototyping is the shorthand of design," Design Management Journal, vol. 12, pp. 35-42, 2001.
Krippendorff, K. 2007 “The Cybernetics of Design and the Design of Cybernetics” Kybernetes 36, 9-10:1381-1392, 2007 available: http://repository.upenn.edu/cgi/viewcontent.cgi?article=1048&context=asc_papers
Norman: http://www.liacs.nl/~fverbeek/courses/hci/AffordancesandDesign.pdf
Rittel & Webber (1973): Dilemmas in a General Theory of Planning, Policy Sciences 4 (1973), 155-169 Elsevier Scientific Publishing Company, Amsterdam
Rust, C. (2007). Unstated contributions – How artistic inquiry can inform interdisciplinary research. International Journal of Design, 1(3), 69-76.
Sanders: http://www.maketools.com/articles-papers/CoCreation_Sanders_Stappers_08_preprint.pdf
Ullman D.G. 1992, The mechanical design process New York p13; MCGRAW-HILL, INC.
