Part of Temporal Constructs Appendix, which contains details behind the research: dates, quotations, descriptions of instruments, and sources.
General definition
The account of authorship the framework implies: decisions fixed at different points in an instrument, by different hands, at different times. It includes the disclosure that follows from knowing where they sit.
Relation to the practice
This page sets out the order in which authorship accumulates along the framework’s ladder, and states what disclosing that distribution is for. The concepts themselves are on Distributed knowledge in Conductor, and the tests that keep them apart are on The three tests.
Distributed agency
Authorship can be read in the order in which decisions were made. Hardware makers decided what the instrument can do, and protocol designers decided which parameters it can reach. A device builder decided how saved settings become events, and a system designer decided how the device and the hardware connect. The performer then chooses presets and shapes a timeline, and the work itself adds context as it develops. Read in this order, creative intelligence is spread across the people and processes that shaped the instrument.
If decisions are made at different points, agency is distributed, and the question is how those points relate to each other. The framework provides terms for this question. Each concept marks a point where a decision was made and names the person or process that made it. The framework does not yet describe how these authors negotiate with each other, or whether the same arrangement applies from one work to the next. The research is testing both questions.
Clark and Chalmers describe the conditions under which a resource counts as part of a cognitive system. They argue that this is the case where there is a high degree of trust, reliance and accessibility (Clark and Chalmers 1998, p. 17). These conditions can be applied to the levels of the ladder. The device relies on a protocol, the performer trusts a preset, and context remains accessible to later work. On this reading, what the framework calls negotiation may be the maintenance or the loss of trust, reliance and accessibility over the life of a work. The question can then be examined in the session record.
Ingold gives a further account of creativity, developed with Hallam. In this account, making is an improvisational correspondence with materials, in which the maker responds to the material as the work proceeds. Novelty is not a condition of creativity. If form develops in the course of making, no single author decides in advance everything that shapes a work. These decisions are then distributed through the process.
Separating the concepts also has a practical use. Each concept names a component with its own author and its own boundary. The framework can therefore be used to give agents smaller, specific tasks. One agent may be given the protocol mapping, and another the device behaviour. A shared vocabulary lets the agents work on different components of the same process without treating it as one undifferentiated job.
Authorship and disclosure
Disclosure means stating the role that each party had in a work, and the influence each party had on it. It shows which choices were the artist’s, which were delegated, and where the system acted on its own. This matters for authorship because credit and responsibility depend on this account. It also affects the creative process, because influence that is visible can be discussed, adjusted or refused. The framework supports disclosure by naming the points where knowledge is fixed and the author of each point. This makes the distribution clear enough for the question of credit to be asked. The framework does not decide how those authors should be credited.
Authorship across models
The arrangement used in the practice adds one further point to the ladder: the inference engine, which is the language model that produces the output. This model can be exchanged while the practice’s own files stay the same. The record names a harness, which is the software that runs the model and holds its skills, instruction files and memory. The models inside the harness come from makers in several jurisdictions. Authorship is therefore distributed across a succession of models as well as across the parts of an instrument. No single model can be named as the author of a passage of the work. For this reason the record names the arrangement and its files, and not a model, except where the behaviour of a model was measured and stated.
Unforeseen
Here the unforeseen situations, sometimes difficult, sometimes unexpectedly interesting, will be described as the project runs. Many of these unforeseen situations have already occurred and have been recorded. They will be added in the time to come.