Abstract
This insight navigates the evolving landscape of modelling, examining its roots in the sciences and its recent emergence in the digital humanities. I contend that despite the digital humanities’ emphasis on the creative and subjective nature of data modelling, visual representations often succumb to adopting scientific graphical tools without due consideration of epistemological biases. Inspired by design practice, I propose a paradigm shift in critical modelling, spotlighting the pivotal role of visual representations as tools for reflective practice.
Although design models can serve as communication devices, in this insight I put an emphasis on their role as enablers of iterative development through reflective conversations between designer and model. By elucidating the qualities supporting reflective practice, this insight forges connections between theoretical claims and actual visual representations in the digital humanities, challenging conventional notions of data visualisation.
While modelling has been a fundamental part of the research process in the sciences, in the humanities it has only begun to gain currency with the advent of the digital and the establishment of digital humanities.1Hereafter referred to as DH. The increased interest in modelling in the digital humanities can be explained by the need for formalisation that arose when computational methods were incorporated into the scholarly process. McCarty speaks of ‘computational tractability’, which forces ‘us to confront the radical difference between what we know and what we can specify computationally’ Full reference in Zotero Library. In the early days, digital humanities modelling was mainly concerned with the creation of data models in order to make humanistic artefacts accessible for computation. In recent years, there has been much debate about how to conceptualise modelling in the DH and about its epistemological premises compared to other disciplines. Although perspectives on modelling between the digital humanities and the natural sciences diverge considerably, there are some commonalities. Generally, it can be said that the terms “model” and “modelling” have a long history of defying definition (cf. Full reference in Zotero Library). However, most disciplines, even with very different epistemological foundations, can agree on certain characteristics of models, namely that models are representative of something (an object or part of an object under investigation) and also for something, so they are created with a certain function in mind (cf. Full reference in Zotero Library, Full reference in Zotero Library, Full reference in Zotero Library, Full reference in Zotero Library).
The difference lies in the way that representation is understood with regard to modelling. While natural scientists often employ models in order to represent certain aspects of an object or system that they see as fixed (by the data produced in an experimental setup, for example), literature on modelling in the DH emphasises the role individual scholars and their viewpoints play in the creation of data:
Furthermore, most digital humanities researchers assume that data modelling is primarily a constructive and creative process and that the functions of the digital surrogate determine what aspects have to be modelled. Full reference in Zotero Library
In a similar vein, Ciula et al. mention subjectivity as part of the modelling process:
[…] in the humanities and cultural heritage modelling targets tend to be human creations and the goal of the modelling is often to describe idiosyncratic phenomena or artefacts of human creation, acknowledging and valuing subjectivity as part of the modelling process. Full reference in Zotero Library
Nonetheless, if we look at the practice of modelling in the digital humanities, we can observe that there is a divide between claims made in the literature and the presentation of modelling results as visual representations. For example, visualisations of social networks of actors in fictional literature often fall back on simple node-link diagrams from the sciences. This is a problem that has been expounded by Johanna Drucker in particular:
As digital visualization tools have become more ubiquitous, humanists have adopted many applications such as GIS mapping, graphs, and charts for statistical display that were developed in other disciplines. But, I will argue, such graphical tools are a kind of intellectual Trojan horse, a vehicle through which assumptions about what constitutes information swarm with potent force. These assumptions are cloaked in a rhetoric taken wholesale from the techniques of the empirical sciences that conceals their epistemological biases under a guise of familiarity. So naturalized are the google maps and bar charts generated from spread sheets that they pass as unquestioned representations of ‘what is’. Full reference in Zotero Library
Drucker as well as Katherine Bode in her insight on ‘Critical Modelling and Measurement’ argue for so-called non-representational modelling. The attribution might be misleading. It is not about modelling that is not representational at all; rather, modelling that is not based on pre-existing data but performative in the sense that it creates data and new knowledge. As such, it is representative of what is yet to be modelled. I strongly agree that we need to think about visual representations that are more suitable for humanistic epistemology. However, I would like to suggest an alternative approach that I think can be beneficial here: design practice.
Models and modelling are an important part of any design process, but there is a fundamental difference in terms of what is modelled: most designers can agree on the notion that modelling describes a process of bringing something to life that does not exist yet, be it an architectural model, a model of a chair, wireframes for a planned software application or a diagram of a service design concept. A model in a design process usually visualises one of many possible representations of an idea. It serves two purposes:
- Communicating an idea to others (colleagues, stakeholders, clients)
- “Conversing” with the model in an iterative way to develop new ideas
The first point is easy to understand. When working together with other designers and for clients, design ideas have to be communicated to provide updates on progress and seek a client’s approval.
The second point requires more explanation. How does one converse with a model? In the next section, I’d like to elaborate on what such a ‘reflective practice’ Full reference in Zotero Library in design looks like and how certain models in the design process (not all) exhibit particular attributes that support this conversation.
In the last section, I draw connections to the digital humanities and show that there are some similarities in perspectives on modelling, but less investigation of visual representations as part of a process of modelling. By making the qualities of visual representations in reflective practice explicit, I hope to contribute a different perspective on visual representations that emphasises their role as devices for modelling, in contrast to just being representations of static data.
Modelling in design practice
The ambiguity of the term modelling in the humanities and sciences is echoed by the use of the term in design practice. In all likelihood, as Willard McCarty argues, this has something to do with the fact that modelling develops out of practice and is ‘rooted in stubbornly tacit knowledge’ Full reference in Zotero Library. In architecture and industrial design, the term “model” usually refers to physical or virtual representations of buildings or products (or parts of them respectively). The purpose of a model can either be to show a finished design at the end of a design process or explore different design options during the process. For my argument, the latter purpose is of particular interest. Here, I would also like to highlight that other visual representations such as sketches, drawings, prototypes and mock-ups are not called models customarily, even though they accord well with the aforementioned definition, whereas a “model” is of something and for something. For this reason, I will treat these exploratory visual representations as models here.
For this discussion on modelling in design, I would like to focus on visual representations that have an epistemological value in the design process, i.e., that can be seen as epistemic things (cf. Full reference in Zotero Library). According to Karin Knorr-Cetina, such objects are characterised by a ‘lack in completeness of being’ and their tendency ‘to unfold indefinitely’. Knorr-Cetina specifies further:
[T]hey continually acquire new properties and change the ones they have. But this also means that objects of knowledge can never be fully attained, that they are, if you wish, never quite themselves. Full reference in Zotero Library
But how are these characteristics expressed in a visual representation? A look at the reflective practice of designers might prove illuminating.
American Philosopher Donald Schön looked at the practical work in different disciplines, among them architectural design. Based on an example in a design studio in a school of architecture, he describes a setting of reflective practice or ‘reflection-in-action’ taking place between a student and teacher. The student explains the problems she is facing with the development of her design concept for an elementary school. As Schön describes it, in a constant oscillation between drawing and talking, the teacher reframes the problem the student has previously set because he deems the framing inadequate. Through drawing, the teacher conducts a ‘web of moves’ that together form ‘experiments’ that lead to unintended consequences and shifts in meaning. The result is a new understanding of the situation and the potential for additional reconceptualisations. Schön speaks of a situation that ‘talks back’ (cf. Full reference in Zotero Library). New consequences that earlier actions produced are judged based on
whether they [the practitioners] can solve the problem they have set; whether they value what they get when they solve it […], whether they achieve in the situation a coherence of artifact and idea, a congruence with their fundamental theories and values; whether they can keep inquiry moving. Full reference in Zotero Library
Visual representation plays a central role in this reflection-in-action process. Schön calls it a ‘virtual world, a constructed representation of the real world of practice’. According to him, reflective practice is dependent on this virtuality, which reduces constraints present in the real world. It is ‘rapid and spontaneous’, but also allows you to omit features from the real world and focus on others. The virtuality leads to a situation, where ‘[n]o move is irreversible’ Full reference in Zotero Library. This sounds like modelling, indeed.
Similar to Schön, Gabriela Goldschmidt describes this practice as ‘dialectics of sketching’ based on a study of sketching at the MIT in Boston Full reference in Zotero Library. Both use the term ‘seeing as’ (or ‘doing as’) (Full reference in Zotero Library, Full reference in Zotero Library) to describe the aforementioned reframing of a situation. Expert designers draw on their past experience to create frames, which they try to adapt to the situation at hand, i.e. they change the visual representation according to a frame and assess the consequences, as described above. The frame they have chosen determines the problem, but also suggests a certain strategy for approaching a solution.
Fish and Scrivener Full reference in Zotero Library elaborate on the interplay between external representations and mental images, and stress the particular role sketches play, in as far as they are ‘records of a sequence of attentive acts combining information from our eyes with images generated from memory’. The conversational quality of sketching is further exemplified by the following statement:
Therefore, when artists sketch, rather than draw, modification of the percept by mental manipulation is always involved. In other words, the visible marks made will generate mental images that in turn may influence the sketch. Full reference in Zotero Library
Fish and Scrivener’s use of the term ‘percept’ here refers to the mental representation of the sketch. They describe three characteristics of sketches that are the basis for this interplay (cf. Full reference in Zotero Library):
- Indeterminacy or ambiguity
- Use of abbreviated sign systems to represent three-dimensional experience
- Use of selective and fragmentary information
This short excursion into reflective practice gives us an idea of the qualities a visual representation can offer to make a process of reflective practice possible.
Reflective modelling in digital humanities
How do these aforementioned forms of modelling relate to data modelling in the digital humanities, when we consider that models of data in digital humanities usually appear as diagrams of a finished model (if at all) and not as sketches, whereas sketches in design practice are commonly used to represent three-dimensional experience?
There are certainly areas of DH practice where representing three-dimensional artifacts or phenomena is relevant, but in data modelling we usually find ourselves at a higher level of abstraction, where representations are predominantly diagrammatical not pictorial. A sketch, on the other hand, typically combines pictorial and diagrammatic elements. Nevertheless, since reflective practice is, for the main part, about entering into conversation with the situation, any type of visual representation that supports this conversational aspect would seem to be an enhancement to the modelling process.
Based on the deliberations in the previous section, I have identified five qualities of visual representations in reflective practice that suggest the benefit of designing and using visual representations for reflective and thus critical modelling in the DH.
Indeterminate
Visual representations in reflective practice exhibit a degree of indeterminacy. On the one hand, this allows a practitioner to fuse external representations with their mental representations, so that the phenomenon to be modelled can be reframed iteratively until a coherent framing is achieved that is in alignment with the practitioner’s individual perspective. In this process, creativity is triggered by the indeterminacy, leaving gaps to be filled by the practitioner’s imagination. On the other hand, indeterminacy also conveys the processuality of the representation: every visual representation is just one of many possible ways to model the phenomenon. And all of these can be documented in the digital quite easily. This appears to be much closer to humanistic epistemology. In a sense, the indeterminacy, the gap, the ‘Leerstelle’ (Full reference in Zotero Library, cited in Full reference in Zotero Library) as Wolfgang Iser calls it, can be understood as an affordance that waits to be filled with meaning.
Immediate
Every change applied to visual representations shows an immediate effect and its consequences. This is what Schön described as a situation ‘talking back’. In the context of modelling, this would mean that every change in the visual representation reveals its effect on the data instantly. Connecting data or data samples would make it even easier to experiment with certain models and immediately assess their value with respect to a research hypothesis.
Reversible
Every action undertaken has the form of an experiment that can be evaluated by the practitioner and reversed if it obstructs/inhibits conversation with the situation or leads to undesirable side effects in the data model.
Selective
Visual representations in reflective practice focus on parts of situations that exemplify an important aspect to be modelled. In the context of sketching, Fish and Scrivener Full reference in Zotero Library mention an example of a sketched footprint. Its shape is represented solely with the curvature of a line. This is just enough information for the viewer to decipher it. There is no need for additional information.
Argumentative
In the process of conversing with the situation, the consequences of individual actions are evaluated by the practitioner and either affirmed or discarded. The way the situation “talks back” can be considered as a process of argumentation that the practitioner can review. The argumentative nature of the process is therefore of value to the practitioner firstly, during their reflective practice, and secondly, when presenting one’s own process within their field of study. In a similar fashion, Soethaert advocates for viewing models as ‘argumentative scheme[s]’ that allow a mode of ‘modelling as’. This is in congruence with the act of reframing, which Schön also describes as a sequence of ‘seeing-as’ and ‘doing-as’ Full reference in Zotero Library. Thus, by viewing a situation in a certain light and comparing it to a similar situation they encountered before, they can also apply a similar strategy of transforming the situation and listening for what the situation has to say to them. The situation is ‘transactional’, however; practitioners not only shape a situation, but their models and experiences are shaped by the situation as well Full reference in Zotero Library.
Of course, some of these qualities are already described as being characteristic of modelling in the sciences and the humanities. For example, Stachowiak describes the selectiveness of models with his ‘Verkürzungsmerkmal’ Full reference in Zotero Library, whilst Lattmann Full reference in Zotero Library stresses the iterative and creative nature of the modelling process. Creativity is also seen as a characteristic by Ciula and Eide Full reference in Zotero Library and Jannidis and Flanders Full reference in Zotero Library. And in his closing keynote of the DHd2020,2Conference DHd2020: Spielräume – Digital Humanities zwischen Modellierung und Interpretation (7. Jahrestagung des Verbands Digital Humanities im deutschsprachigen Raum). Alan Liu stressed the importance of making protocols, procedures and processes explicit, which he identified as one of the important tasks of Digital Humanities Full reference in Zotero Library.
But what is usually left out of the discussion is how to achieve these qualities in modelling. For example, can creative engagement with a model be encouraged? How do we achieve that? Looking at the reflective practice of design work might open up new pathways to conceptualise visualisation in the DH and satisfy the high expectations with respect to modelling practice.