A Systems Theory for Today — Project Report

Version: 1.0 · Status: Reading-surface synthesis (not canonical theory; placed by Code) · Last updated: Session 4e

A synthesis of the project as of Session 4d. Written to be read on its own by someone who has never seen the repository. It covers three things, in order: the history (why the kind of theory this project attempts stopped being possible in its old form), the current situation (what exists in the field now, and the gap this aims at), and this project — its method, its architecture, and the three rival “tester” theories it has built and is beginning to test. It is written in the project’s own register: explanatory, opinionated where opinion is earned, and skeptical of its own claims. One sentence to hold throughout: the project has built an unusually disciplined design for finding out, and has not yet found anything out — the first empirical test is only now cleared to run.

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

I. The question, and the wager

The project begins from a deceptively simple question: can there be a systems theory adequate to the 2020s — and if so, by what method?

For most of history the answer to “who explains the whole?” was a single mind and a single book. Aristotle, Aquinas, Hegel — a person held a total account of reality and wrote it down. That form is now foreclosed, for reasons set out below. The project’s wager is that the need the single-author system served — a total-enough understanding of how things hang together — has not vanished, and that what died was the method, not the ambition. So it asks whether a theory adequate to today can be built by a new method: living (continuously revised rather than finished), plural (many vantages in structured disagreement rather than one author), computationally augmented, and self-correcting.

Crucially, the project does not merely argue this. It attempts it — it is itself an instance of the method it proposes, on the theory that the only non-self-refuting way to claim “no single mind can hold the whole” is to build something that isn’t a single mind holding the whole. This reflexivity (“the medium is the thesis”) runs through everything, and it is both the project’s most interesting feature and its greatest risk of self-flattery, a tension the project tries to keep honest rather than resolve.

II. The history — the single-author total system, and why it died

(The project’s full account is in its History document; this is the compression.)

For most of human history, total explanatory systems were the norm, not the exception. Theological cosmologies gave way to the great rational systems of antiquity through the Enlightenment — Plato and Aristotle, the Scholastics, then Descartes, Spinoza, Leibniz, Kant — reaching an apex in German Idealism, where Hegel offered history itself as reason unfolding through contradiction: arguably the most ambitious total system ever built. Then the inversions began. Marx turned Hegel’s idealism onto its material feet; Nietzsche took a hammer to the whole enterprise of systematizing, exposing the hidden value-posits beneath every “objective order.”

Between roughly 1880 and 1970 the form itself broke, from several directions at once:

The limits of formal system. Gödel showed that any formal system rich enough to contain arithmetic cannot be both complete and consistent — a permanent crack in the dream of a closed, self-grounding total system.
Specialization. Knowledge grew faster than any one mind could hold. The polymath became structurally impossible; the last people who could plausibly know “everything known” lived generations ago.
The meaning vacuum. The “death of God” (Nietzsche’s phrase for the collapse of a shared metaphysical ground) removed the anchor that had let total systems feel inhabited rather than merely constructed. A system can be built; a shared why cannot simply be authored.
Computational irreducibility. Later, complexity science revealed that some systems have no shortcut — they can only be run, not solved. No closed-form theory can outrun them.

But the ambition did not die — it migrated. It moved out of philosophy into systems science, cybernetics, social theory, and complexity science: von Bertalanffy’s General System Theory, Wiener’s cybernetics, Luhmann’s theory of society, the Santa Fe complexity tradition. The impulse to understand the whole went looking for new homes and new methods.

The project’s reading of this history is precise and load-bearing: the single-author total system is genuinely foreclosed — but the need it served is not. That gap between a dead form and a living need is exactly the opening the project aims at: a total-enough understanding held by a process rather than a person.

III. The current situation — the field now, and the empty corner

(The project surveys eight serious contemporary attempts and scores each on four axes: does it join the domains, does it carry lived meaning, is it adequate to today’s pace, and is it falsifiable.)

The contemporary field is not empty — but nothing in it delivers all four. In brief:

Cliodynamics (Peter Turchin) — history as a quantitative, falsifiable science; structural-demographic models of instability (elite overproduction, popular immiseration, the “wealth pump”). This is the project’s rigor bar: it proves a systems theory can be built by new, testable methods. Its limit is scope — it models political instability, not the whole.
Complexity science / system dynamics — the mathematics of the whole (feedback, emergence, networks, tipping points). Powerful and general, but it is a toolkit, not a theory of the present.
Luhmann’s social systems theory — the strongest post-Heidegger attempt to explain all of society as self-reproducing communication among functionally differentiated subsystems. It is the total theory that already exists — and its cost is instructive: it has no place for the human being, no subject, no meaning. It is a cautionary success.
Polycrisis / metacrisis — names the target (many crises entangling into one) but, by common admission, leaves the arrows undrawn: it is a label and a sensibility, “neither a comprehensive theory nor an approach.”
Big History — the whole story (cosmos to now, organized by complexity thresholds), but not the whole system: narrative, not mechanism.
Metamodernism — the stance the project already occupies (provisional, ironic-and-sincere proto-synthesis after postmodernism), more posture than theory.
Integral Theory (Ken Wilber) — the most explicit “theory of everything,” and the cautionary tale: comprehensive to the point of being unfalsifiable, explaining everything and therefore predicting nothing.

Positioned on two axes — interpretive → predictive and partial → total — the field clusters everywhere except one place. The top-right corner — total and testable — is nearly empty. That emptiness is the niche. The survey commits the project to two disciplines its neighbors lack: to be falsifiable where Integral is not, and to draw the couplings where polycrisis only lists them.

IV. The project — method, architecture, and the loop

The project’s answer to “by what method?” has three parts: a way of thinking, a way of remembering, and a way of correcting.

The method: designed disagreement

The engine is a panel of ten core thinkers plus twenty-two advisors, run as designed disagreement. The core ten are chosen not for agreement but for the fights they pick: Aristotle (build the whole from observed particulars) against Kant (the order is partly imposed by the knower, and knowledge has limits); Heidegger — kept deliberately as the internal adversary — arguing that the whole systematizing project is itself the disease, the technological “enframing” that turns everything into optimizable standing-reserve; Nietzsche insisting the meaning-vacuum is the central fact no diagram can fill; Ibn Khaldun and Braudel on the deep cycles and slow structures of history; Meadows on feedback, overshoot, and leverage points; Turchin as the demand for falsifiability; Luhmann as the coldest structural voice and the most direct rival to the very idea of a unified (rather than differentiated) theory; and Le Guin keeping ethics, imagination, and the question of how to live on the table when the systems-talk goes cold.

The rules are explicit (there are 23 of them). The most important: steelman before critique; the chair synthesizes but never votes; disagreement is manufactured when it’s missing; and dissent is never erased. Panelists are treated as reconstructed positions, not the actual persons — a fabrication guard. And the project’s single most-watched risk is C-006, the chair’s drift toward tidy closure — the danger that the “plurality” is cosmetic and one voice is quietly deciding everything. Two rules exist specifically to guard it: draft the dissent before the consensus, and label the strength of every agreement (consensus / majority / plurality / contested).

The architecture: a living document, two surfaces, two tools

The project is built as a living document — not a file that gets edited, but 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 not an embarrassment, forkable so its life exceeds its authors, refreshed so it tracks a moving world, and distributed so no single mind or context holds it whole. This is the concrete form of “the medium is the thesis”: the document is the third theory (below) made real.

It now runs across two surfaces — a public reading website (live) and a private, until-stable contribution repository — and two tools: Chat deliberates, Code executes (writes files, runs checks, deploys), and the repository is the shared memory. Neither tool holds the whole project in view at once; a “multi-register memory” lets a working session hold the map, the frontier, and a compressed gist of everything while pulling full detail only on demand. This is not incidental plumbing — it is what lets the project outgrow the context limits of any single mind or session, which is precisely the condition the theory says defines the present.

The loop: how it corrects

Experience enters as catches (errors, near-misses, flaws — 13 logged), which are distilled into learnings (patterns — 9), which, when they imply a rule, are folded into the constitution (the 23 ground rules). Metrics track whether the thing is actually improving; a reflection document keeps honest stock. The loop has demonstrably closed more than once — a prior learning caught a live error before it shipped; author-caught defects became standing disciplines; and, most tellingly, the loop once caught a flaw in itself (a learning that sat un-promoted for three sessions), and turned that latency into a new rule. A system that can catch and correct its own procedural failures is doing the thing the project claims is possible.

V. The three tester theories

The project does not assert one theory of the present. It seeds three rivals, holds them in tension, and is building falsifiable tests for each. They are deliberately of different kinds — one diagnostic, one mechanistic, one reflexive — and they are not mutually exclusive; the interesting questions are where they agree, where they diverge, and where each can be proven wrong.

Theory A — the Adaptation Gap (diagnostic)

Claim: today’s defining condition is a widening gap between the rate of change of the systems humans build (R_c) and the rate at which institutions, cognition, and culture adapt to them (R_a). Most symptoms are downstream of G = R_c − R_a.

Operationalized: build each rate from pre-registered, domain-normalized, equal-weighted proxies (technology-adoption speed, capability benchmarks, firm turnover for R_c; regulatory lag, legislative cycle time, skill/curriculum half-life for R_a). Predict that G positively tracks a symptom-severity composite, cross-sectionally and over time, with G leading — and, the decisive test, that G survives controlling for the wealth pump (Turchin’s inequality mechanism). If inequality alone explains the symptoms and G adds nothing, A is redundant; if both carry independent weight, A and Turchin are complementary. Drawn as a signed causal loop (a reinforcing “trap” loop versus a delayed “balancing” loop that encodes the accelerationist counter-argument).

Weakness: the best integrative reach, the weakest falsifiability — a gap between two rates is only a theory if both rates can be measured, and the composite risks becoming an elegant restatement of “things change fast.” Its first study (≈30 OECD countries) is the heaviest data lift.

Theory B — the Optimization Ecology / Autopoietic Capture (mechanistic)

Claim: self-reproducing optimizers — markets, media, technical systems — capture human drives (attention, dopamine, outrage, desire) as fuel and increasingly run without human ends, by selection, not by any designer’s plan. Not a conspiracy; a selection dynamic.

Operationalized: measure optimization intensity (O) — from inputs (the platform’s stated objective, the depth of algorithmic personalization, the presence or absence of user control) — independently of the capture/pathology (P) it predicts. Predict O→P, and that deliberate de-optimization lowers P. B’s integrity rests entirely on its make-or-break selection-not-design test: in cases where a designer’s intent and the market pressure diverge, the outcome must follow the pressure (systems built to “connect people” that drift into outrage machines) — otherwise B collapses into the conspiracy theory it disavows.

Status and honest revision: B is the first theory being tested — but the recent methodological deliberation sharpened it considerably. It turns out to be O-easy, P-hard: O is now genuinely measurable from newly-mandated platform disclosures (and is non-circular with P, since one comes from inputs and the other from outcomes), but P — compulsive use, polarization, misinformation, well-being — is fragmented, low-frequency, often societal-rather-than-per-platform, and in the well-being case causally contested. So the study’s inferential weight moved from a weak cross-platform comparison to within-platform natural experiments around documented de-optimization events, with sober expectations. B is the cheapest to begin, but for a humbler reason than first assumed.

Theory C — Distributed Coherence / the Commons of Sense-Making (reflexive)

Claim: because no single mind can hold the whole, an adequate theory must be distributed, living, plural, and self-correcting — a governed commons (in Ostrom’s sense), not a monograph. The project itself is the prototype; the three theories together are an instance of the third one.

Operationalized (the epistemic half only): a designed-disagreement, dissent-preserving, catches→learnings→rules process should out-catch a lone author on the same questions — testable by internal ablation (run comparable questions with designed disagreement ON versus OFF, and measure the catch rate, the provenance of catches, and how many live objections survive). This needs no data pipeline; it is runnable now.

Its two honest limits, kept in view: first, the baseline problem — in the project’s current form the “panel” and the “lone author” are the same underlying model in different prompts, so the ablation tests only the narrower claim (does structure beat no-structure?), not the grand one (does human plurality beat individual genius?); the clean baseline arrives only when the repository opens to real external forks. Second, the meaning half — whether sense-making together restores a shared “why” — is deliberately left outside measurement (Nietzsche’s cut: a method is not a meaning). C’s epistemic claim can be tested; its deeper hope cannot, and the project refuses to dress the second in the metrics of the first.

VI. The pressure-tests — thirteen symptoms in four coupled layers

The theories are stressed against the present through a set of thirteen pressure-tests, arranged in four layers — and, importantly, presented as a contested causal hypothesis, not a settled taxonomy:

M — the coherence vacuum (the master condition; whether it is the base or the summit is left open).
Drivers: D1 acceleration · D2 optimization · D3 wealth-pump / inequality · D4 machine intelligence.
Dynamics: Y1 epistemic breakdown · Y2 coordination failure · Y3 institutional / legitimacy decay · Y4 ecological overshoot.
Symptoms: S1 fertility collapse · S2 the attention / dopamine economy · S3 populism · S4 anomie / loneliness.

The arrows between the layers are the research program — which driver drives which dynamic and surfaces as which symptom, with what sign and what delay. The layering is itself contested and the disagreement is preserved: Marx rejects the driver/symptom cut as ideology (the “symptoms” are the system working as designed for someone); Nietzsche insists the coherence vacuum belongs at the summit, not the base; Le Guin notes that any ordering already encodes a worldview. Of the thirteen, only three so far carry a clearly falsifiable claim — raising that count is the central near-term task.

VII. Where it stands — built, tested, and the honest coordinate

What is built: the full reference architecture (history, landscape, goals, method, a 23-rule constitution); the panel and its recorded deliberations; three operationalized theories, each stated as a falsifiable claim with a pre-registered study design; a public reading website with a summary, a facts-and-figures page, and per-page export; a private control-room dashboard; the shuttle’s standing checks and its compressed-memory artifact; and gated study scaffolds for all three theories.

What is tested: nothing — yet. This is the honest center of the report. All three operationalizations are stated, not run. The project has produced a large, coherent body of argument and research design and zero empirical results. That is the live form of its own named risk (grandiosity). The pivot now underway — running Study B as the first empirical test — is what turns the project from “an impressive design for finding out” into “finding out.”

The honest fragilities, named plainly:

The “distributed commons” has so far been, in practice, one human and one model in a conversation. Theory C’s whole bet is untested until real external vantages arrive, which is why the repository’s eventual opening matters for testing, not just for sharing.
The chair-resolution-bias (C-006) is real and still unmeasured: the plurality could be partly cosmetic, and the metric that would expose it (where catches actually come from) has not yet been computed.
Campbell’s Law lurks at the empirical turn: the moment a measure becomes a target it gets gamed — which is literally Theory B’s mechanism. The study’s central methodological risk is the very phenomenon it studies.

The immediate frontier: the flagship open question is operationalizing “the gap” (Theory A) into a tested claim — the heaviest lift. Study B’s pre-registration has been finalized and author-ratified (Gate 2 open), but its empirical run needs external data the executor does not have — so its pre-registered pipeline is built and frozen while the world-test waits. Study C’s ablation is the one that needs no data pipeline and can run soonest — and has now had a first, directional-only pilot. The standing meaning-governors — is understanding structural or meaning-first; is there one master condition or an irreducibly coupled field; is a better system the cure or the disease; is the reflexive theory a third path or a dodge; whose crises are on the list; is the gap a pathology or a feature — are kept open by design. They are not problems to be solved before proceeding; they are the tensions that keep the project honest.

VIII. Coda — what would make it real

The project’s most useful self-description is also its most modest: it is, so far, a design for finding out — with an explicit honesty discipline whose reliability only an outside reader can judge (the self-grade is not the project’s to award) — and it has not yet found anything out. Four things would change that, in order of leverage:

Run Study B. The single act that turns design into evidence. Everything else waits behind it.
Run Study C’s ablation and compute the catch-provenance metric — even if it embarrasses the chair. The honesty is the product.
Open the repository when the skeleton is stable — the only way to get a real, unentangled baseline for testing the reflexive theory.
Keep the loop fast — promote what is learned the moment it is learned; measure whether the thing is actually improving, not merely moving.

Whether any of the three theories survives contact with data is unknown, and that uncertainty is the point: a project that could not be proven wrong would have failed by its own first principle. What it has demonstrated so far is narrower but not nothing — that a plural, self-correcting, honestly-bounded process can be built and sustained, can catch its own errors including errors about itself, and can hold real disagreement without collapsing it. Whether that process can also produce a true account of the present is the question the next phase, finally, begins to test.

Report ends. For the working documents behind every claim here — the full history, the landscape survey, the three operationalized theories, the pressure-test causal hypothesis, the open questions, and the reflections — see the repository. Author: Hulki Okan Tabak — with Claude · License: CC BY-SA 4.0.