The Importance of Sketching with Code

2026-09-08 · page · source ↗ · source summary

Gorilla Sun (gorillasun.de) argues that sketching with code — small, exploratory programs that conjure visuals onto a canvas — is a fundamental creative practice, not a warm-up for "real" work. The sketch is a planted seed that fruits across a whole body of work, a system whose output space is the artwork, and a flow-state dialogue with a machine whose flawed interpretations are the engine of progress. Its sharpest claim: sketching is solving problems you don't have yet.

Key Quotes

"Sketching is a practice where you're solving problems that you don't have yet."

The thesis in one line, and the reason this essay belongs in a wiki full of pages about ROI, moats, and incentive structures. Every other framing of side work — portfolio, skill-building, product R&D — justifies the sketch by a future problem it will solve. Gorilla Sun inverts it: the sketch has value precisely because no problem is waiting for it. This is the same move The Usefulness of Useless Knowledge makes for basic research, applied to a weekend in a code editor.

"When you make a sketch, you basically conjure up a new concept, but at the same time, you're also planting a seed that might carry fruit at some point in the future, be that in the near future or many years down the line."

The seed metaphor carries the essay. He grounds it in Zach Lieberman's "sense of time" — ideas develop "across the span of your entire body of work," not a single project — and in his own return to an old irregular-grids algorithm, where "just one step further" unlocked an entirely new output space. The practical corollary is that the archive matters: sketches are worth saving because you can't predict which seed will fruit.

"In the advent of long-form generative art... the ultimate goal ceases to be output curation — but rather, the entire output space and the system itself become the artwork."

The systems turn. Once a single output is no longer the point, the craft shifts from making one good thing to designing the space of all possible things — a parametric system that "spans an interesting output space." He traces this back to Vera Molnar, who "always began with a system from the get-go," before she even worked with computers. This is a genuinely different discipline from one-off making, with its own set of "considerations and challenges" the essay only gestures at.

"There's interest in these flawed results though. These flaws are the essence that push the overall discourse ahead... It would be a boring dialogue if the machine knew exactly what you wanted right from the get-go."

The dialogue-with-the-machine passage. A sketch is an ongoing conversation: you input thoughts as code, and "it answers with what it thinks this thought should evaluate to" — sometimes accurate, sometimes a flawed interpretation lost in translation. The flaws are not bugs to eliminate but the thing that "pushes the discourse ahead," because they force you to "double down, and adjust your input... just like steering a very complicated vehicle." Note the honesty of the follow-up: you also "don't have a crystal clear image of what you're looking for," so the machine is a collaborator in discovering the destination, not just a dumb executor.

"In an age where attention spans are increasingly shorter, coding is an incredibly effective antidote."

The happiness close. Sketching gets him "into the zone" — in control, an empty canvas, and ctrl+z for every mistake. This is the direct rebuttal to the attention economy: where feeds extract, sketching replenishes. It's the creative-practice version of the Dopamine Fracking critique, stated from the inside rather than the outside.

Key Themes

#concept Ideas as seeds. Ideas are the "unitary building block" of creativity; a sketch plants a seed that may fruit far in the future. Implies an archival obligation — keep the sketches, keep the seeds — because you can't predict which one matters.
#pattern Solving problems you don't have yet. The reframe that rescues low-stakes making from the ROI lens. A sketch earns its keep by generating problems later that it will have already, quietly, been working on.
#concept The system as the artwork. Long-form generative art's goal is the output space, not the output. Lineage via Vera Molnar. Connects to any systems-first craft, including the Software Engineering Craft emphasis on fundamentals over artifacts.
#pattern Dialogue with the machine. The sketch as conversation; imperfection as engine. The machine's flawed readings — not its accuracy — are what push the work forward.
#person Zach Lieberman. Creative coder and educator; his "sense of time" — ideas developing across a whole body of work — is the essay's temporal frame.
#person Vera Molnar. Pioneer of generative art who began with a system before computers; the essay's evidence that systems-thinking predates the web3 long-form hype.
#tool Creative coding. The medium (implied p5/Processing lineage) where code conjures visuals; the essay is its defense as a practice, not a hobby.

Critical Analysis

The "problems you don't have yet" line is the durable contribution. Most essays on side projects justify them by the future: this builds a portfolio, this sharpens a skill, this becomes a product. Gorilla Sun cuts the justification entirely. The sketch is worth making because you're "solving problems you don't have yet" — problems that only become problems once you've already been working on them. That's a cleaner statement of why basic research and creative practice need no external justification than most policy arguments manage, and it's why this essay survives translation into a wiki otherwise obsessed with moats and economics.

The systems turn is the essay's most underdeveloped idea. "The system itself becomes the artwork" is a sharp observation, and the Molnar lineage is well-chosen, but the essay stops at the gesture. What are the "entirely new set of considerations and challenges" of system-oriented sketching? How does one evaluate an output space rather than an output? These are the interesting questions and they're left open. Fair enough for a short blog post, but it means the essay opens a door it doesn't walk through.

The dialogue-with-the-machine passage is elegant but pre-agent. Written in the web3-long-form era, the essay's "machine" is the compiler-plus-randomy — an interpreter that returns "flawed interpretations" of your code. In 2026 the same phrase names a different thing: a coding agent that doesn't misparse your intent but over-agrees with it. The flaws that drove the old dialogue — the machine "not knowing what you wanted" — have to be deliberately reintroduced when the machine is a sycophant rather than a stubborn interpreter. The essay's frame survives; the "flawed result" now has to be manufactured. Compare Creative Firewall's argument that what's worth protecting is the human's intention, not the tool's accuracy.

The seed argument implies a discipline the essay doesn't spell out. If sketches are seeds and seeds fruit unpredictably, the practice needs an archive — and the archive is where this connects to the wiki's own project. Gorilla Sun's concrete failure (losing the random seed his mother preferred, "basically gone forever") is the creative-coding version of the impermanence problem If AI Is Doing the Investigation, Version the Investigation names for reasoning: if you don't capture the state, the good one is unrecoverable. The sketch isn't just the output; it's the seed plus the parameters that let you replant it.

The antidote claim is the essay's quietest and most radical point. "Coding is an incredibly effective antidote" to shrinking attention spans. This flips the usual polarity — code as the opposite of the feed, not a faster way to build a feed. It's the personal-side answer to the question Dopamine Fracking leaves hanging ("what does non-fracked engagement look like?"): for a programmer, it looks like a sketch in the zone, not a timeline.

The Importance of Sketching with Code

Key Quotes

Key Themes

Critical Analysis

See Also