CDP: Multiple Perspectives on Computational Graphic Design

This discussion is based on six interviews, which took place in 2017 (five years ago from today). They were edited for clarity and additional commentary was inserted for emphasis.

Embracing code in the design process encourages designers to consider time and interaction in creative ways as they construct algorithmic and parametric relationships between data and form. In this way, designers can perceive forms as variations generated out of algorithms When building computational systems, designers craft a flexible visual language that dynamically changes with a person’s input or complex data, morphing one set of forms into another, and constantly making them evolve over time. Beyond a static visual representation, the process of creating this autonomous, permutable, and improvisible visual language emphasizes the impact that time and interaction can have upon visual patterns. Through responses from multiple contributors working adjacent to computational graphic design, this CDP discussion shares their comprehension, excitement, and apprehension about computational design.

I am curious how computation has shifted your design process? In particular, how has computation informed you to create generative systems that yield variable artifacts? Could you describe your process for selecting parameters, creating the relationships between them, and defining their ranges?

Each visual parameter we define determines the visual boundaries of the ‘playground’(eg. narrow or wide color palette, fast or slow movement, ordered or chaotic behavior). This becomes interesting when the parameters are pushed to the extremes–between ultra minimalism and ultra maximalism–this is where surprises emerge which no human could have invented. With commercial generative branding projects, the parameters are defined by the brand guidelines–to ensure any iteration of the design strikes a balance between expression and brand coherence.

To me, this is a conceptually driven process of designing. You have an idea for a project that hopefully manifests as a set of specific instructions that when implemented (by a machine or human) design the project themselves, in a sense. That explanation really oversimplifies the process, but in some respects having a reduced and specific set of restrictions is what makes the process generative and context-specific. If there is too much freedom then you can do anything and as a result, nothing happens. Since I don’t have programming skills, in my early work I would create a set of instructions and manually implement them, trying to programmatically execute many iterations without worrying about the outcome, as a machine would do it. I think a big part of this process is letting go of critical judgment while making and seriously considering all possible outcomes, even if they are ugly, boring, or incomprehensible. Critically analyzing the outcomes is the next step and, in many cases, I discover new ideas through the making process that were not part of the initial concept.

Generative parameters, like any design asset, need to have some relative conceptual justification. Once a generative system has the right ingredients plugged in for a project, they become amazing tools to execute very dynamic work across all platforms. We can create limitless variations of images and kinetic content that transfer them into context quickly. At this point it still boils down to the designer’s eye, taste, and intuition to select the moments that feel the best.


Computation has not so much determined how the variations are produced but how many of them there are. In developing a design we like to produce a lot, testing out small differences. Decisions are in the end based on conceptual clarity and formal consistency. Generative design methods have definitely informed my practice. For example, much of my work is either black and white or multicolored because I find picking a specific color from a palette of 16.7 million possibilities, particularly challenging. I prefer automated systems to do that work for me. However, I am also a strong believer of reviewing digital output. Any piece of software is the result of a thinking process that happened in the brain of its maker at the time of the implementation. So there is a natural bias in every outcome produced by a computer. Assessing these outcomes will become especially relevant in the context of artificial intelligence and algorithmic accountability.

Working with data allows us to endlessly iterate on the design process, since it gives us the ability to resort and parameterize our design. When we design a visual algorithm or analyze a dataset, we have no particular form in mind yet. It is the process of visualizing that shape and every variation on that form. This sounds uncontrolled or even random, but we tend to think of it as shaping a design process whilst embracing serendipity.

What might be some (new) types of graphic design artifacts enabled or informed by computation? It’s hard to predict! It seems technology evolves so quickly and is so widely available, that I cannot possibly imagine what someone is doing right now in some lab or even in their garage. Perhaps some of this work suggests a time where computers can learn and make decisions and designers are not necessary, which I do not think is a preferable future. Posters that design themselves, graphic identities which evolve on their own, web to print capabilities where the translation between mediums are automated…all these things are already happening! I’m simultaneously fascinated and weary of the affordances that the computer offers. Computation is the process of creating work which will result in a more complex and dynamic visual form. The artifacts will always be contingent on what formats we have available for our design output. We will see more and more specialized design applications. Brand identity guide books will eventually become obsolete, to be replaced by generative toolkits. If a software application can be defined as a design artifact then yes. Otherwise, the only change is that the aesthetic output in existing and future design formats will represent a computational process, the formats or artifacts themselves will only enable the software.

Computation in graphic design is especially useful when dealing with complex data. I mean complex data in terms of (1) Quantity–they will be especially large datasets; (2) Speed–data will become available as soon as it is collected; and (3) Difficulty–the data will become increasingly hard to understand. The graphic design artifacts that will be enabled by computation will on the one hand be screen-based artifacts that use live data such as maps and news apps. On the other hand, computation can be used in those artifacts that step aside from the data flow to reflect on our lives like visualizations of complex data in the context of cartography and science.

We are now seeing the emergence of self-improving design based on user feedback. Once the work is released into the public realm, it can evolve to adapt to the tastes and needs of the user.

Let me focus on artifacts understood as undesired byproducts since there is a very interesting property of computation that I believe graphic design is unfortunately inheriting. By means of Moore’s law and the resulting growth of storage capacities, computation creates an ever-increasing waste heap of data that we will eventually need to learn to dispose of. As stated by Bruce Schneier, cryptographer, and computer security specialist, “data is the pollution problem of the information age”. Until now, it seems that the world is still deliberately ignoring this problem of data pollution, much like the effects of climate change. But even though data is virtual by nature, we will inevitably reach the point where digital concerns hit physical constraints. Obviously, there is an urgent need to scrutinize the environmental and psychological impacts of our ever-accumulating software and data waste.

Ironically, we feel that the word “artifact” is in fact becoming more and more obsolete as a way of describing the output of contemporary graphic design projects. To us, contemporary graphic design outputs are much more fluid than the typical and historic graphic design artifacts like a book or a poster. Their existence is verified only by the audience interacting with them, which gives them a form of agency. They exercise some influence on their surroundings and their surroundings in turn influence them. They are much more dynamic, and therefore the finality of the traditional graphic design artifact is something we do not recognize in these outputs. Also, as we live in the age of open-source code sharing, we see that design outputs have gained the ability to evolve with and respond to the medium that is carrying the design. Design outputs nowadays can be preserved, conserved, maintained, rediscovered, reenacted, remixed, and redesigned by the community at large. In a way, you could say the artifacts become more and more adaptable and resilient to the negative influence of time and trends. Maybe there will no longer be single and final graphic design outputs in the near future.

I have two questions. Using code, designers can instantly achieve myriad permutations of visual artifacts by tweaking variables. Beyond the quantity, the outcomes of these computational systems are rendered in real-time which means they can be reflective of various and dynamic contexts at the moment. I am curious what kind of impact might ‘real-time’ and ‘data-driven’ aspects have on your practice? What is your opinion on graphic designers authoring ‘tools or systems’ that generate variable artifacts?

Graphic design is moving away from ‘form giving’ to the production of tools and concepts to transform data into visual information. The consequence of this is that graphic design is moving away from the end stage of a process to a step prior to that. Others – non-designers, users, etc. – will generate forms using the tools developed by graphic designers.

The exciting thing would be that graphic languages could be in constant flux, almost alive in a sense. Every time you see something it would be in a slightly different state. This seems to imply that more design will be viewed or experienced digitally, which would allow a viewer to perceive incremental shifts as they happen. I think of Dexter Sinister’s Kadist Identity, a 10-year identity program based on Donald Knuth’s Metafont, where the form of the typeface shifts on a daily basis. I see this idea being used in an interesting way by architects, like Future Cities Lab for example, where built structures “harvest” online activity through search engines and social media and translate the data into form using LED lighting technology. In this same way, I see a lot of potential for environmental and exhibition design leveraging these opportunities in the spatial realm. I also think this way of experiencing data may open up the way people interpret information…I may be speaking more to an American context specifically where data visualization tends to lean toward more pragmatic (graphs and charts) and conventional tropes of “infographics”…I see the tolerance for abstract interpretation of data to be far greater in Asia and Europe.

If you follow through this thought, you will ultimately end up in a discussion about the know-how and skillsets of the future graphic designer. Real-time design is based on templates and pre-designed software environments, in which graphic designers become dispensable. Unless of course, they do not simply apply the design tools anymore, but rather develop them themselves. Consequently, the future of the discipline lies in creating viable alternatives to the ruling business model of Silicon Valley and becoming involved in the processes of decision-making.

This is great because it separates the critic from the maker. Designers now have to rely on their taste and intuition to make creative decisions reacting to generated work. We can control the parameters and then curate the surprising results. I feel that designers that have extensive traditional training will benefit the most because this relieves a lot of the uncreative manual labor and puts more emphasis on the creative parts. Taste and intuition generally are a result of experience with manual labor and traditional training. So I’d say automation, computation, and generative systems will have a major impact on our work moving forward, but they will not replace aesthetic criticism and curation.

We are excited to witness our generative works evolving in real time. Once they have left the studio, they are released ‘into the wild’–with unpredictable consequences, which continue to surprise. Compared to traditional hand-crafted graphic design where we are familiar with every detail created, these results feel as fresh to us as the general public. There was much joy in watching our work Communion unfold into new forms of graphic life, beyond what we had imagined.