Dictionary

A spectrum, not a binary — and the axis is who the work is *for*. Made purely for yourself, it sits at the art end; the moment you start accounting for someone else's perspective (an audience, a user, a client), it leans toward design. Almost nothing is 100% one or the other; most work lives somewhere along the dynamic range between those two extremes.

A casual game session where you play with friends against computer-controlled opponents (bots) instead of against other people. Used in the RTS community, for example for a weekly group night against the AI. Informal slang with no official definition. Also called a comp stomp.

The practice of using unusual or adversarial message patterns to get a language model to do something its guardrails are meant to prevent. The name borrows from *phone phreaking*, the hobby of exploiting the telephone network to do things the phone company didn't intend, like free long-distance calls. Where "jailbreaking" names the outcome, chat phreaking names the craft and the subculture around it: the tinkering, the swapping of patterns, the curiosity about how the system works.

The principle that technologists should be mindful of the ways their technologies could be misused by authoritarian powers. Coined by Bruce Schneier in 2010.

Playing a game against computer-controlled opponents ("comps") instead of against other people. Used in the RTS community as a synonym for a bot bash. Informal slang with no official definition.

A small interactive widget inside an explanation that lets the reader change one thing and watch what happens. A toggle that switches a bug on and off, or a button that makes two people edit at the same time, are toys. Each one is built around a single idea, runs on tiny made-up data, and exists to be played with, not used. The word comes from *toy example* and *toy model*, where "toy" means a deliberately simplified stand-in for the real thing. It is an informal term with no official definition.

A single, one-shot LLM call that produces its output directly — no turns, no tool use, no observation loop. Less a formal architecture than a useful counterpoint to the ReAct agent: where ReAct interleaves reasoning and action over many steps, a generator agent just generates and stops.

Rendering a pre-defined expectation or output. Where painting is an open-ended conversation with materials, illustration starts from a known target — a brief, an idea, a message — and executes toward it. The goal exists before the first mark does.

A feral pig that is trapped, tagged or fitted with a tracking collar, and released so it rejoins its group and leads hunters back to other pigs. Adapted from the older "Judas goat" technique. Sows tend to work best, since they stay with their group.

A person who mindlessly relays an AI's output to others instead of engaging with it themselves — passing along generated text without reading, understanding, or validating it, adding nothing beyond being the human in the loop. Popularized by Niklas Gruhn's post "Don't be a meat proxy" (2026): the antidote is to read, understand, and rewrite the AI's answer in your own words as proof you actually did the work.

A conversation with materials. In the action-painting sense, it's loose and open-ended — responding to what the paint actually does rather than rendering a predetermined image. There's no explicit end goal; the work emerges through the process itself. (Contrast with illustration, which renders a pre-defined output.)

An agent pattern that interleaves reasoning and acting: the model alternates between natural-language 'thoughts' (reasoning traces) and actions (tool calls, queries), feeding each observation back into the next reasoning step. Introduced by Yao et al. in 'ReAct: Synergizing Reasoning and Acting in Language Models' (2022).

The juxtaposition of two materials — a conversation between them, where the work lives in how they're set against and combined with each other rather than in any single medium alone.

An ETA that is just total bullshit.