Data visualisation does not have to be the data vis that we know—it does not have to be graphs, plots, or diagrams → IT can be more. It can be something else, something more like a map, some kind of new structure, new aesthetic, new system -- Now data can be applied as system parameterisation.
Multiple stages of system data vis
1. Data brutalism: No over-designed layer 2. Data visualisation: Charts, Graphs, or sometimes 2D/3D Data vis 3. Dashboard: Combination of visualisation - designed somehow. Somehow an elementary system. 4. WorldBuilding: Decorative. Ian Cheng/Teamlab-like. Unity/Unreal visuals… 5. System Parameterisation: Web/Software based: (System) = f(param1, param2, param3, ..) all visual/audio output composed as the collage of these different parameters. When one dimension of the parameter is changed - either the whole visual, the part of the visual, the context of the visual, the style of the visual, the layout-composition of the visual is changed. Sort of like metaphor from VAE disentanglement but more direct/customised-for-situation/explicit approach.
All algorithmic output is eventually a sort of a data visualisation
---- Object: Now automatically easily generated (website and stuffs) Key: Object - not only generate - but mass generate (even further) - key to this is modularisation and retrieval of key parameters
Back then what was random parameters - only colors or hex cods or smth that can be explicitly parameterised/numerically transformed Now: for the context, thx for llm, all-natural languages can be modularised and mass-produced So whatever you build - look for a systematic pov, not as a slave of the system, but the creator/manager of the system, make it as a combination of parameters, a combination of data – and thus you can massively create different systems (which generate objects), when other just focus on mass generating objects on their own
COLORS, MINOR SIZES RANDOM/ADJUSTED → ALL PARAMETERS RANDOM/ADJUSTED
--- Gpt-4o: Image generation. Very good. All Images, Videos generation are telling me: Lots of static stuffs can now be auto-generated.
BUT this is not dynamic, this is not interactive - This is not a system. Objects are easily generated, and static objects are also easily generated. However, interactive objects—a system that defines the result of these interacted results—cannot be generated. Now we should focus on creating these systems. Systems which operate on top of parameters, and user adjusts -- and the parameters (some numeric, some contextual) redefine and reshape the system.
“FROM OBJECT TO SYSTEM” -
SYSTEM: Dynamic system? Ecological system?
For example, within an ecological system art, the user interaction (i.e. difference in system parameters) might effect
The small part of the visual (unit visual) The overall small degree The style of the visual -- Minroily, Majorily -- JUST DIFFERENTIATING THE WHOLE STYLE (pretty much metaphor from style transfer) The composition of the visual / The layout of the visual The signified description of the visual The image used in the visual
The temperature of the system Other different ecological parameters of the system Each temperature of each system
And the change in the ultimate ‘status’ ‘state’ of the system - in this case, the temperature - might affect ALL visual’s status, style, specs, texts, and graphs.
Collage - Parameterising Basquiat?
Here, for example: The subsection might be a graph. The graph itself might change, as well as the presentation of the graph (style, animation) accordingly The partial image might change - real-time generation? Getting image from real-world? Obviously the composition/layer might change. In the course of this change, the asset that’s used might also change. Or the overall theme might change! From Basquiat to more childish (minor adjustment), from basquiat street-style to minimalistic conventional media art style (major adjustment) The ingredients are the same: Globe visual on the left, 16 charts on the right But as the theme change: the visual output drastically changes. From minimalist neon-ish data visualisation aesthetic, to chaotic basquiat-like street style data visualisation. Or the overall composition might blink: Showing on/off, turning on/off
Other strategies to well-communicate the changing in data - but not explicitly, not using graphs, but more implicitly? Via other parameters? Any idea? --- Another automatic system: The ideal automatic system is self-evolving, self-generative, and self-calculating. Here, the parameter might play smth more than merely adjusting the visuals: Its parameter is a meta-parameter for the parameter that’s shown on the visual, adjusting the visual, but not directly accessible. Think of: 지수들. Poission Lambda, for example. ------
The aesthetic of automatic system (and apramter-based system): 개미들아 일하게 하여라! 개미 한명 한명이 일하는것을 보여주는것 (Single Agent/Single System Unit) 만으로는 충분하지 않다. 개미 여러명이 한꺼버ㅓㅂㄴ에 분주하게 일하게 하여라. 그리고 그 방향성, 그 일의 objective 응은 우리 인간들이 제공한다. – 그러니 여러 개미들아, 일하여라 (Objection - optmimsation to the objective) – 그 일하는 목적이 무엇인지는 질문하지 말고 (no doubt on the objective), 그냥 열심히 하라면 하여라. 오늘 내가 시키는 일이 내일 내가 시키는 일과 정반대이더라도 (시지프스), 그냥 열심히 일하라. 일종의 현대적인 파리 인종 동물원 – 그러나 이제 모든 인류는 평등하다. 그러나 이것을 인류와 기계가 평등해야한다는 것으로 착각할 필요는 없다. 기계들이여, 일하라. 우리는 양산을 쓰고 그대 기계-개미들이 열심히 일하는 ㄱ성르 고나망하고 있겠다. 우리는 너네 기계들에게 여러 먹거리를 주겠다. 이런 방향서응로 가여라, 저런 방향성으로 가여라. That’s system parameteric adjustment. And system & units gmove/morph accordingly. So they are really good at optimisating withi n the given objective, but not capable of reasoning beyond the given objective.
Text written by Jeanyoon Choi
Ⓒ Jeanyoon Choi, 2025