Introduction

Design, in general, has to solve ill-defined, or "wicked", problems, the understanding of which is concomitant to the act of their resolution. In the words of Rittel and Webber, “the information needed to understand the problem depends on one’s idea of solving it” (Crowley and Head, 2017; Rith and Dubberly, 2007; Rittel and Webber, 1973). As such, the act of design does not have a definitive stopping rule, and neither can its output be judged by a binary evaluation of good vs. bad. Most importantly, the stakeholders involved in the design process do not necessarily have a set of shared values. Essentially, design can be seen as an iterative act that aims to reduce uncertainty at an ontological level (Hanna, 2014) by simultaneously searching for the appropriate problem representation and resolution.

The main goal of this project, as set out in the original InnoChain call (Work Package 3: Design Communication, ESR05: Alternate Means To Communicate Measure), is to analyse how complex digitally based design can be communicated and collated internally, within a design team, and externally, with the various stakeholders involved in the design process. Literature establishes that communication and dialogue are the base for solving wicked problems, as through these means shared understanding can be constructed amongst the actors involved in the (design) process (Dawes et al., 2009; Lawson, 2005; Roberts, 2000; Conklin, 2005; Walz et al., 1993; Bechky, 2003). In other words, shared understanding can be construed as a set of matching ontological representations of meaning that gradually emerge through a process of conceptual displacement (Koestler, 2014; Schön, 2011, 1991). Communication and dialogue are increasingly reliant on digital means, nevertheless they cannot be fully analysed through the lens of a purely technical model. As such, in the context of this thesis, communication is understood as a transactional phenomenon that has both a technical or mechanistic manifestation (digital) and, as well, an intrinsic psychosocial component. The former approach corresponds to the Shannon-Weaver model of communication (Shannon and Weaver, 1963, 1948), whereas the latter draws from inferential models developed in the philosophy of natural language communication (Grice, 1991; Sperber, 1995; Wilson and Sperber, 2008).

Currently, the design process exhibits a certain friction in its communicative processes that is perceived by the stakeholders in the Architecture, Engineering and Construction (AEC) industry as holding back the application of the full potential of new digital technologies, contrary to the developments in other domains of the global economy (Barbosa et al., 2017). In part, this research is based on the hypothesis that this friction results from the technical limitations of the software tools, standards and methods in use. Nevertheless, one cannot dismiss the human aspect involved in the communication process. Particularly, this unfolds into issues around authorship translated into disciplinary and professional boundaries (Giddens, 1991; Abbott, 1992), and materialises into issues of trust stemming from disembodied data (Tsoukas, 1995). Disembodied data, as discussed at length in Section 2.2.3, is the result of separating information from meaning, or intention. To this extent, the investigation will not be limited to purely software aspects of the problem, but out of the necessity to offer a complete picture, will include the psycho-social context in which digital communication unfolds, is shaped by, and simultaneously shapes.

The design process, as currently unfolding in the contemporary societal milieu—coupled with the design problems' complexity and scale—requires a large array of stakeholders, from both technical backgrounds as well as from outside specialist disciplines, to engage in a continuous (but to a certain extent temporally bounded) act of communication and creation in order to complete a given design assignment. These networks are seen as the basis for resolving wicked problems (Weber and Khademian, 2008). Nevertheless, at the beginning of a design project, one cannot assume the existence of a defined ontological base upon which stakeholders may interact. Furthermore, one cannot fully know in advance who the stakeholders will be and in what parts and capacities they will be involved in the resolution of the problem—the communicative topology of the network, as well as its informational needs, is not fully known. As such, communication in design cannot rely fully on known data representation and classification patterns, and neither on predefined interaction networks.

Consequently, this research project aims to investigate critical aspects of design communication, whereby communication is understood as having both technical and social dimensions that reinforce each other (Garfinkel and Rawls, 2006). Following that, at the beginning of the design process, one is bounded by an incomplete definition of the given problem and by the fact that the relationships between actors need to first emerge before subsequent shared ontologies are defined. Three main research directions have been selected: (1) data representation and (2) data classification juxtapose different approaches to ontological models of design objects and, finally, (3) data transaction looks into the mechanisms of interactional exchange and how they change the nature of design communication.

Data representation

Currently, design data exchange is dominated by the Industry Foundation Classes (IFC) object model, which positions itself as the official interoperability standard of the AEC industry, and has been in active development since 1994 (Froese, 2003; Plume and Mitchell, 2007; van Berlo et al., 2012). It is widely used throughout all stages of the design and build process and underpins most, if not all, Building Information Modelling (BIM) techniques, regulations and practices. Similar to Cyc, an artificial intelligence project, started in 1984 and still ongoing, that attempts to assemble a complete ontology spanning the basics of “how the world works” (Lenat et al., 1985; Lenat and Guha, 1991), IFC classes aim to specify a complete vocabulary and grammar of all elements (both high level—walls, doors, etc.—as well as low level—points, lines, etc.) that are used in architectural design, engineering and construction (Hamil, 1994).

Nevertheless, many large engineering and design companies have been developing in-house alternative design data exchange standards positioned as internal replacements for the official IFC standard, rather than as additions or extensions, to better suit their own internal needs. These developments result from the problematic nature of IFC, whose scope, while vast, is limited, and extensions of it are not easily shared: similarly to Cyc, it is bound to be (forever) incomplete and difficult to evolve (Bertino et al., 2001). These ad-hoc informed standards have scopes that vary from a single-project base to a company-wide norm. To a certain extent, the emergence of such standards results from the need to further articulate domain- and organisation-specific knowledge that accumulates over time: most observed examples underpin computational techniques, workflows and methodologies that quintessentially represent a given organisation market advantage, or a given domain’s internal language constructs that allow for its effective functioning (Giddens, 1991; Kuhn, 1996).

Regarding the representation of information, modern computer science techniques allow for transparent and reflective approaches to constructing higher level standards. Based on self-describing structures (i.e., XML or JSON), only a low level, primitive standard is enforced thus allowing for the bottom-up emergence of expressive, ad-hoc defined object models whose definition, interpretation and validation are relegated to the communicants, rather than the transmission medium itself. The simultaneous readability of data by both machine and human allows for the development of rigorous informational exchanges and applications that consume said data.

The emergence of multiple deregulated project- or organisation-based standards hints at a larger question than IFC’s erosion of the one standard approach. To a certain extent, ontological uncertainty has been embraced and enabled by computer science through the development and increasing adoption of self-describing structures for object representation. Within AEC, it is not yet known what is the best model of data representation that can serve communication and collaboration needs.

Q1: Consequently, one of the central research aims of this thesis is to identify the limitations (and advantages) of a schema agnostic, self-describing and composable object model compared to an ontologically complete “one standard” approach.

Data Classification

Following the questioning of data representation, this research project will look at existing methods of data collation and classification pertaining to existing digital collaboration practices in design. Currently, AEC software is reliant on monolithic data structures that may be traced to the implementation choices of the data representation standards. BIM practices currently advocate a collaborative approach centred around the idea of “one model”, or one single source of truth, that contains, synchronises and validates the information produced by all technical stakeholders involved in the design process. The “one model” approach is encouraged by the mandatory inter-relation of building objects which is prescribed by the data representation standard in use or by the architecture of the software used. For example, as of the latest IFC release, “round-tripping“, or sending and receiving of model information, is stated to be out of scope and, in general, not supported (buildingSmart, 2016a, 2016b, 2016c; Bentley Systems, 2008; Thein, 2015).

Nevertheless, when confronted with the reality of communication in a multidisciplinary design process, the monolithic model prevents natural dialogue to emerge due to its high processing cost when used as a communication vehicle. This problem is mostly visible at the early design stage but is not limited to any particular point in time of the design process. The symptoms of this problem, revealed through industry exchanges undertaken within the research project, manifest as the proliferation of “residual models” (Scheurer and Stehling, 2011), i.e. files containing partial models with selective information that responds to various stakeholders’  specific needs at one point in time. The emergence of lean data exchange workflows can be attributed to the fact that communication tends towards an optimally relevant state (Sperber, 1995; Wilson and Sperber, 2008) in a specific given context. Relevance theory, based on Grice’s inferential model of communication, suggests that communication effort is minimised through the reliance on interpretation and “expansion of meaning” at the receiver end (Grice, 1991; Lindblom, 2001), and through contextually-based informational enrichment mechanisms (Sperber, 1995; Wilson and Sperber, 2008).

Q2: Consequently, existing standards of data classification in digital design need to be questioned. Specifically, the investigation will focus on what extent design data can be compartmentalised but still maintain logical coherence through an analysis of the advantages and limitations of an object-centred classification approach to digital design data as opposed to a file-centric collaboration methodology.

Data Transaction

Data is not the only ingredient necessary for a communication contract to be enabled at a social level, and data alone does not suffice to allow for the emergence of shared understanding (Aish, 2014; Aish and Fleming, 1977; Bechky, 2003; Strathern, 2000; Tsoukas, 1997). As such, the following investigation focuses on another critical aspect of digital design communication, namely its transactional nature: specifically, how data is exchanged. In this regard, literature establishes nextness and sequentiality as the key variables describing communication contracts (Lindblom, 2001; Sacks et al., 1974). Nextness, or adjacency, ties participants into a cursive dialogue by requiring that informational transactions happen within a certain time (and potentially spatial) limit. The latter, sequentiality, looks at the order of in which communicative transactions occur and is especially important in an asynchronous digital environment.

Existing BIM collaboration workflows, due to the nature of existing standards for data classification and representation as well as current software limitations, do not lend themselves towards spontaneous transactions with actual design data. Thus, within the digital design process, spontaneous dialogue is delegated to classical means of communication such as chat applications (i.e., Slack), email and attachments composed of residual models, screenshots, etc. and can be seen as introducing friction by not being intimately tied with design data.

Conversely, literature reveals a large number of custom workflows being developed to serve the interoperability needs of large-scale design projects (Bhoosan, 2017; Deutsch, 2017; Goldup et al., 2017; Heumann and Mullenix, 2014). Many of these workflows show a common effort to achieve responsiveness, or nextness. They achieve this mostly through unpacking design information into its most relevant parts for the task at hand (i.e., centrelines and section for a beam element), thus minimising the amount of information that is transacted.

Q3: Complimentary to data representation and classification, data transaction requirements for digital design communication are not yet fully understood or supported. Consequently, this research project looks at what are the advantages and limitations of an object-based collaboration methodology as opposed to an file-based one in regards with enabling the key requirements of a communicative contract, namely nextness and sequentiality.

Methodology and Research Instrumentation

In order to be able to answer the questions outlined above, this research project necessitated software design development so as to be able to counterpose existing approaches to a feasible alternative. This resulted in the software platform named Speckle, which served as base research instrument throughout this project.
Firstly, the Speckle platform can be described as being schema agnostic: it does not have a standardised ontology per se, but rather a small set of user-defined, composable object models that can be swapped in and out, and used in together. Secondly, Speckle is object-centric, as opposed to file-centric: instead of persisting data in monolithic blocks, it stores each object individually (and immutably) and allows for overlapping groupings thereof. Third, Speckle is embodying data: as opposed to existing approaches, where files are just “shared” and there is no overview of who is consuming the information, and to what effect, Speckle traces the communicative network and aims to inform end-users of their transactions and their implications.

Speckle enabled a theoretical, technical and applied analysis into the three research questions outlined above. The collaborative network within which this research project was undertaken, InnoChain, consisted of both academic and industrial partners. This network served as the seed for a living laboratory which grew throughout the project to include many other industry participants. As such, this setting served both as the basis for the technical development and testing of Speckle, as well as the pool from which the quantitative and qualitative data required to answer the questions outlined above was gathered from.

Followingly, in the Literature Review chapter, the key preliminary issues raised by the research questions will be tackled. Broadly put, the analysis will examine the act of design as a communicative act. Specifically, it investigates the causes behind the divergence of disciplines and how inter-professional communication unfolds, as well as how disciplines re-converge as dynamic networks of stakeholders when tackling wicked (design) problems. Further, an analysis of existing communication models – both technical and social – frames, when coupled with the state of the art in digital design communication, the theoretical context in which the counterpoint to existing approaches to digital collaboration in AEC was formed and assessed.