Article one of a series

What is new media art?

You walked into a room and a wall twenty feet tall was covered in slowly churning color. It was genuinely beautiful. You also had three thoughts in quick succession: I don't know what this is called, I don't know how it was made, and I'm not sure it has anything to do with me.

All three are reasonable. Let me take them in order.

You probably arrived with the wrong words, and that isn't your fault

People searching for this end up typing video art, new media art, digital art, or generative art more or less interchangeably, because those are the phrases that are available. They are not the same thing, and the field itself does not fully agree on them, which is why searching is frustrating.

The short version:

• Video art is a recording. It plays back the same way every time. Nam June Paik, Bill Viola, Gary Hill. Sixty years of serious practice, and probably not what you saw on that wall.
• New media art is an umbrella term from the 1990s that has aged awkwardly — "new" is a strange word for something with a sixty-year history — but it is still the phrase embedded in museum department names, curator titles, and grant categories.
• Generative art means the artist builds a system and the system produces the work. Rules, plus usually some controlled randomness. The output can differ every time it runs.
• Creative coding describes the practice and the tools — writing software as the act of making, rather than using software someone else designed.

If the wall you were looking at was different at 2pm than it was at 11am, and not because it was on a loop, then what you saw was generative, made with creative coding. That distinction is the whole subject of this article.

The history is much older than you would suspect — and it starts with weapons

The modern computer was not built for art, or for business, or for communication. It was built to aim things.

ENIAC, completed in 1945 at the University of Pennsylvania, was commissioned by the US Army's Ballistic Research Laboratory. Its job was computing artillery firing tables — given this shell, this powder charge, this elevation, this air temperature, where does the projectile land? Doing that by hand took a room of human calculators weeks per table. That specific problem, putting a projectile precisely where you intend, is what justified the machine.

Three things followed from that lineage, and each one turns out to matter for art:

Feedback. During the war, the mathematician Norbert Wiener worked on anti-aircraft fire control — not where the plane is, but where it will be in four seconds, and how a gun should correct as new information arrives. Generalizing that problem produced his 1948 book Cybernetics, about systems that sense, respond, and self-correct. Artists read that book. Cybernetics is the intellectual bridge between missile aiming and interactive art, and it is a direct line.

The screen. MIT's Whirlwind, begun in the late 1940s and folded into Cold War air defense, was among the first computers with a real-time cathode-ray display and a light pen — a person pointing at a screen and the machine responding. Every interactive graphic thing you have ever used descends from air defense.

Drawing. In 1963, Ivan Sutherland at MIT Lincoln Laboratory — a defense lab — completed Sketchpad, the first interactive computer graphics program. It is the ancestor of every drawing, drafting, and design application in existence.

None of this was intended to produce art. But by roughly 1960 there existed, in a handful of universities and military-adjacent laboratories, a machine that could execute a rule precisely, repeat it endlessly, and draw the result. Some people noticed what that was good for.

But the artists were already there, without computers

Here is the part that surprises people, and it is the most important section in this article: the ideas underneath computer art were fully developed by artists before artists had computers.

Working from a rule rather than from intuition is an old practice. So is the grid. So is deliberately handing part of the decision over to chance.

Vera Molnár (1924–2023)

From 1959 she worked with what she called her machine imaginaire — the imaginary machine. She would invent a rule, then execute it by hand, step by step, as if she were the computer. She did not get access to an actual computer until 1968. Nine years of algorithmic art performed manually, because the idea did not require the hardware.

Sol LeWitt (1928–2007)

His wall drawings are instructions. The artist writes the procedure; somebody else executes it on the wall; when the show closes it is painted over. What exists permanently is the instruction. In 1967 he wrote the sentence that this entire field runs on.

François Morellet (1926–2016)

From the 1950s he built paintings from explicit systems, and when he needed randomness he took digits out of the telephone directory. A pre-digital random number generator, used exactly the way a seed is used now.

Josef and Anni Albers

Josef's Homage to the Square is systematic permutation of colour through a fixed format. Anni's weaving connects to something more literal: the Jacquard loom of 1804 was controlled by punched cards, and punched cards are what ran computers for the next century and a half. The grid and the computer share an ancestor, and it is the textile industry.

John Cage (1912–1992) and Marcel Duchamp (1887–1968)

Cage used chance operations to determine musical structure. Duchamp's 3 Standard Stoppages (1913–14) let dropped threads decide a form. "Controlled randomness" as a legitimate artistic method was settled long before anyone could type random().

"The idea becomes a machine that makes the art."
— Sol LeWitt, Paragraphs on Conceptual Art, 1967

LeWitt wrote that about pencil on plaster. It is also, without a single word changed, an accurate description of generative software. That is not a coincidence or a clever comparison — it is the same idea, and the computer simply turned out to be a very good machine for it.

The first computer artists

When artists did reach the machines, two exhibitions in 1965 opened within months of each other. In February, Georg Nees showed plotter drawings at Stuttgart Technical University, in a circle around the philosopher Max Bense — who had coined the term generative aesthetics, which is where the word "generative" in "generative art" actually comes from. Frieder Nake, a mathematician, exhibited the same year. In April, A. Michael Noll of Bell Labs showed at the Howard Wise Gallery in New York, the first computer art exhibition in the United States.

What followed over the next decade:

• Charles Csuri at Ohio State — Hummingbird, 1967 — helping establish computer graphics as a field.
• Vera Molnár gains computer access, 1968.
• Cybernetic Serendipity, Institute of Contemporary Arts, London, 1968, curated by Jasia Reichardt — the landmark early survey.
• Jack Burnham publishes "Systems Esthetics" in Artforum, 1968, and curates Software at the Jewish Museum, New York, in 1970. The art world had a critical vocabulary for this by 1970.
• Manfred Mohr — a solo exhibition of computer-generated work at a Paris museum in 1971, having come to it from jazz and painting.
• Harold Cohen begins AARON around 1973 — a program that makes drawings, which he developed for decades. Cohen had already represented Britain at the Venice Biennale as a painter.

So: major museum exhibitions, a critical literature, and career artists committing decades to it, all before the personal computer existed. When somebody tells you this is a recent trend, that is the record.

What creative coding means now

The thing that changed the scale of this field was not a concept. It was tools.

Processing, released in 2001 by Casey Reas and Ben Fry, made programming usable by artists and designers who were not computer scientists. openFrameworks followed in the mid-2000s from Zach Lieberman and collaborators. Both are free. Nearly every artist working in this field today came through one of them, and their existence is why the field went from dozens of practitioners to many thousands.

The people making this work are artists, but a fair number are also computer scientists, engineers, or researchers, and many are all of those at once. That mix is normal here and it is not a sign of dilettantism. The output runs wherever there is a processor: in a browser, on a single screen, across a wall of LED, in a room full of sensors, or on a plotter drawing with a real pen on real paper.

Blockchain, briefly, and without enthusiasm

You cannot read about this field for long without hitting blockchain, so here is the plain version.

A blockchain is a ledger — an append-only list of records kept simultaneously on many independent computers. Because no single party holds the only copy, entries are very difficult to alter or delete after the fact. That is the entire technical idea. It is a record-keeping mechanism.

A cryptocurrency is a token that exists on such a ledger. Worth being clear about what it is not: it is not the artwork, it is not required in order to make or own this kind of art, and it is not what any of the artists above were working toward.

What the ledger genuinely provides for this work is two things.

Provenance. An unbroken, publicly checkable record of who made a work, when, and who has owned it since. Provenance is not a crypto invention — it is a core museum concern, and registrars maintain it with paperwork. A ledger is simply a version of that record that does not depend on the paperwork surviving.

Seeds, and reproducibility. A generative program uses a starting number — a seed — to initialize its randomness. The same code with the same seed produces the same result every time, permanently. That is what makes a specific output identifiable rather than arbitrary, and it means a work can be re-run correctly in ten years by someone who has the code and the number. For anyone thinking about long-term care, that is the interesting part, and it has nothing to do with prices.

This is not the ape thing

When people hear that digital art and blockchains are in the same sentence, they usually think of the cartoon-primate profile pictures from 2021. What I am describing is different. I want to be precise about how, because this is a structural distinction and not a moral one.

Both used similar infrastructure for a period, which is exactly why they get confused. But the artists in the right-hand column were doing this before there was any market for it and continued after that market collapsed, which is the most reliable way to tell the difference.

So what are you actually buying — the code, or the output?

This is a live argument in the field, and Casey Reas has been one of the more thoughtful voices in it. If an artist writes a program and the program produces an image, which one is the artwork?

It is not a hypothetical question. It determines what appears on an invoice, what a registrar catalogues, what a conservator is responsible for, and what you own in fifteen years.

Reas made this argument in the most direct way available in 2004, in a project called {Software} Structures for the Whitney Museum's online art space. He took Sol LeWitt's written wall-drawing instructions and implemented them as software — closing the loop between conceptual art's instruction pieces and generative code, and demonstrating that they are the same practice in different materials.

Which leads to the point I most want you to take away, because it is the one that makes this legible to any institution:

Buying the code rather than the output is not a strange new digital arrangement. It is the LeWitt model, and museums have handled it since 1968.

When a museum acquires a Sol LeWitt wall drawing, it does not acquire the painted wall. It acquires the certificate and the diagram — the authorization and the instructions. The physical drawing gets executed, exists for a while, and is painted over. The work persists as instruction. Registrars, conservators, and collections committees have been managing exactly this for over fifty years.

So if a proposal for generative work ever sounds unfamiliar in a collections meeting, the useful move is not to explain blockchains. It is to point out that the institution already knows how to do this, and has since 1968.

Where this leaves you

The wall of churning colour you saw was probably generative work, made with creative coding, in a practice that runs back through Processing to Bense and Molnár and LeWitt, on hardware whose ancestors were built to aim artillery. It is not a trend, it is not the cartoon apes, and it is not difficult to talk about once the words are sorted out.

Whether it has anything to do with you depends on what you do next. If you run a small museum, sit on an arts commission, or have to explain any of this to a board, you now have the vocabulary and the dates, which is most of what you need.

Next article: creative coding and blockchain — how the work is actually made, and what the ledger does and does not do for it.