CONCEPT
Specification Affordance
The first and most fundamental of the
directional affordances: the possibility, unprecedented in computing history, of describing desired outcomes in natural language and receiving implementations — the action possibility the
large language model introduced to the
builder's environment.
The specification
affordance is the most fundamental new action possibility in the AI-augmented environment. Prior interfaces afforded implementation: the builder could write the code that produced the behavior, and the writing was mediated by formal syntax the builder had to translate into. The AI-augmented environment affords something categorically different: the direct specification of outcomes in the natural language of the builder's thought. She describes what she wants the system to do — not in syntactic structure, not in API call sequences, but in the same language she would use to explain the desired behavior to a knowledgeable colleague. The environment produces the implementation. This is not an improvement on the old affordance; it is a different class of action possibility entirely. In
Gibson's framework, the specification affordance transforms the builder's perceptual orientation to the problem: she now perceives problems in terms of outcome
affordances (this need affords a conversational interface, this constraint affords a microservice architecture) rather