# System Art: The Elephant and the Blind

Canonical URL: https://jeanyoon.ch/oi/texts/system-art-the-elephant-and-the-blind
Type: essay / research note
Author: Jeanyoon Choi (최정윤)
Date: 2025-02-17

## Summary

Using the parable of the elephant and the blind, Jeanyoon Choi argues that we mistake observable symptoms, even sophisticated data analyses, for the dynamics of whole systems, so local solutions create problems elsewhere. He proposes multi-device artworks that run an underlying system while screens show only particular manifestations, from dashboards and maps to advertisements and agent profiles. He identifies four ways to enter such a system: controlling it, navigating it, becoming part of it, and encountering unexpected consequences, as when actions in Room A harm Room B, and asks whether AI could generate the system's graph.

## 요약 (Korean)

최정윤은 '장님과 코끼리' 비유로 우리가 시스템의 징후를 전체로 착각한다고 지적하고, 화면마다 시스템의 일부만 드러내는 멀티디바이스 작품과 참여자가 시스템에 들어가는 네 가지 방식(제어, 탐색, 참여, 예기치 않은 결과)을 제안한다.

## Questions this answers

### What are the four ways a participant can enter a system artwork, according to Jeanyoon Choi?

In 'System Art: The Elephant and the Blind' he lists controlling the system, navigating it, becoming part of it as an avatar or agent, and encountering unexpected consequences elsewhere or later.

### How can art show that local solutions create problems elsewhere?

Jeanyoon Choi proposes linked spaces in which someone in Room A achieves a local goal while harming Room B, whose response then harms Room A, so both work hard and the system goes nowhere; delays in time can work the same way.

Topics: system art, complex systems, multi-device web artwork, unintended consequences, data visualisation, zero-sum dynamics

References: [Blind men and an elephant](https://en.wikipedia.org/wiki/Blind_men_and_an_elephant), Mobile Jockey

## Full text

We live within systems that exceed our understanding, but we repeatedly mistake an observable symptom for the dynamics of the whole. The elephant and the blind people remains a useful image: each encounter is real, yet none is complete.

Even sophisticated data analysis can deepen this mistake if it treats the available measurements as the whole system. We solve the problem we can see, then discover consequences elsewhere or later. A local solution becomes another part of a larger problem.

I want to make the relationship between a system and its symptoms perceptible. A multi-device artwork can operate an underlying system while its screens display only particular manifestations. The algorithm is not fully visible. The participant senses that something larger connects the fragments without being given the entire elephant.

Those fragments need not look like conventional data visualisation. They could take the form of graphs, grids, dashboards, tables or stock-market displays; diagrams of relationships; advertisements, booking interfaces, shops, business cards and logos; generated images and videos distributed in two or three dimensions; typography and streams of text; repeated small multiples; clocks; avatars and agent profiles; maps and geographical plots.

I see four broad ways for a participant to enter such a system.

1. Control it. An action changes its parameters or dynamics. A small interest-rate adjustment could have amplified effects across several screens. A command could propagate through a bureaucratic hierarchy. The challenge is to make the action meaningful and engaging rather than simply provide a control panel.

2. Navigate it. The system already operates and evolves. The participant encounters different perspectives, perhaps as in a historical archive. A text request could generate a view of a particular part without changing the underlying system. Here the aesthetic of autonomous operation matters.

3. Become part of it. An avatar or agent enters a competitive or cooperative field. Mobile Jockey suggests one possibility, but I also want to explore what participation means with only one to five visitors in a physical installation.

4. Encounter unexpected consequences. An action succeeds nearby but creates a problem elsewhere. In Room A, someone achieves a local goal while producing a negative effect in Room B; Room B’s response in turn harms Room A. Both work hard and the system goes nowhere. Competition, ecological damage and zero-sum structures can be approached through this spatial relation.

The same structure can operate through time: a desirable immediate result carries delayed costs. Distance and delay become ways to experience the limits of a local viewpoint.

A further question is whether AI could generate some of the system itself: its inputs, parameters and relationships. That would require thinking about the graph, not merely generating new material for the screens.

---

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.