A Systems Theory for Today

A living, forkable attempt to build a systems theory adequate to the present — by designed disagreement.

THE LIVING DOCUMENT

Version: 1.0 · Status: Ratified baseline (author, Session 4) · still living · Last updated: Session 4

What it means for this project to be a living document rather than a book, a paper, or a wiki — stated in essence, and held to its own falsifiable standard rather than accepted as a nice phrase. “Living document” is one of those terms that flatters a project while meaning almost nothing; most so-called living documents are just files someone edits now and then. This document says precisely what would make this one alive, what would make it merely pretending, and what it must avoid to keep the difference real.

1. The problem a living document solves

The History (HISTORY_OF_SYSTEMS_THEORIES.md) argues that the single-author total system died for reasons that are not going away: specialization outran any one mind, computation revealed irreducibility, the meaning-vacuum opened, and — the reason that matters here — the pace of the world outran the form. A book is finished on a date. The world it describes keeps moving. For a slow world that was tolerable; a treatise could stand for a generation. For today’s world (the adaptation gap, D1) a finished artifact is stale before its ink dries, and worse, it has no mechanism to notice or repair its own staleness. It cannot learn. It can only be replaced.

A living document is the form that answers this specific failure. It is not a stylistic preference or a publishing gimmick. It is the minimum viable response to the demand that a systems theory for today be adequate to today’s pace — one of the three unreplaced gifts the project exists to recover. If the diagnosis (the world changes faster than our forms adapt) is right, then the form of the answer must itself be adaptive, or it commits the very error it diagnoses.

2. What “living” means here — and the line against the wiki

A document is living, in this project’s sense, when it has all six of the following. Fewer than six and it is merely edited.

  1. Versioned and dated. Every file’s state at any moment is knowable (Version · Status · Last updated: Session N), and its trajectory is visible, not just its latest state (the snapshot discipline in METRICS.md). You can see not only what it says now but how it got there and where it changed its mind.
  2. Self-correcting. It runs a metabolism that turns error into rule change: catches → learnings → amended ground rules (logs/CATCHES.md → logs/LEARNINGS.md → docs/GROUND_RULES.md). It does not merely accumulate corrections; it changes how it operates in response to them.
  3. Disagreement-preserving. Its unsettled questions and standing splits are a first-class part of the document (logs/OPEN_QUESTIONS.md), not an appendix or an embarrassment. A living document is proud of its frontier.
  4. Forkable. Others can branch it, contradict it, merge into it, or replace parts of it, under a governance that keeps a commons coherent (CONTRIBUTING.md; Ostrom’s rules). Its life is not confined to its original authors.
  5. Refreshed on a cadence. Its time-sensitive claims (the contemporary-landscape survey, any live data) are re-checked on a schedule, because “what’s true now” decays (COMPREHENSIVE_PLAN.md, Phase 4). It has a pulse, not just a birth.
  6. Distributed in substrate. It is held by a process and a repository, not by one author’s working memory — which is why the Chat/Code shuttle and the multi-register memory (CHAT_CODE_WORKFLOW.md) are load-bearing, not incidental. A document that can only live inside one mind (or one context window) is not distributed, and dies when that mind moves on. This one is built so it literally cannot fit in a single context — and is therefore forced to live in the shared, durable form instead.

The line against the wiki. A wiki is versioned (1) and forkable-ish (4) and edited. It is not, by construction, self-correcting in sense (2), disagreement-designing in sense (3), or disciplined by falsifiability. Anyone can edit a wiki; nothing makes it learn, nothing makes it stage productive disagreement, nothing tracks whether it is getting truer. The difference is a metabolism. This project has one: designed disagreement as the engine, the catches/learnings loop as the digestion, the metrics as the sense of whether it is improving, and the governors as the immune system. That metabolism is what separates living from merely editable.

3. The essence: the medium is the thesis

Here is the whole idea in one line: this document does not merely describe the kind of theory today needs — it attempts to be one.

The project argues (Theory C, CANDIDATE_THEORIES.md) that because no single mind can hold the whole, a systems theory adequate to today must be distributed, living, plural, and self-correcting. A monograph making that argument would refute itself by its own form — a finished book by one author, claiming that finished books by single authors are no longer possible. The only non-self-refuting way to make the claim is to enact it. So the living document is not one feature of the project among many; it is Theory C made concrete. The form is the argument. If the form works — if this living document genuinely out-thinks what a lone author could produce — that is the strongest evidence for Theory C. If the form produces only the appearance of rigor while a single hand quietly makes every real decision, Theory C is being falsified in real time (which is exactly why the chair’s resolution-bias, C-006, is the most-watched risk).

This is the sense in which “living document” is not decoration here. It is the project’s central bet, wearing the clothes of a file format.

4. The metabolism (how it actually stays alive)

Concretely, the life runs on machinery already built:

  • The four-document loop — catches (raw), learnings (distilled), metrics (trajectory), handoff (continuity). This is the digestion: experience in, rule change out. Session 2 showed it close for the first time (a prior learning, L-001, caught a live error, C-007, before it shipped).
  • The frontier — OPEN_QUESTIONS.md, where the live questions and standing disagreements are curated as the most valuable part of the record and the first place a contributor is sent.
  • Versioned trajectory — session-stamped headers and per-session metric snapshots, so the document’s history of changing its mind is legible.
  • The refresh cadence — the standing obligation to re-scan the landscape and any live data, so the document tracks a moving world.
  • The fork model — the contribution and governance rules that let the document be extended by many without dissolving into noise.
  • The substrate: shuttle + multi-register memory — CHAT_CODE_WORKFLOW.md. The repository is the memory; Chat is the mind that holds only a slice; Code is the hands that keep the whole repo coherent and propagate every structural change. Multiple registers of the document (a pointer, a frontier, a working set, a maintained gist-of-the-whole, and deep pulls on demand) are held at once, at different resolutions — so the document can be worked on without any one worker holding all of it. This is what lets a living document outgrow the container that would otherwise kill it.

5. The honest test — what would make it actually alive

Skepticism first, because the term invites self-flattery. The falsifiable test of whether this is a living document or a pretentious static one is Goal S5: does the loop demonstrably change the rules and the roster in response to what the catches reveal, and does the plural process measurably out-catch a single author on the same questions? Two concrete, checkable signals:

  • Rule change under pressure. Count the rules and roster entries that changed because a catch demanded it (not because the author preferred them). If that count stays at zero over several sessions, “living” is marketing. (Current status: L-001 caught C-007; L-006 and L-007 turned author-caught defects into standing discipline. Early, but non-zero.)
  • Out-catching the lone author. The reflexive metric (COMPREHENSIVE_PLAN.md, Phase 1): the rate at which the panel-plus-loop surfaces caught errors and retained live objections, versus a single-author baseline on the same material. If the commons does not out-catch the individual, the plurality is cosmetic and the “life” is theatre.

A living document that cannot pass these is not alive; it is a well-formatted corpse with a version number. Naming that plainly is itself part of the discipline.

6. Failure modes (and their guards)

A living document has characteristic ways of dying that a static one cannot. Each has a guard already in the system:

  • Hoarding — accumulation mistaken for life: the document grows without distilling, volume masquerading as vitality. Guard: subtractive bias (Ground Rule 13); the demand that catches become learnings that change rules, not just entries; L-004 (layer, don’t lengthen).
  • Thrashing — churn without direction: endless revision that never converges or compounds. Guard: versioned trajectory and metrics (is it improving, or just moving?); author ratification as a real gate; the append-only respect for prior records so change is additive, not amnesiac.
  • Capture — the document optimizing for its own continuation and mistaking activity for progress (Luhmann’s warning, and Theory B’s disease turned inward). Guard: the S5 falsifiable test above; the standing governors that can declare the whole enterprise mistaken (Heidegger, C-002); never counting output as success.
  • Ossification — the opposite death: the “living” document that quietly stops changing, its headers still updating but its rules and roster frozen. Guard: the refresh cadence; the frontier that must stay populated; the periodic self-audit asking whether learnings are actually reaching the rules.
  • Fragmentation — forking that produces noise, not distributed coherence (Theory C’s second word failing). Guard: commons governance v1; a defined integration process; the branching-and-reconciliation discipline (Q-011).

A document that names its own ways of dying, and builds guards against each, is doing the thing that makes it alive: watching itself.

7. Two surfaces: published before open

A living document can be published (readable, versioned, evolving in public view) before it is open (forkable, accepting contributions). The project uses this split deliberately:

  • The reading surface — public early. A companion website presents the question, the landscape, the theories, the diagrams, and the open disagreements, in public, from an early stage. The ideas are legible, attributable, and evolving where anyone can see them. This is the living document shown.
  • The contribution surface — opened at Session 4k. The repository (forking, pull requests, rival theories) was kept private until the skeleton was stable enough that contributors fork a settling structure rather than a molten one — opening it prematurely invites forks of a shape that then changes underneath them (churn, not commons). At Session 4k the author judged that gate met and opened the repository to the public, closing the build and handing the attempt to whoever might carry it — in the honest knowledge that a community may never form. Opening is the one move that could someday supply the genuinely-foreign vantage the closed build conceded it would not get; it does not summon one.

So: private repo, public website, until ready — then open the repository for contribution. Publishing the reading surface early honors the living-document ethic (evolve in the open); gating the contribution surface honors the coherence half of Theory C (distribution with coherence, not before it).

8. In one paragraph

A living document, in this project’s sense, is not a file that gets edited — it is a process with a memory and a metabolism: versioned so its trajectory is visible, self-correcting so error becomes rule change, disagreement-preserving so its frontier is a feature, forkable so its life exceeds its authors, refreshed so it tracks a moving world, and distributed so no single mind or context has to hold it whole. It is the only non-self-refuting form for a theory that claims no single mind can hold the whole. It is alive exactly to the degree that it changes itself in response to what it learns — and dead, whatever its version number, the moment it stops.

Author: Hulki Okan Tabak — with Claude · License: CC BY-SA 4.0