E.H. Ong, P. H. T. Janssen, T.T. Lo (eds.), Open Systems: Proceedings of the 18th International Confer-
ence of the Association of Computer-Aided Architectural Design Research in Asia CAADRIA 2013, 000–
000. © 2013, The Association for Computer-Aided Architectural Design Research in Asia (CAADRIA),
GROUP FORMING: NEGOTIATING DESIGN VIA WEB-
BASED INTERACTION AND COLLABORATION
ONG EU HO, FABIAN,1 PATRICK H.T. JANSSEN,2 and LO TIAN
1 Singapore University of Technology and Design, Singapore, Singa-
2 National University of Singapore, Singapore, Singapore
3 The Chinese University of Hong Kong, Hong Kong, Hong Kong
Abstract. This research project proposed to create spatial and com-
munal qualities of Group Form architecture via a web-based user par-
ticipation design method. The proposed method allows multiple users
to simultaneously design houses on the same site, encouraging
spatiotemporal negotiation as users interact and collaborate with one
another. In order to assess the feasibility of this approach, a prototype
of a web-based Group Form design tool was implemented using the
Processing environment. An experiment using the web-based tool was
conducted with the objective of exploring the actual user behaviour.
Keywords. User participation; group form; web-based tool; Process-
ing; collaborative design.
Online user participation in design is becoming more prevalent in today’s
world. Involving users in design can lead to a better fit between design and
different user needs and can result in a sense of ownership over the design
outcome. The Internet has become an invaluable tool to enable such user
participation as users can interact via the web from any location over ex-
tended periods of time. Furthermore multiple users could be logged in at the
same time, thereby creating opportunities for user!user interactions.
2 E.H. ONG, P.H.T. JANSSEN AND T.T. LO
Collective Form, or Group Form (Maki, 1965) architecture can be found
in medieval cities and traditional vernacular communities. They typically
emerged over a long period of time, from spatial negotiation between neigh-
bors as land was urbanized and as new buildings replaced the older ones.
This process can be seen as a bottom-up planning approach where a micro
single house unit is the generator of the macro village form.
Many architects today find the spatial relationship inherent in Group
Forms as having desirable qualities. In some cases, they attempted to create
these Group Form qualities in a top down manner though following certain
spatial rules. For example, in Japanese architect Koji Tsutsui’s Annular Or-
phanage in Uganda, Africa, clusters of housing facilities are formed by posi-
tioning rectangular units at a 45 degree angles to the adjacent units (Tsutsui,
2010). However, these attempts to create Group Form are missing certain
key elements of how Group Form emerges in a bottom up manner.
This research project proposed to create spatial and communal qualities
of Group Form architecture via a web-based user participation tool. The re-
search begins with a precedence study of existing web-based tools available.
Various types of user-design method are brainstormed and a prototype is
created based on the proposed system. Also, the concept of user-to-user
interaction via the web-based tool will be developed. Finally an experiment
is carried out to test the approach and the outcome will be discussed.
2. Existing Web-Based Tool
Researchers have developed a wide range of web-based tools that allow end-
users to design spaces and buildings. Three recent tools that allow end-users
to customise residential designs will be briefly reviewed: WIDE Kingdom,
Hinterland Koeln, and Bar_Code Housing System. Table 1 below gives an
overview of key features of each of these tools.
Table 1. Comparison between three precedence studies
tion of space
WIDE Kindom (2001)
Hinterland Koeln (2010)
Yes (Façade only)
Bar_Code Housing (2011)
Yes (2D only)
Proposed Web-based Tool
WIDE Kingdom (WIDE Kingdom, 2001) allows users to customise the
design of their apartment units. It employs information filtering as the key
mechanism for the customization of apartment units. Users can select the
GROUP FORMING: NEGOTIATING DESIGN VIA WEB-BASED INTERAC-
TION AND COLLABORATION 3
apartment unit, arrange the layout, try different interior finishes, and choose
furniture via drop-down lists and text-field. While the manipulation of space
is limited, the outcomes are rendered realistically in 3D for visualization.
Hinterland Koeln (Hinterland Koeln, 2010) is a tool that allows citizens
to build their houses within the courtyards of existing building plots. Users
can outline their building plot on the map by drawings a shape using the
mouse. Various parameters, including the elevation and floor height of the
building, can be adjusted via sliders and the changes are reflected simulta-
neously. These features make the tool very intuitive for the laymen.
Bar_Code Housing System (Bar_Code Housing System, 2011) is a build-
ing system for the construction of affordable mass housing blocks with in-
dustrialized process. A tool has been created that generates a floor plan
automatically based on a set of parameters input by the user, including the
number of inhabitants and programmatic requirements to have the system.
The tool is accompanied by a comprehensive web environment that facili-
tates communication and the collaboration amongst the stakeholders
involved in the building project.
All three tools allow users to create customised designs via a web-based
interface by specifying selections, parameters, and forms. However, it is
noted that in all three cases the interaction is solely between the user and the
tool. For a design method capable of generating Group Form, a tool would
be required that supports user-to-user interaction. More specifically, one of
the key elements of Group Form architecture is spatiotemporal negotiation
between people with multiple conflicting goals. A Group Form design
method is proposed that enables users to negotiate their design decisions via
online interaction and collaboration.
3. Proposed Group Form Design Method
The proposed Group Form design method focuses on residential design, and
involves both architects and end users in a structured process. The method
consists of three distinct stages: schema development, design development,
and detailed design.
• In the first stage, the architects would develop a set of spatial rules, referred
to as the design schema. These rules would be encoded into a user-friendly
web-based Group Form design tool that would allow non-experts to design
their own residential units.
• In the second stage, end-users would use the Group Form design tool to de-
velop their own designs. The design possibilities would be constrained by the
architect’s spatial rules embedded within the tool. Multiple users would work
4 E.H. ONG, P.H.T. JANSSEN AND T.T. LO
on their designs at the same time within a single site over a specified limited
period of time. Each participant would be able to see their neighbours design
emerging over time, and various mechanisms would be implemented to allow
multiple conflicting design decisions to be negotiated through online interac-
tion and collaboration.
• In the third stage, the architect would take over the final negotiated design
and develop it into a detailed design proposal.
4 Prototype Group Form Design Tool
A prototype Group Form design tools has been implemented as a web appli-
cation written in the Processing language (Processing 2010). The tool uses a
client-server architecture, where the server hosts all design data, and clients
are used to connect to the server and edit the designs.
The user works with the client application on their local computer, and
over a period of time develops their design. Each time they make any chan-
ges to their design, the modifications will be sent to the server and the cent-
ral model will be updated. If multiple users are logged in at the same time,
any changes made to their design will be reflected on other users screen al-
most instantaneously. In addition, each user will have an avatar, and will see
other user’s avatars walking around and designing their housing units.
In order to explore the concept of spatial negotiation, it was decided to al-
low users to compete with one another in defining the spatial location of
their house. Thus, rather than assigning each user to a predefined plot, they
were allowed to claim any part of the site as their own. Conflicts could then
be resolved either through consensus or through auction-style bidding. The
former method involves users communicating with one another through on-
line messaging. The latter method involves each user placing financial bids
on the area of the site in question, with the highest bid wining. In addition, a
complex spatial organisation was proposed, whereby individual housing
units could be built on top of each and interlock in non-standard configura-
tions. In order to ensure that no lifts were required, the height was limited to
4.1 SPATIAL RULES
In order to implement the tool, the first step was to define a set of spatial
schema rules that allow designs to vary widely in terms of their overall or-
ganisation, but that also constrain them to be viable. These rules were then
embedded in an easy to use software tool that would allow non-experts to
create designs for their own housing units.
GROUP FORMING: NEGOTIATING DESIGN VIA WEB-BASED INTERAC-
TION AND COLLABORATION 5
Initially, a range of alternative approaches were experimented with, in-
cluding free-hand drawing, component assembly, and cell selection. Free-
hand drawing has few constraints, and allows users to draw on a canvas.
Component selection presents the user with a set of components that can be
assembled in various ways to ‘build’ the housing unit. Cell selection discre-
tises space into a set of regular cells, and allows the user to define the vol-
ume of their housing unit by selecting cells in 3d space.
A hybrid approach was finally selected that combined cell selection with
a number of other techniques. In the first step, the volume of the site was
discretised into a cellular grid and users define the volume of their house by
selecting a set of cells. In the second step, the wall areas are divided into a
smaller cellular grid, and doors and windows are inserted by specifying their
The cellular grid was developed through a number of iterations. In the
earlier iteration, various grid sizes were experimented with. However, the
same grid has to be used to define all the spaces in the housing unit, which
makes it difficult to select an appropriate cell size. On the one hand, if the
cell size is too large, then the design moves became too coarse, with users
being forced to add unnecessarily large cells. On the other hand, if the cell
size is too small, then the design moves become too fine-grained, forcing the
user to painstakingly select numerous cells to define their housing unit.
After exploring various options, a specialised technique was developed
that allowed users to adjust both the position and size of the inhabited room
volume in the cell. The inhabited room volume was set to be smaller than the
size of the cell. The cell size was set to 8 x 8 meters, while the inhabited
room volume was set to 6 x 6 meters. Once the user had selected a cell, they
could then offset the inhabited room volume by pushing it into any of the
four cell corners. In a second step, they could further tweak the dimensions
of the inhabited room volume. The original 6 x 6 meter space could be re-
sized in steps of 0.5 meters, with a maximum of three steps. Hence, in plan
the area of each inhabited room volume could vary from 25 to 36 square me-
ters. This allowed users to easily define their housing unit by selecting a
small number of cells, while at the same time allowing users to have fine-
grained control over the dimensions of each inhabited room volume. Figure
1 shows the inhabited room volume being offset within the grid cell to the
four different positions and being resized to various different sizes.
6 E.H. ONG, P.H.T. JANSSEN AND T.T. LO
Figure 1. An inhabited room volume within a single cell offset to 4 positions (left) and chan-
ging size (right)
In order to ensure that reasonable housing units are generated, three key
of constraints are enforced: gravity, connectivity and size. The gravity con-
straint ensures that users are not able to select floating cells. They can only
select cells that are positioned either on the ground or on top of other cells.
The connectivity constraint ensures that users select cells that form one sin-
gle connected habitable room volume. When selecting the first cell, any non-
floating cell on the site can be selected. However, after that, they can only
select cells adjacent to (including above and below) the already selected
cells. The size constraint ensures that users cannot select beyond a certain
limited number of cells. There is a maximum of four floors, and the user can
select a maximum of four cells per floor.
Figure 2. Connecting cells with enforced constraints
Finally, once all the cells have been selected and the habitable room volumes
defined, the user can then progress onto defining more detailed components
of the design, including in particular the positions of the windows and doors.
The walls are divided into square ‘bricks’ measuring 0.5 x 0.5 meters. Win-
dows and doors can then be inserted into the walls by specifying the inser-
tion point in this wall grid. Figure 3 shows the process of adding windows
and doors to the walls.
Figure 3. Adding windows
Connecting Vertical Overlap
GROUP FORMING: NEGOTIATING DESIGN VIA WEB-BASED INTERAC-
TION AND COLLABORATION 7
In order to resolve conflicts of cell occupation, the tool allows users to
compete for a cell by bidding money. Each user would offer bids for the cell
and the user with the highest bid would eventually win the cell. Due to site
conditions, certain areas might have higher demand such as those with better
view. This approach will therefore allow the financial cost of the space vol-
ume (and land) to emerge in a bottom-up manner through individual negotia-
Figure 4.Resolving conflict with an auction style bidding system.
4.2 GROUP FORM TOOL
The final Group Form tool allows users to progress in a fluid and users-
friendly manner through the two design steps of defining the habitable room
volume by selecting cells, followed by inserting doors and windows.
Figure 5 shows screen shots of a user’s interaction with the Group Form
tool. For clarity, this scenario only contains a single user. The top three
screen captures show step 1, in which grid cells are selected and habitable
room volumes are defined. The bottom three screen captures show step 2, in
which windows and doors are inserted.
Although an underlying grid is used, the resulting arrangement of units in
plan and section do not suffer from monotony, as each user’s customisation
creates significant variability. Figure 6 shows an example of a plan gener-
ated using the schema rules described above.
8 E.H. ONG, P.H.T. JANSSEN AND T.T. LO
Figure 5. Screen captures of a single user working with the Group Form tool.
Figure 6. Floor plan generated using the spatial rules
4.3 GROUP FORM EXPERIMENT
In order to test the feasibility of the proposed Group Form design method, an
experiment was conducted. The objective of the experiment is to explore ac-
tual user behaviour when using the tool as the feedback is invaluable for fur-
ther improvement and implementation. The experiment was held in School
GROUP FORMING: NEGOTIATING DESIGN VIA WEB-BASED INTERAC-
TION AND COLLABORATION 9
of Design and Environment, National University of Singapore’s computer
Ten architecture students from fourth and fifth year participated.
Participants were each given a character card with a role they had to as-
sume. On the card, basic information such as name, age, gender, budget and
occupation was stated. Some characters were related. In addition, some of
the families were related to each other and therefore included preferences to
live closer together. Participants were encouraged to consider the site condi-
tion (a square plot of land with forest, river, carpark or road on each side),
the number of rooms required, their budget and land cost.
The experiment lasted for 40 minutes. After the experiment was com-
plete, participants were asked to write down their aims, whether they have
achieved their aims, and if they are satisfied with the outcome. They were
then asked for more general feedback on the Group Form tool and how the
prototype could be improved.
4.4 EXPERIMENT OUTCOME
The experiment resulted in a complex and spatially rich outcome. Users suc-
cessfully developed designs for housing units that performed reasonably well
from a functional point of view. From the point of view of developing a
Group Form, the process of online interaction and collaboration was varied,
and differed significantly from one user to the next. However, in general the
feedback was positive as users could see the potential of such as design ap-
Figure 7. From left to right: Worst-case outcome, experiment outcome, expected outcome
Based on the experiment, three key weaknesses were identified. First,
better tools are required for user-to-user communication, such as support for
private and group discussions and for the ability to annotate objects in space
with comments. In addition, the resolution of conflicts through auction-style
bidding required refinement.
10 E.H. ONG, P.H.T. JANSSEN AND T.T. LO
Second, additional constraints can be added to disallow design moves
that are clearly not desirable. For example, constraints could be added to
disallow one user to place a window directly opposite and other users win-
dow, thereby avoiding privacy problems.
Third, additional feedback can be added. For example, in the current sys-
tem, feedback on the land cost and construction cost are continuously dis-
play to the user. As these users are not expert designers, additional feedback
such as predicted energy consumption and daylight levels could help im-
prove the designs that are generated.
This research proposed a web-based design method for developing Group
Form that supports spatiotemporal negotiation through online interaction and
The method consists of three stages: in the first stage, architects encode a
set of spatial rules within a design tool; in the second stage, non-expert end
users use the tools to design their individual housing units; and in the final
third stage, architects take over the negotiated design and develop it further
into a detailed design proposal.
In order to demonstrate the feasibility of this method, a prototype of the
web-based Group Form tool was implemented. This tool allowed users to
design individual housing units within a shared virtual site, with real time
feedback. The experiment highlighted the potential of the proposed method.
Maki, Fumihiko: 1965, Some thoughts on collective form, Structure in art and in science,
Kepes, Gyorgy, ed., 116-127.
Koji Tsutsui: 2010, “11 Koji Tsutsui”, JA 77: Contextual Algorithm, 102.
Sheng-Fen Chien and Shen-Guan Shih: 2010, Design through Information Filtering, National
Taiwan University of Science and Technology
“Hinterland Köln,”: 2010. Available from: Youtube
<http://www.youtube.com/watch?v=ZAfw5HSzJto> (accessed 10 December 2010).
“Bar_Code Housing System,”; 2011. Available from: Bar_Code Housing System
<http://www.barcodehousing.net/> (accessed 1 February 2011)
“Processing,”; 2010. Available from: Processing <http://processing.org/> (accessed 20 Octo-
“Group Forming”; 2011. Available from: Vimeo < http://vimeo.com/26638743> (accessed 7