Professional Documents
Culture Documents
Tasks
Acknowledgments: The authors would like to thank the Engineering and Physical Sciences
Research Council U.K. for funding this research, and the participants who gave their time to
Cite as: Coughlan, T., & Johnson, P. (2008). An exploration of constraints and end user
Abstract
Creative tasks are performed within constraining structures. However, the exploration
and development of these structures is central to the creative process. This paper summarises
research on the role of constraint in creative tasks, defines the role of constraint development
in a model of the creative process and classifies types of constraint in order to inform the
design of environments for creative tasks. We explore links between constraint development
and end user development in software environments through analysis of the design and use of
Music Builder, a prototype for the user development of musical instruments in a collaborative
Tasks
Creativity has been studied across a wide range of disciplines, leading to a detailed
understanding of the processes involved from multiple perspectives. Our aim is to take what
is appropriate from this body of knowledge and translate it into useful information for the
design of interactive systems. Therefore we need to understand how people interact with the
environment and each other in creative tasks. Creative work can be understood inclusively as
a tension between divergent and convergent thinking (Baer, 2003), an iterative process of
ideation, representation and evaluation (Coughlan & Johnson, 2006), and as occurring where
creativity occurs and shapes the evaluation of creative outputs (Csikszentmihalyi, 1996).
motivations, tools and conceptual structures are necessary to direct and support creative
activities, but that the ongoing questioning and development of these structures is central to
the process.
Across domains, creative work occurs within various structures, from a physical
practitioners have the ability to explore and modify aspects of these structures. The novelty
required in a creative output commonly arises through the use of novel variations or
combinations of concepts and tools, compelling creators to explore paths less followed. We
argue that computer-based environments for creative tasks both constrain the creative process
in new ways and provide new possibilities for creators to work with constraints, particularly
Constraints and End User Development 4
through the notion of end user development. To move from theoretical understanding to
practical design knowledge we analyse the design and use of a system for the user
The atomic elements of creative work are ideas – a thought or suggestion of a possible
course of action. The nature of an idea is domain-specific, as are the forms through which it
is represented (Coughlan & Johnson, 2006). Ideas do not occur in a vacuum, but exist with
reference to specific situations, knowledge and stimuli, and in response to problems we are
motivated to explore. Finding these problems of interest is key, whilst problem solving is ill
structured, because the path to a suitable solution is not clear as the task is performed.
Spatial metaphors have been used to represent the creative process from this
perspective. Creators approach a new task with a wide, unexplored solution space, forming a
vast array of possible actions (Boden, 1991). Decisions are made to place boundaries within
this space or choose areas to explore. A minimalist overview of a creative process would
begin with initial exploration of some of a vast range of possible actions – perhaps by making
instrument. The space of possibilities diminishes in scope as choices are made to keep or
discard ideas and concepts that have emerged – for example if a particular musical phrase or
subject matter is identified as valuable, subsequent ideation and evaluation will take it in to
account. In reality there can be a great deal of deviation and reversal as creative practitioners
creative domains: Scientific theories are created as possible solutions within a set of observed
Constraints and End User Development 5
constraints. Designers are generally presented with initial constraints required of any
solution, such as health and safety laws or limited budgets. Artistic work is perhaps less
constrained by external prescription, but in its place practitioners enforce constraints to focus
their activity, for example by concentrating on a single subject matter or restricting the
materials with which they work. Stokes argues that useful constraints preclude conformist,
tried and tested responses and promote novelty. Successful creators generate novel, valuable
products by finding or developing demanding task and subject constraints that partially
In design, constraints are commonly identified as a basis for bounding the design
space in which solutions must fit. A range of constraints from cost or available materials to
the functions required of the design can be identified and used to frame the process.
Chandrasekaran notes that “all design can be thought of as constraint satisfaction” but in
design tasks of any complexity the space satisfying constraints that can be formally defined
Pérez y Pérez and Sharples classify constraints in creative writing either as external
factors - tools and resources existing in the world, or mental constraints that reflect
requirements for content or rhetorical issues of how best to compose for the audience.
Writing involves periods of engagement and reflection. Engagement entails realising ideas
that are expected to comply with constraints. Constraints “drive the production of material
during engagement”, but writers also enter periods of reflection when they run low on ideas
or feel compelled to depart from current constraints due to divergent thinking (Pérez y Pérez
& Sharples, 2001). In modelling the process of musical composition, Pearce and Wiggins
identify three types of constraints on the composer: Internal - the requirement to fit with
existing ideas in the composition, external - the limitations on what it is possible to create,
Constraints and End User Development 6
given current skills, tools and knowledge, and stylistic - constraints relating to the intended
existing computer environments; for example there is a stark contrast in the musical domain
between the highly collaborative nature of most musical endeavours and the single user
design of the vast majority of music software (Coughlan & Johnson, 2006). Therefore this
studies (Coughlan & Johnson, 2006), we have developed and refined a descriptive model of
the creative process. Creative work is defined as a cyclical, iterative process of ideation – the
generation and representation of novel thoughts - and evaluation – reflection on the value of
generated ideas. These cycles are interspersed by periods of constraint development, a sub-
process that considers the fit of existing constraints, how new constraints could affect the
outcome and then implements changes to the constraint structure. The model has been
developed to aid general understanding of creativity support needs and to define the role of
constraint development.
Forms of Constraint
Our model defines three forms of constraint from the perspective of designing
technological support. Tangibility is the essence of the external constraints of Pearce &
Wiggins and the task constraints of Stokes. We therefore define tangible constraints as
Constraints and End User Development 7
properties in the virtual or physical environment that have defined qualities and limitations,
such as a painter’s palette or the range of a piano. Subject and goal constraints in Stokes’s
definition are, like Pearce & Wiggins’s stylistic constraints and Pérez y Pérez & Sharples
mental constraints, conceptual notions that have no defined form. We therefore define
conceptual constraints as those constraints that exist in the minds of practitioners, such as
aiming to create a painting in an impressionist style. Finally, Pearce & Wiggins’s notion of
internal constraints include a drumbeat that a collaborating musician should play to, or a
In addition to considering forms of constraint, the model considers that at some level
constraints are either malleable or static. In any creative process there will be supporting
tangible and conceptual structures that remain static and provide the basis from which
malleable constraints are built – for example a writer might only use the English language, or
a musician only a specific software environment. By considering the static constraints in any
given situation, we can differentiate and analyse the areas in which practitioners consciously
explore constraints, and the underlying structures used to frame this exploration.
Constraints and End User Development 8
Explanation of Process
ideation in the existing constraint structure. Each phase of ideation is informed by the current
conceptual and internal constraints, and bound by the enforced tangible constraints. Tangible
constraints and ideas are evaluated with reference to conceptual goal constraints, leading to
decisions on whether to retain an idea – in which case it becomes an internal constraint, and
also whether to maintain or transform current constraints. While constraints are maintained,
cycles of ideation and evaluation form the basis of the creative process. If a decision to
constraints modified where possible. A cycle of ideation and evaluation then tests the validity
For example, musicians may begin composing by choosing instruments, a scale, key
and / or a theme. Next they perform ideation within this space, for example through
exploratory play. They evaluate what they play with respect to aesthetic sensibilities and
Constraints and End User Development 9
conceptual goals, and use this to consider whether to retain played ideas in the composition,
and whether to modify or add to the constraints they have built. They may later find and
retain a specific musical phrase and build the rest of the composition around it, using the
Constraints form an important conceptual lens through which to view the design of
environments for creative tasks. Environments that are open to manipulation through end user
development mirror the creative mind by allowing users to question and explore boundaries
(Smith, 1993). The virtual can be far more malleable than the physical environment, as our
prototype study highlights. However limitations can form useful points of reference, setting
fixed structures within which creative work occurs and is understood. The computer
“enhances freedom for exploration, but also contains within it the potential tyranny of
continual choice” (Haworth, Coldwell, Gollifer, Kemp, Faure-Walker & Pengelly, 2005).
Conversely, well-designed software environments can aid the development and use of
constraints by promoting reflection and ensuring that constraints are satisfied. Candy notes
that a tool for scientific visualisation “presents the existing constraints, the human revises
them and the resources of both are employed in the process of considering and negotiating
The interface defines a tangible constraint structure that is to some degree comparable
interface define the scope of possible actions. Our theoretical understanding suggests that it
should aid users to be aware of the tangible constraints imposed upon them and be
Constraints and End User Development 10
empowered to apply and manipulate them where feasible. The structures provided to
manipulate the environment then define the static constraints on the user.
It is also common for environments for creative tasks to support the capture and
internal constraints inherent in the combination of ideas expected to form the outcome. As the
computer understands elements of user’s ideas in more detail, it becomes more of a partner in
the process by providing useful feedback and a powerful interface for manipulation, but this
requires a formal structure that enforces specific processes and restricts possible forms of
representation. Environments for musical composition in particular provide only a fixed set
of unambiguous representations through which users can define, evaluate and combine ideas,
allowing the computer to interpret them and produce music. In contrast, composition with
paper and standard instruments supports ambiguous representation and an ad-hoc process of
reassessment in light of two issues. Firstly computers structure creative processes to a far
greater extent than other tools and demand greater formality from users. Secondly, computers
can conceivably empower users to explore constraints, and to develop and exchange not only
ideas, but also structures for creativity. There is great value in the artist / technologist
collaboration (Edmonds, Weakley, Candy, Fell, Knott & Pauletto, 2005), as creative
these fruitful situations, can we free end users to develop and manipulate interfaces for
creativity without extensive technical expertise? The prototype study and analysis presented
here is used to explore this question and consider the value of this concept to users.
Constraints and End User Development 11
A prototype system was designed to ground our theoretical work and identify
design aims:
platform with which to create music through screen-based instruments, and hold some
appealing qualities as a musical interface: Interaction using a pen is less restrictive than with
a mouse or track pad, and the direct mapping between the elements displayed on screen and
presented. The three spaces are designed to support fluid movement between creating
instruments, playing music and developing compositions. In the build space shown in figure
3, users can develop their own instruments in order to manipulate tangible constraints. These
instruments can then be used to play and record musical ideas in the play space, shown in
figure 4. Recordings made in the play space can then be replayed, combined and manipulated
collaboratively in the composition space shown in figure 5. This supports the evaluation of
The underlying structure of the building interface focuses on the essential properties
(Hunt, Wanderley & Kirk, 2000). Instruments are developed by adding interaction shapes to
a free-form space, and defining how interactions with each shape (such as pressing, holding
or dragging) map to output values for sonic properties such as pitch, volume or voice. Users
link shapes to sound sources, which receive output values from the shapes, add these values
to their initial settings and produce sound. Figure 6 depicts shapes and sound sources linked
to form an instrument. To support informal representation, users can also draw anywhere on
Scaffolding
In designing a system through which users can create instruments, a trade off is
apparent: Complex environments usually involves a considerable learning process and rarely
support immediate engagement with ideas, while simplification enforces additional static
constraints on possible actions, for example by limiting the range of functions and options
available. We wished to reduce constraints without creating difficulties for new users. To
supporting environment for a student to learn through action, with structure provided but
removed as students develop the skills required to work independently (Wood & Wood,
1996). We consider this notion important to support for constraint development, aiming to
provide an initial level of constraint that is supportive of immediate action, but open to
The scaffolding in the prototype takes the form of seven template instruments,
provided to engage users immediately and to make visible some of the possibilities of
instrument building. These represent a cross-section of possible interaction styles and sounds,
Constraints and End User Development 16
including a drum kit, a piano and a Theremin style instrument where the pitch and volume
from both machines. This allows users to collaborate in a variety of ways, for example by
developing ideas individually then sharing them, or by jamming together and recording the
outcome. To support composition using these recordings, the prototype provides a shared,
networked space, based on the Sonic Sketchpad software developed and evaluated previously
(Coughlan & Johnson, 2006). This version employs the same notion of a library of recordings
and a paper-like space for composition where users can freely manipulate, arrange and link
recordings and add annotations, avoiding excess constraints on the representation of ideas.
After an initial pilot study and work to address the basic usability issues identified
through this, 12 participants took part in the main study in 6 collaborating pairs. Musical
experience was varied, in keeping with our notion of support for a wide range of users.
Constraints and End User Development 17
Participants included a pianist with 17 years of musical experience, a guitarist and a drummer
both having over 10 years of experience and a computer musician with 4 years of experience
using the software environments Cubase and Reason. At the other extreme, several of the
participants had only limited musical education, mainly from school lessons. Collaborators
were in some cases unknown to each other, but were generally friends who took part
together.
The studies took place in a usability lab, where screen-capture devices and cameras
provided a video record. The software logged actions to provide an accurate picture of user’s
interaction with Music Builder. Participants filled in questionnaires after each session and a
As one of our aims was to explore requirements for the design of systems to support
collaborative constraint development, we used the evaluations of the system to examine the
utility of shared visual and physical space to collaboration. Participants used the system
Participants could talk freely and hear output from the same set of speakers across the
conditions. A written tutorial was provided walking users through the features of the
environment. Participants were then asked to use the environment to produce a short
composition they were happy with under each condition, with the system wiped and restarted
Analysis
In analysing use of the prototype, we wanted to learn if, why and how musicians
would develop instruments in order to understand how interaction with these constraints
occurred. We looked for evidence supporting the validity of a user development approach
and issues relating to this, and the utility of the scaffolding and screen sharing. A further aim
was to relate our model to observations of interaction between creators and a software
environment.
Users spent between one and a half and three hours creating the three compositions.
In many cases users were still exploring new and interesting ways to interact with sound
throughout this time, and opinions of the instrument development concept were generally
very positive, particularly amongst participants with a previous interest in computer music.
Development of Constraints
conditions, while the remaining 4 used the template instruments throughout. As each
participant used the system 3 times, there were a total of 36 sessions. Each session was
• Multiple Instrument Edits – Users cycled between instrument development and play /
composition
• Single Instrument Builder – Users developed an instrument initially which they used
development of instruments, and also that development commonly occurs cyclically and in
tandem with composition. Instrument development occurred in the majority of sessions, and
in almost all of these cases instruments and musical ideas were developed in cycles.
Participants used the ability to develop tangible constraints that defined conceptual
instrument to allow only the notes in a musical phrase to be played. The ability to manipulate
goals. An experienced drummer linked shapes to multiple drum Sound Sources, enabling him
to play two or more drums with each tap. Through this he developed complex drum patterns,
Scaffolding
In 30 of the sessions the participant’s first action was to load a template instrument,
while in the remaining 6 sessions it was to build an instrument on a blank canvas. We learnt
through our observations that scaffolding fulfilled multiple user needs, providing a starting
point for instrument development, a basis for exploring the possibilities that exist in the
environment and an immediate ‘pick up and play’ structure. Users generally focused on
manipulating sonic constraints such as the range of notes or voices, rather than exploring the
various forms of interaction available, and the templates provided an existing structure for
interaction from which these sonic properties could be quickly modified and tested.
Constraints and End User Development 20
which a creative output can emerge. The design of each other’s instruments was a common
theme of the dialogue between collaborators. Instrument design not only affects the group’s
musical capabilities, it provides a platform through which roles are coordinated and
sharing tangible constraints was a successful strategy for coordinating collaborative play.
Extended discussions of tonal properties and instrument structure were common. In one
session a user asked if he could pass his instrument over to his collaborator. Unfortunately
this was not supported here, but could provide an interesting extension. In the short-term, the
which collaborators negotiate conceptual and tangible constraints for the intended
composition. Further to this, instruments could evolve as tangible constraint structures passed
describing the instrument as it is built and identifying a shared concept to realise, the
effectively constrained:
Three reasons for the success of this behaviour are suggested here: Firstly the
externalisation of constraints focuses effort on working within those constraints, rather than
developing ideas solely through conceptual notions which collaborators may define
predictable and provide a form of awareness information, narrowing the possible actions a
collaborator could take and increasing the likelihood of coherent collaborative play. Finally,
by restricting the instrument to notes that fit within a structure such as a scale, there is no
longer the need to use cognitive resources to evaluate whether the desired constraint is being
fulfilled, as it is impossible for it not to be so. These resources can instead be used to evaluate
The excerpt also shows the value of shared visibility as a basis for explanation of
concepts. Collaborative developers used the shared screen when provided as a means to
understand each other’s instruments, and questionnaire responses provided further evidence
for the value of being able to view and compare instrument development. Further support for
this process could focus on allowing collaborators to better ‘read’ their collaborator’s
Conclusions
environments for creative tasks. Supporting the development of tangible constraints has
Constraints and End User Development 22
structure within which successful collaborative play was significantly easier. When possible,
the movement from conceptual constraints towards tangible structures promotes the
tasks involve the realisation of ideas fulfilling a set of constraints that often begin as
ambiguous or intangible concepts. The user development of the environment can support the
Additionally, these systems can reduce the constraining influence of the designer on
the user by supporting the modification of the environment. There were multiple occasions
where users created instruments with properties that we as designers would never have
realising ideas.
Provide scaffolding to support the immediate use and exploration of the environment
Creativity support environments need structure, but structure that can be pulled apart
by users at will. However the consistent availability of initial structures to adopt and explore
provided inspiration and was preferred to building from a blank screen. The ‘tyranny of
choice’ arises through a lack of available structure, and making tangible constraints in the
environment flexible may compound these problems. This was not apparent with Music
Support the sharing of developed structures and the visibility of actions in constraint
development.
The shared screen afforded users the ability to see collaborator’s actions, aiding the
may be more appropriate. The possibilities for sharing a single space are interesting, but
structures between collaborators and across networks could provide a better model for user
user’s intentions, so the ability to read properties of each other’s instrument and gesture
Support fluid movement between ideation, tangible constraint development and evaluation.
Constraint development does not occur only as an initial stage of the creative process.
fluidly integrate the representation of ideas and of constraints and allow users to explore the
interdependence of the two. Music Builder provides an example of the rich design space for
References
Baer, J., (2003) Evaluative Thinking, Creativity, and Task Specificity, In M. A. Runco
Boden, M. A., (1991) The Creative Mind: Myths and Mechanisms. Basic Books.
Candy, L., (1997) Computers and Creativity Support: Knowledge, Visualisation and
Winter 1990, Association for the Advancement of Artificial Intelligence. pp. 59-73.
Csikszentmihalyi, M., (1996) Creativity: Flow and the Psychology of Discovery and
Edmonds, E. A., Weakley, A., Candy, L., Fell, M., Knott, R. & Pauletto, S., (2005) The
452-481.
Constraints and End User Development 25
Haworth, J., Coldwell, P., Gollifer, S., Kemp, T., Faure-Walker, J. & Pengelly, J., (2005)
Freedom and Constraint in the Creative Process in Digital Fine Art: An AHRB
Hunt A., Wanderley, M. M. & Kirk, R., (2000) Towards a Model for Instrument Mapping
Pérez y Pérez, R. & Sharples, M., (2001) MEXICA: A Computer Model of a Cognitive
(ed.), Watch What I Do: Programming By Demonstration. MIT Press. pp. 19-47
Springer Press.
Wood, D. & Wood, H., (1996) Vygotsky, Tutoring and Learning, Oxford Review of
Figure 1: Model
Constraints and End User Development 27