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

General definition

A research approach in which designing and making are the means of inquiry. Knowledge is produced inside the practice, through the process of building something and through the artefact itself.

Relation to the practice

The plan adopts research through design, because building and performing with the ecosystem produces knowledge about human-machine collaboration that theoretical study alone cannot reach. The approach unfolds in four modes that scale over the trajectory: archaeological recovery, in which the pioneers reconstruct their working methods and the languages are brought back to use; real-time interfaces that the legacy systems constrain, so a model has to collaborate; performance, with the model mediating between the artist and the machines alongside comparative validation with educational partners; and laboratory integration, for artistic demonstration and institutional pilots.

Documentation is itself the research data: session recordings and video analysis distinguish three-way collaboration from simple tool use and compare it with delegation-based tools, and performance documentation shows when the constraints of the old systems produce more human-centred behaviour. Educational pilots measure learning against traditional routes, and cost-benefit and adoption studies test whether the frameworks remove barriers or only produce novel tools.

Name for the method

The practice calls this research through engineering, and the doctorate has that name. Its mode of inquiry is building, and the doctorate’s first year shows what that means when the object of study is a technology that changes while it is being studied. The term sits beside research through design, as the form the approach took here.

First-year form

Engineering was the form the first year took, and the plan allows for it. When building is the means of inquiry, the year produces the system the argument is made in. Two of the plan’s four modes were worked in it: archaeological recovery, with HMSL ported and running, and KML written by an agent into its own editor. The other mode is the one in which the constraints of the legacy systems keep a model collaborating. That mode is CONDUCTOR, its command grammar and its sixteen modules. The modes that depend on other practitioners, with their educational partners and institutional pilots, belong to the later part of the trajectory.

The decision in April 2026 to set the educational commitments aside was a sequencing decision with a reason. The modes were worked in an order the plan did not set, because a capability the plan could not anticipate arrived while the year was running. It was an agent framework that could build its own tools around the period instruments. That framework is part of the second mode and changed what the mode could attempt. Nothing can be handed to others until the person handing it over understands its behaviour. Establishing that understanding was the work of the year.

The plan already states where the workshops and the studies of adoption sit when they return: in the third and fourth modes and in the later years of the trajectory. What is deferred is deferred to a point the plan already names. A year of building can be read as production. The answer is the one the plan already gives: documentation is itself the research data. That answer stands only while the decision record behind the build is kept. Keeping that record is the work the method still has outstanding.

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.