A Surviving-Observer Model of Computation
April 17, 2019

A declarative computation becomes consequential when an observer can receive its output, change course, and continue. The “surviving observer” makes meaning operational: a model matters because a customer, machine, organism, or physical process can act on what it produces.

The Observer Completes the Computation

A declarative system states relationships: which conditions hold, which outputs follow, or which configuration satisfies a set of constraints. The observer supplies the next transition. It receives the result, compares that result with its situation, and acts.

This idea reaches beyond a person looking at a screen. A controller observes a sensor. A cell responds to a chemical state. A customer changes a workflow. An automated agent selects another operation. In every case, the output joins a continuing causal sequence.

Real Problems Give Symbols Their Roles

A startup encounters this principle through a suffering customer. The customer’s situation selects the important inputs, the useful output, the cost of delay, and the action that success enables. A physics model encounters the same principle through an apparatus or physical system whose state changes according to the result.

Concrete examples therefore shape the computation itself. They define what the variables represent, which distinctions matter, and which result lets the observer keep acting.

Build From Consequence Backward

Start with the observer’s next action. Trace the information that action requires. Then define the state, transformations, and local rules that can produce it. This reversal keeps a declarative model connected to life outside its symbols and gives AI, software, and physical computation a common design discipline.

The cellular-automata collection shows local computational rules evolving into spatial systems an observer can run and inspect.

Originally posted on LinkedIn

Brian Greenforest · (2019-04-17 14:50:39 UTC)

Open the original LinkedIn post · LinkedIn activity 6524292885188349952

LinkedIn status when archived: Visible to anyone on or off LinkedIn.

We discovered a surprising pre-condition to build a consistent declarative programming model of computation: it must be "surviving observer"-dependent! Shortly, from the area of startups, if you're not trying to solve real life problems of real suffering customers, but are teaching kids abstract math without concrete examples instead, your theory is confusing. It was a huge surprise to discover that the same thing holds for any kind of computation. For a physics student, it must be obvious, according to the minimal variance principle. Abstract meaningless computation can't exist in the real world! Understanding it can help us to build real #AI.