← Articles

The method under the studio

Studio case note · June 2026

Underneath the studio and its production system sits a model. It's called Then.ODM, the Then. Organization Design Model, and it's the reason the rest holds together.

The model starts from a plain way of describing any piece of work. Every action has four parts: the source it comes from, the action itself, the target it operates on, and the intended state it's meant to produce. The vocabulary sits close to the cybernetic tradition — Ross Ashby's account of regulation in An Introduction to Cybernetics (1956) and Herbert Simon's framing of design as the science of how to reach intended states (Simon, 1969). A picking session has a source (me, with taste-clarity), an action (pick), a target (a set of candidate variants), and an intended state (one variant chosen, the downstream work unblocked). Name those four parts and the work becomes legible. You can see what each action costs, what resists it, and whether it landed.

The useful move is that source and target aren't fixed categories. They're points of view at different zoom levels. Zoom out and a whole pipeline is one action. Zoom in and that action is its own chain of smaller ones. This is the recursion principle Stafford Beer built into the Viable System Model — the same organizing shape repeating at every level (Beer, 1972). The studio at full zoom is a thing that produces work; zoom in and it's the CPS configured to run; zoom in again and it's a single task firing. Same four-part shape at every level. That's what lets one vocabulary describe a single task and the whole studio at once.

The part I lean on most is what it does to operators. Most systems treat the people and tools running them as fixed inputs. You show up rested or you don't. Then.ODM treats operators as sources, so their condition becomes something the system itself acts on: rest is an action whose target is the operator. (In this studio, that operator is often me.) Loading canon into a Claude session is an action whose target is that Claude. Maintaining the system is work, not overhead, and it sits in the same structure as the production. The instinct is older than it looks: Ashby modelled the regulator's own internal state as part of the system in Design for a Brain (1952), and his Law of Requisite Variety holds that only sufficient internal variety can absorb the variety a system faces (Ashby, 1956) — which is to say a tired operator is a regulator running short of what the work demands.

That's why a production company has a method under it. The method keeps the scaling honest. The same leverage can build or extract, and the direction depends on how the systems are designed. Then.ODM is where that direction gets set, and the studio is one place it gets exercised. (For the build-or-extract direction itself, see The Lever and the Debt.)