Exploring (or, Searching)
Generating through rulemaking is a powerful tool for designers working computationally. But the sheer size of the output poses a new problem: How does one navigate an immense field of possibilities? For digital architecture historian Mario Carpo, the answer lies in increasingly powerful computational search algorithms, such as those used by Google to find information from among the billions of indexed web pages on the internet.1 In this technologically determinist view, if one knows that a certain optimized solution exists, it is possible to arrive at it through algorithmic processes. While computational techniques like genetic algorithms and neural nets trained on relevant datasets are adept tools for parsing the results of a generative procedure and determining optimal results, a technology-first approach downplays the role of the designer in the process. Rejecting the dominant narratives of data and computation as panacea to all problems (design or otherwise), artists, architects, and designers are devising approaches to exploring generative possibility spaces that harness technology while transforming the role of the human.
Natalie, a software developer and designer who creates interactive, web-based visualizations, subverts established generative practices to restore agency to the artist. Describing a piece that resembles a modular, gridded, sprawling circuit board, she reveals that, “Although at first glance it does look pretty symmetrical, pretty generative, if you were to really dig in there… There are minor inconsistencies and stuff like that, which I think lends it that organic-ness and its realness. There’s a way to do it with code, but do I want to? No.”2 Natalie uses an aesthetic associated with generative art (which I mistakenly assumed the piece was in our interview) but maintains complete, manual control over the creative process. Her work critiques a view of the role of the designer as merely selecting from a deterministic set of choices, trading away creative agency in exchange for breadth of possibilities.
However, for others, giving up control over individual objects is desirable for increasing corresponding agency at a level of abstraction. In this view, exploring a possibility space is neither randomly selecting from it nor exhaustively searching it via algorithm. Instead, it can be an aesthetic experience that provides a starting point for further iteration, whether manual or generative. Max, an artificial intelligence researcher, worked on a project to generate plunderphonics music tracks — a style which samples and combines various audio sources into a musical collage — from a ‘seed’ song. Using, for example, an album by Prince as an input, the software might pick up on the use of falsetto, drum machine, and synthesizer (that is, the rules of a generating system), and assemble tracks with overlaid, found audio clips approximating that sweet Minneapolis sound. Importantly, while the algorithms perform a great deal of ‘labor’ in data analysis, searching for audio clips and assembling them according to rules, what to do with them from there is expressly left to the designer-listener. The produced track comes with an index of its source clips, making it not so much a listenable artifact as a new starting point. Max elaborates, “It’s like a little toolkit where you have the samples, you have the track listing, and you have one possible track that could be created with the samples, and you can create something totally different from that if you want.”3 In his view, exploring a generated space of possibilities isn’t about arriving at one optimal solution so much as discovering something with potential to be further investigated and expanded on by the human designer.
Another aspect of this example stands out: As opposed to the work of Anna, Paul, and Natalie, who all dictate the initial rules themselves to a generative algorithm, Max’s software takes over the analytical step. Automating information processes is central to the fields of artificial intelligence and machine learning, and while there are certainly risks involved, it also serves as a strategy that allows designers to generate systems in new ways.
1. Carpo, Mario. The second digital turn: Design beyond intelligence. MIT Press, 2016. ↩
2. Google Hangout interview with Natalie, August 1, 2017. ↩
3. Phone interview with Max, August 2, 2017. ↩