Part of Temporal Constructs Appendix, which contains details behind the research: dates, quotations, descriptions of instruments, and sources.

General definition

Distributed knowledge in Conductor is a conceptual framework that sorts the knowledge built into a musical instrument and the performance around it into four kinds. The kinds are constraint (what is possible), procedure (what acts), decision (what was chosen) and memory (what is remembered).

Relation to the practice

Sorting knowledge by where it is fixed follows Hutchins (1995). Hutchins showed that cognition is distributed across people and things. The four kinds are that sorting made for an instrument.

The seven concepts of that system are instances of these kinds. The kinds are constraint, procedure, decision and memory. A work can involve more than one person and more than one model. The account can fail to record a human choice that shaped the result. It can also fail to record where an agent’s contribution entered. A maker of such a work meets one question: where the decisions happened. This question is practical before it is theoretical. It concerns authorship and how the creative process is understood.

Concepts

This entry describes seven concepts for answering that question. In dependency order they are hardware, protocol, device, controller, preset, timeline and context, and each names a point where knowledge is fixed, who fixed it, and which part of the work it governs.

The framework comes from Conductor, the instrument this research built, a modular system in which human operators and AI agents work on legacy instruments through one shared command language, and it applies to the instruments Conductor works with. ALTERITY is the work built with it so far. The framework offers one account of what such an instrument does and where its decisions are made, and that account matters because it changes how authorship is read.

The concepts form a sequence in which each constrains what the next can do, so a decision at one point is shaped in part by the points before it, and that sequence suggests how distributed agency can work.
| Concept | Kind of knowledge | What it constrains | When its knowledge is fixed |
|---|---|---|---|
| Hardware | constraint | everything after it | before any session |
| Protocols | constraint | what can be mapped | long before a performance |
| Device | procedure | how states become sound | mostly before a performance |
| Controller | constraint | what presets can capture | before a performance |
| Preset | decision | what is easy to reach for | between performances |
| Timeline | decision | the shape of change over time | during composition or in the moment |
| Context | memory | later work, through accumulated context | after the fact and still active |

Kinds of knowledge

The framework proposes four kinds, grouped by the question each answers, and the grouping does not follow the dependency order. The device is a procedure lying between the protocol and the controller, both of which are constraints.

KindQuestion it answersConcepts
ConstraintWhat is possible?Hardware, Protocols, Controller
ProcedureWhat acts?Device
DecisionWhat was chosen?Preset, Timeline
MemoryWhat is remembered?Context

These definitions are working ones, and the tests below show where they hold and where they need revision.

Rest of the framework

The tests that keep the concepts apart, and the places where they overlap, are on The three tests. The order of authorship the sequence implies, and the disclosure that follows from that order, are on Distributed authorship. What the framework changes about reading the two works is on Reading Conductor and ALTERITY.