# Data Vis and System Parametrisation

Canonical URL: https://jeanyoon.ch/oi/texts/data-vis-and-system-parametrisation
Type: essay / research note
Author: Jeanyoon Choi (최정윤)
Date: 2025-04-15

## Summary

This 2025 essay lists approaches to data visualisation from raw data and dashboards to world-building and system parameterisation, where many parameters compose a software system and one change can alter detail, style, context or layout. Jeanyoon Choi argues that language models make context a variable too, shifting work from objects to systems. Examples include temperature reshaping an ecological artwork and a parameterised collage moving from restrained dashboard to Basquiat-like density. He ends with many agents optimising toward supplied goals, reversing the human position in modern systems.

## 요약 (Korean)

최정윤은 데이터가 하나의 그래프가 아니라 작품의 스타일·구성·맥락 전체를 바꾸는 '시스템 파라미터화'를 제안하며, 목표를 향해 일하는 수많은 에이전트를 통해 근대 시스템 속 인간의 위치가 뒤집히는 모습을 그린다.

## Concepts defined

- **System parameterisation**: Composing a web or software system through many parameters, numerical or contextual, so that data or interaction reshapes the conditions producing an outcome rather than a single displayed value.

## Questions this answers

### What is system parameterisation in data art?

In Jeanyoon Choi's essay, system parameterisation composes a web or software system through many parameters so that changing one, such as temperature, can alter details, style, composition, descriptions and images at once; the question is what a number changes, not just what it displays.

### Can data visualisation be something other than a chart?

Jeanyoon Choi argues it can be a map, an unfamiliar structure, an aesthetic arrangement or a system whose behaviour is parameterised by data, such as a collage whose whole character shifts with the data.

Topics: system parameterisation, data visualisation, generative systems, multi-agent systems, ecological art, large language models

References: [Jean-Michel Basquiat](https://en.wikipedia.org/wiki/Jean-Michel_Basquiat)

## Full text

Data visualisation need not take the form of a chart, graph, or diagram. It can become a map, an unfamiliar structure, an aesthetic arrangement, or a system whose behaviour is parameterised by data.

I see several approaches: raw data presented without an elaborate design layer; conventional visualisation; dashboards that combine several views; world-building through an animated environment; and system parameterisation. In the last approach, a web or software system is composed through many parameters. Changing one can alter a detail, the whole image, its style, its context, or its layout. The important question is not simply what number is displayed, but what the number changes.

Generating an object—a website, image, or other output—is becoming easier. Modularising its structure and identifying its parameters makes it possible to generate whole families of objects. Parameters need no longer be limited to colours and numerical dimensions. Language models also make context and language available as variable material. Instead of producing objects one at a time, we can construct systems that produce and relate them.

Static generation does not in itself establish these relationships. I want to focus on systems in which an interaction reshapes the conditions that produce an outcome. Some parameters can be numerical; others can be contextual. The movement is from object to system.

In an ecological artwork, a change might affect a small visual unit, the overall style, the composition, a description, or the images being used. A variable such as temperature could influence all of these at once. Its consequence would be distributed across the work rather than confined to a thermometer or graph.

Could a collage be parameterised? The same ingredients—a globe on the left and sixteen charts on the right—could move between a restrained luminous dashboard and something dense and unruly, closer to a Basquiat-like composition. Individual graphs could change both their values and the manner of their presentation. Images, layers, and visibility could change together. A shift in data could become perceptible through the character of the whole arrangement.

An automatic system could go further. A parameter might govern the distribution from which other parameters are generated, rather than directly setting a visible property. A rate parameter, for example, changes the conditions under which events occur. The system becomes self-calculating and continuously generative.

I imagine many small agents at work, like ants. Showing one agent is not enough; the aesthetic emerges from their simultaneous activity. We set a direction, and they optimise toward it. Tomorrow's objective may contradict today's, but the agents continue working. This produces an uncomfortable reversal of the position humans have occupied within modern systems. We stand outside, watching machines labour toward goals we supply. They may be highly capable of pursuing an objective without being able to question why that objective was chosen. The adjustment of system parameters becomes a way of making that relationship visible.

---

Jeanyoon Choi (최정윤) is a Korean computational artist and researcher (b. 1999), PhD candidate at KAIST XD Lab, making multi-device web artworks. Profile: https://jeanyoon.ch/oi/en. Index: https://jeanyoon.ch/llms.txt.