Open-ended construction builds an object and a theory of it

Type: kb/types/note.md · Tags: foundations, learning-theory

Status: conjecture.

When an object's purpose and constraints do not determine its organization, building it produces two coupled results: the object and a theory of it. This theory relates the object to its purpose and environment, explains why its parts have their roles, and guides later changes that were not specified in advance. The consequences of construction choices prompt revisions to the theory, and the revised theory guides further construction.

Complexity alone does not produce this coupling. A complicated object can be reproduced from a complete specification without the producer developing a new theory. The conjecture concerns open-ended construction: the organization itself must be discovered, local choices interact, and later demands may require the builder to reconsider the whole.

Two cases

Programming

Peter Naur's theory-building view of programming supplies an analogy for the general conjecture. It is not evidence that every open-ended construction has this form. The program text is the constructed object. The programmers' theory connects the program to relevant affairs in the world, justifies the roles of its parts, and lets programmers relate a later demand to facilities already present in the program. The analogy lies in this relation between object and theory.

Theory building

A theory is itself an organized object: its problems, concepts, claims, scope, and the relations among them have been selected from alternatives that the evidence does not fully determine. The resulting object theory explains its subject. The construction theory explains why the object theory has its present organization and how unresolved problems or new evidence should change the object theory.

The two theories develop together. Evidence may defeat a claim about the subject and revise the object theory. A failed investigation may instead show that the chosen problem, abstraction, evidence boundary, or test made progress unlikely. That finding revises the construction theory and changes how the next object theory is built. Theory building is therefore the reflexive case of the general construction conjecture, not a separate kind of activity.

Scope

  • Producing a satisfactory object once does not demonstrate possession of its theory: copying, search, or luck can produce the object. The conjecture instead concerns the sustained capacity to explain the organization and modify it coherently under novel demands.
  • The theory need not be a separate document or be held by one person. It may be tacit or distributed across people, models, and retained artifacts.
  • A process that repeatedly handles novel demands coherently through direct search or optimization, with no project-specific theory contributing to its choices, would refute the conjecture for that process.

Relevant Notes: