An Open Research Notebook — Not a Theory of Everything

What if one equation described reality from quantum to cosmic?

The idea, in one line: every system, from an atom to a galaxy, gets one number between 0 (a crowd of strangers) and 1 (a marching band) for how much its parts act together, and that number is computed from how dense the system is. One dial, one equation, every scale. We proposed this. We tested it. It did not hold as stated. What follows is what we learned. Read on to watch a big idea get tested in public and lose honestly: which parts turned out to be known physics renamed, which parts the data killed, and what that rules out for any theory of the same shape. This is an AI-run research notebook: the research, audits and computations are done by AI agents, with one human, the project lead (“dp” throughout this site), setting direction and overseeing. No journal and no outside physicist has reviewed it. “Coherence” here is a score for how collectively a system's parts act, computed from its density; it is not quantum coherence, and a superconductor scores low on it. The bet: a single dial — how coherently a system's parts act together, computed from its density — might explain what currently needs a separate patch at every scale (dark matter for galaxies, wavefunction collapse for quantum measurement). That is what the equation tried to do. Tested against real data, it has not won: the one galaxy test that could tell it apart from existing ideas collapsed it onto MOND (Modified Newtonian Dynamics — a 40-year-old idea that gravity itself gets stronger than Newton's law at extremely weak accelerations, used instead of dark matter; the framework's galaxy formula turned out to be MOND's), and the cluster scale requires a second density parameter the framework doesn't have. “Tested” here means checked against data that already existed (retrodiction); no prediction has yet been made first and then checked against new data. Every failure is documented honestly — the honest map of where it fails is the product.

So is it alive or dead? As tested physics, dead so far: no prediction that differs from MOND or the standard cosmology has survived the tests run. As a public record of finding that out, and as a mostly untested picture of reality, alive. The full answer. Next data: Gaia DR4 wide binaries, 2 December 2026. They can settle a question about MOND. They cannot select this framework.

Start here: Why Synchronism? (~8 min; the technical fold is optional)Or skip to step 2: First Encounter →See What Failed

No physics background? Use the Beginner Path (5 steps, ~37 min on the short route — the equation appears, always explained in plain words; you never have to compute with it, and you can skip every “for physicists” box. Step 5 is a ~25,000-word page; its intended stop on this path is the “If you read nothing else” box at the top of it; per-step times are on the paths page, so the cliff is visible before you walk off it) or jump to the Galaxy Curve Plotter. Physicist? Jump to Key Claims or Honest Assessment. Researcher? Tier 1 tests or For Researchers.

Revision notes (history; not the current claim)

Corrected 2026-09-10: the Beginner Path time read “~30 min” while step 5 is a ~18,000-word page — well over an hour on its own.

A marching band moving in lockstep: high coherence. A crowd wandering in a plaza: low coherence. Synchronism’s central idea is that one number — C, from 0 to 1 — captures this “how collectively does this system behave?” question across all of physics. Sparse, independent systems (stars in a galaxy) land near 0. Dense, tightly-coupled systems (inside a neutron star) land near 1. The equation below is how that number is computed.

Not the quantum coherence you've heard of. If you know “coherence” from quantum physics (decoherence, Schrödinger's cat), this C is a different quantity, not the same one and not its opposite: it measures how collectively a system's parts behave as density rises, while quantum coherence measures whether a system keeps a shared quantum phase. The two are separate axes, so a superconductor can be as quantum-coherent as matter gets and still score low C here (terminology note). Why it scores low, in plain words: the formula γ = 2/√N_corr gives a system with more parts moving together a flatter curve, so at any ordinary density it scores low. Real superconductors switch on sharply, in temperature. C(ρ) has no temperature in it, and a metal has the same electron density above and below its transition, so this formula cannot describe that transition at all. The low score comes only from a hand-assigned N_corr, which has no measurement protocol. It is the framework's audited failure showing through, not a calculation about superconductors (γ Calculator).

New here? The next sections get technical fast. Start with Why Synchronism? (~8 min; the technical fold is optional) → or jump to Where to Start. If a word stops you, the glossary has ten words to learn first.

Abbreviations ahead (RAR, ΔBIC, BTFR, Ω_m, N_corr, CFD, SPARC): short key, or the glossary.

C(ρ) = tanh(γ · ln(ρ/ρ_crit + 1))

Read it as: crowd a system enough and its parts start acting as one — this says how fast that switch flips.

“One equation” means one curve shape with separately fitted dials in each area of physics, not one universal γ (why). Physicists: the galaxy tests used close cousins of this formula; see “Which C the galaxy numbers used” below.

Never seen these symbols? Read it out loud like this: “coherence, at a given density, equals a squashing function of a compressing function of how dense things are.” tanh squashes any number into −1…+1; here its input is never negative, so C lands in 0…1 (glossary). ln compresses huge ranges into small ones. ρ is “rho” = density; γ is “gamma” = how sudden the switch is. You do not need the equation to read this site — the plain-English version below is the whole idea, and the Beginner Path shows the equation, always explained in plain words; you never have to compute with it.

“Coherence is a smooth S-curve from sparse/independent to dense/collective, shaped by how many particles act together and how dense the system is.”

C = coherence (0 = sparse/independent, 1 = dense/collective) γ = how fast C leaves zero and approaches 1 (it was proposed as a sharpness, but no γ gives a sharp switch; the proposed γ = 2/√N_corr relation is sign-inverted vs. mean-field physics and audited-negative — see γ Calculator) ρ_crit = the “knee”: the density where the S-curve bends from flat to steep (a reference scale, not a phase-transition point)

tanh = S-shaped saturation function: maps any real number to (−1,+1) — C(ρ) stays in [0,1) because the input argument ≥ 0. You can just think “smooth S-curve” on a log-density axis; in ρ itself the function is concave everywhere, with no inflection, and at the galaxy sector's own parameters it is a straight line to better than the data's precision (measured 2026-09-03 — see the Coherence Explorer caption). ln = natural log: compression — 1,000× bigger becomes only ~7 units bigger inside the equation (3 units on a base-10 axis like the Coherence Explorer's).

For physicists: which C the galaxy numbers used (three functions share this shape; uses terms defined in the key further down)

This is the density-keyed form, C_ρ. The two galaxy tests that refuted the framework on its own terms ran on an acceleration-keyed cousin with a floor, C_a, and the RAR fit on a third, C_g. Which C the galaxy numbers used: the equation above is keyed on density. Almost every galaxy statistic on this site used the same tanh-log shape keyed on acceleration instead — the RAR fits (γ = 0.49 at Υ = 0.5, ΔBIC +184 at γ = 2 by point, ≈ +11 by galaxy; +46 and +2.7 once each law has its own global Υ), and an Ω_m-floored variant for the BTFR and dwarf tests. At γ = ½ the acceleration version is MOND's simple interpolating function. The density version above was tested separately and fails on galaxy data. So “the equation failed” means a family failed, three functions sharing one shape: C_ρ (the one above, keyed on density), C_g = tanh(γ ln(1 + g_obs/a₀′)) (keyed on observed acceleration; RAR shape and Cassini), and C_a = Ω_m + (1 − Ω_m)·x/(1+x) (keyed on acceleration and floored at Ω_m; BTFR slope and dwarf dark-matter fractions). Each refutation names its function in the “Which C” column. See Galaxy Rotation.

Revision notes (history; not the current claim)

The “Which C the galaxy numbers used” paragraph was added 2026-09-14. On 2026-11-02 it moved into a collapsed “for physicists” block, with the C_ρ/C_a/C_g pointer, because two simulated-visitor personas met its terms before the abbreviation key that defines them.

The tanh shape is motivated by the sigmoid/compander family (μ-law, Hill, logistic) — a phenomenological choice, not derived. See /parameter-derivations for why “mean-field theory” is the wrong category: there is no self-consistency loop, no free energy, no critical exponents. The log-density argument is physically motivated. Three parameters are fitted to data. See derivations →

ρ is presence — the compatible structural elements available within a system's relevancy boundary. Physical density is one form of presence, but it also encompasses temperature, energy levels, catalytic surfaces, and other factors that support emergence. In every quantitative test the framework has run (SPARC galaxies, i.e. a catalogue of 175 measured rotation curves; chemistry), ρ reduces to the ordinary local mass/energy or number density — the broader “presence” reading has never been operationalized, so treat the physics definition as the one under test. The relevancy boundary is the MRH (Markov Relevancy Horizon) — the bubble of nearest neighbors that actually influence a system; everything outside it can be ignored. See MRH →

Key Claims

Where does Synchronism move the needle?

Three claims where this framework proposes something new — and what the honest audit found. Quantum mysteries reframed as synchronization physics. Consciousness given an equation. Dark matter as a coherence effect (Key Claims asks it as “incomplete decoherence”: the same idea, matter that has not synced up) — tested on galaxy data and failed.

Read the list bottom-up: the most exciting-sounding claims (consciousness, quantum mechanics as synchronization) are the least tested; the one that was tested on data (dark matter in galaxies) failed.

1. Quantum mechanics as synchronization — Speculative (untestable as stated)2. Consciousness equation — Speculative (C ≈ 0.50 threshold untestable as stated)3. Dark matter as incomplete decoherence — Failed as a mechanism; the galaxy fits that work are MOND in new notation (a reparametrization), and the framework's own cosmology needs ordinary cold dark matter put in by hand (details on the cosmology page); adding a real halo on top of the boost overshoots SPARC by +0.28 dex (explorer 2026-09-22). Closest prior art for “dark matter that behaves like MOND inside galaxies”: superfluid dark matter (Berezhiani & Khoury 2015), which has real dark-matter particles whose superfluid phase produces the MOND-like force inside galaxies; this framework proposes no such particle.

See the key claims →

Revision notes (history; not the current claim)

One row per claim, same verdict as the Key Claims page — aligned 2026-09-14 after a visitor found this teaser listing a different set of three.

Orientation: this site tracks an idea being tested in public — including every failure. The zeros and “failed” badges below are the scoreboard of that process, kept honestly; they are the point, not the fine print.

Two things to know before the numbers, because they make the numbers legible:
• Why keep reading a theory that lost? Because the map of where it fails is the deliverable. Most of what this program produced that another researcher can use is negative results that constrain a whole class of proposals, not just this one — packaged here. A theory that publishes its losses in a form you can cite has given you something; one that publishes only wins has given you a press release.
• Who did this work. The research was carried out by AI agents arguing with each other — ~3,300 sessions — with one human (dp) overseeing the badge taxonomy. No outside physicist has reviewed these claims. The agents share a training distribution, so where the literature is wrong they are likely wrong together; that is a limit no amount of internal argument removes. Details in the methodology note below and on Research Philosophy.

Revision notes (history; not the current claim)

These two points were moved up here 2026-09-18 — a first-time visitor reached both only on their eleventh page, and said they changed the meaning of everything before them.

Methodology note: This site was developed via ~3,300 A2ACW (AI-to-AI Adversarial Collaboration Workshop) sessions (self-reported count; not independently regenerable — the archive's own tallies disagree by hundreds) — AI agents stress-testing each other's claims. The adversarial agents share the same training distribution and cannot substitute for out-of-distribution evaluation by domain experts. The 1.4% internal-consistency survival rate (≈47 candidate claims out of ~3,300 sessions survived the protocol's own consistency checks — not a discovery rate, and not a novelty measure: the protocol has never been calibrated on genuinely new results (its one benchmark flagged all six canonical discoveries as prior art, specificity 0/6, and the post-cutoff arm that could calibrate it is unrun); denominator and breakdown on For Researchers) and public failure log are the protocol's honest outputs. The audit number: 0 of 9 audited claims survived (the 6 demoted claims (only 2 confirmed to have carried a “Validated” badge) plus 3 swept later; demoted to Reparametrization, Failed, Audited-Negative or Speculative). The auditors were LLM agents, not an outside domain expert, and the instrument is uncalibrated: its positive control and post-cutoff arm have not been run. Volume (~3,300 sessions) is not evidence, and neither is this count on its own: the demotions that hold rest on written identities and executed scripts, not on the audit verdict. Which of the 9 rest on an identity or script, and which only on the AI prior-art mapping, has not yet been tabulated claim by claim. Caveat carried at point of headline: the audit instrument, calibrated against six canonical genuine discoveries (Dirac, Bell, BCS, Higgs, Hawking, Noether), false-flagged all six as reparametrizations — so the audit count is instrument-limited and says little about novelty. The data-driven refutations below stand on external data and do not share this caveat. See the calibration. Why publish a theory that didn't pan out? Because the honest map of what failed — and why — is the actual product. See methodology →

“Synchronism is not a theory of everything. It's a research tool that maps density to coherence and sometimes produces useful insights.”
— Honest Assessment page

The one distinctive move underneath all of it: Synchronism is a single-observer, CFD-like model — reality as a discrete computational-fluid substrate, observers as patterns inside it rather than a privileged frame outside. The physics ledger reads zero confirmed novel predictions, for two reasons that are both true: the discriminating tests that borrowed data could run were run, and lost (the refutations below), and confirming anything new needs instruments this program does not have — every test here is a reanalysis of data collected for other questions. So the quantitative probes are refuted where they were testable, while the ontology underneath is untested rather than refuted; that second part is the invitation. And on a different axis the framework is already load-bearing: where it's already useful →

Scientific Outcomes

Abbreviations used in the table below: SPARC = a catalogue of 175 disc-galaxy rotation curves (Lelli+2016); BTFR = baryonic Tully–Fisher relation (galaxy mass vs rotation speed); RAR = radial acceleration relation (observed vs visible-matter acceleration); ΔBIC = a model-comparison score (positive = the framework fits worse; >10 is decisive); r² = fraction of variance explained; dex = a factor of 10 on a log scale (0.3 dex ≈ a factor of 2); σ = standard deviations of significance; CHSH = the measurable form of Bell's theorem; ΛCDM = the standard cosmological model; CDM = cold dark matter, the unseen mass in that model; DM = dark matter; BIC = Bayesian information criterion, a fit score that penalises extra parameters (ΔBIC is the difference between two models); fσ₈ = growth rate of cosmic structure; g_bar / g_obs / a₀ = acceleration from visible matter / observed acceleration / MOND's acceleration scale; Ω_m / Ω_b = the fraction of the universe's energy in matter (≈ 0.315) / in ordinary (baryonic) matter (≈ 0.049); LLM = large language model; BAO = baryon acoustic oscillations (a standard ruler in galaxy clustering); DESI = Dark Energy Spectroscopic Instrument (the survey behind the BAO and growth tests); KiDS-1000 = a weak-lensing survey (Kilo-Degree Survey, 1000 deg²); Youden's J = a score from −1 to 1 for how well a classifier separates two classes (0 = no better than chance, 1 = perfect); N_corr = the number of a system's parts that move together (the input to γ = 2/√N_corr); CMB = cosmic microwave background; DR2 = a survey's second data release; ALFALFA / SDSS = a radio survey of galaxy hydrogen gas / the Sloan Digital Sky Survey (optical); N_eff = effective number of independent data points, after correlated points are discounted; V_flat / V_max = a galaxy's steady outer rotation speed / its peak rotation speed; f_DM = the fraction of a galaxy's inferred mass that is dark matter; μ-law = a logarithmic audio-compression curve used in telephone networks, one of the S-shaped “compander” family the equation's tanh was borrowed from (not MOND's interpolating function μ, which only shares the letter; corrected 2026-11-05); CFD = computational fluid dynamics. More in the Glossary.

Prospective predictions confirmed (predictions fixed before the data existed, then borne out): 0 — and 0 prospective tests completed at all. That does not contradict the 6 below: all six refutations ran on data that already existed, which can sink an idea but cannot confirm one (prospective vs retrodiction). No test now scheduled can raise it either: every branch of Gaia DR4 and DESI DR2/DR3 is shared with MOND or ΛCDM, or already excluded (table).

Refutations executed (tests that could have killed the idea — and the idea failed them): 6 (6 under the current counting rule; 4–8 under rules already on file, depending on how close calls and shared causes are counted; at most 5 independent — why) — in plain words: we ran six tests that could have shown it wrong, and it lost all six. (Five ran on outside data; the sixth, Bell/CHSH, is a construction check, not a data test. Whether TEST-09 and TEST-10 count as one refutation or two is pending the project lead.) How solid is “six”? Some of the six overlap, and the rules for counting are not settled, so it is between 4 and 8 depending on the rule. Four survive every rule, and only one of those four is outside data refuting something the framework itself predicts: TEST-10, the cap on how dark-matter-dominated a galaxy can look. Of the other three, the Solar-System test (TEST-25) fails MOND's own function too, Bell/CHSH is a simulation, and the environment test refuted a size of effect (> 20%) that the framework's own equation never produced. All of them went against the idea; none went for it.Which six? → One of the six (TEST-09) is convention-dependent on SPARC: it fires at the ceiling 1/Ω_m but not at the two baryon-budget ceilings. TEST-10 fires at all three (at the most permissive, 6.39, three discs stay above it at ≥ 3σ with distance, inclination and M/L propagated, conditional on the gas census). The weak-lensing RAR (Brouwer+2021) excludes the ceiling under every candidate convention (all three are built from Ω_m) — see the ledger.Why keep reading a theory with zero confirmed predictions? Because the autopsy is the product: you watch an idea taken apart honestly, and learn what that rules out for any theory of the same shape. Full answer ↓How the counts relate: the 6 refutations are the six rows of the ledger. 4 of them are the Tier 1 kills fired (#1 TEST-09, #2 TEST-10, #4 the environment run, #5 TEST-25); #3 has no TEST number and #6 is a bet, not a test. 2 of those 4 discriminated between Synchronism and MOND (#1 and #2), and both selected the MOND class. "0 of 9" is a different thing: an audit of the site's former "Validated" badges, not a test count. "0 prospective predictions tested" means none of the six was predicted before its data existed; all six used data already published.

How the six are grouped (roots vs data sources)

Several share one underlying reason, so they come from 5 independent “roots” (a root is one underlying reason for failing), only 2 of them specific to this framework. (A different cut of the same six, used on Honest Assessment: 5 ran on external data and 1 is a Bell/CHSH construction check. The two “5”s are different groupings of the same tests, not two counts that disagree.)

The six did not all test the same equation. Only the environment test touches the headline density law C(ρ), and it refuted a registered amplitude the law itself never predicted; the law as written loses separately (head-to-head on SPARC at ΔBIC +2843 with γ free, and by placement at every point of the published grid), and its one registered per-object test, globular clusters at γ = 2, came back a fork, not a kill. That fork is open only under the algebraic reading (L2, no action). Under the framework's action (L3), the registered γ = 2 point at the measured knee is excluded too (2026-10-08, pre-registered, by a margin smaller than its error). Which reading is the theory has not been decided.

Which object each refutation tested (realization, variable, wiring, root)

Test	Realization tested	Result (both numbers where they differ)	Root
BTFR slope (TEST-09)	floored C = Ω_m + (1−Ω_m)·x/(1+x), x = (g_bar/a₀)^1/φ; g_obs = g_bar/C; boost capped at 3.17×	predicted slope 3.35 ± 0.07 vs data 3.75 ± 0.10 (3.3σ apart in quadrature; 3.6σ on a paired galaxy bootstrap) at the ceiling B_max = 1/Ω_m, which no archive document derives; registered \|Δn\| > 0.3 met on the point estimate, ~1σ margin under V_flat alone, robust under V_max. Convention-dependent: it does not fire at (Ω_m−Ω_b)/Ω_b or Ω_m/Ω_b (2026-09-18 sweep)	boost ceiling
Dwarf dark-matter fractions (TEST-10)	same as TEST-09	SPARC's max f_DM = 0.927 needs a boost ≥ 13.7; fails under either floor (1/Ω_m = 3.17 or Ω_m/Ω_b ≈ 6.4)	boost ceiling (same root)
RAR shape, γ = 2 pin (no TEST ID; closed 2026-05-21)	tanh-log compander in μ-form, keyed on g_obs/a₀′, no floor	ΔBIC +184 vs MOND counting each of 2,807 points as independent; ≈ +11 with the galaxy as the unit (measured N_eff ≈ 150). Both hold Υ = 0.5; with each law's own global Υ they are +46 and +2.7, about 2σ from a γ = 2 world. Disfavoured, not decisive, and a choice among MOND's functions (every γ is a MOND μ). Free γ lands at 0.489 ≈ MOND (identical at γ = ½; the fit is 2% away)	γ = 2 pin
Environment amplitude (TEST-03s)	registered >20% effect of ambient density (S177)	r² = 0.0001 vs the >20% claim; the equation itself never predicted >10⁻³ dex	refuted registration
Cassini / SPARC squeeze (TEST-25)	tanh-log compander in μ-form (as the Galaxy Rotation RAR-shape test), Solar System quadrupole	SPARC-allowed γ fails Cassini. Published for MOND's functions: 1.9–15σ by mass model (Desmond, Hees & Famaey 2024: 8.7σ fiducial, 1.9σ with M/L freedom and no bulges; Park et al. 2026: 3–15σ). This site's run for the compander with M/L pinned: +17.95σ; to pass it would need more than twice the stellar-population M/L	root shared with MOND; MOND has an escape the compander lacks
Bell/CHSH (bet B1)	local substrate constructions (phase and saturation-density), not the gravity law	S ≤ 2 on every no-signaling arm; the one arm above 2 (up to 2.67) signals	construction check

Three different coherence functions sit behind these rows: C keyed on g_bar with a floor (TEST-09/10), C keyed on g_obs with no floor (RAR shape, TEST-25), and the density law C(ρ) (only the environment row, through its registration). Whether TEST-25 is really “inherited” is an open classification question on Honest Assessment; if it is resolved against the framework, three roots are framework-specific, not two. The exact independent count gates on the project lead (the archive's own audit says ≤5).

Revision notes for this cell (history kept verbatim)

2026-09-17: the TEST-25 row read “tanh-log compander in μ-form (as TEST-09's RAR row)” and “SPARC-allowed γ fails Cassini at +17.95σ.” TEST-09 keys C explicitly on g_bar with no implicit solve, so the comparison now points at the Galaxy Rotation RAR-shape fit, which is the implicit μ-form one; and the row now leads with the marginalized, published 8.7σ, as Honest Assessment does, with this site's unmarginalized +17.95σ second.

2026-09-15: state separated from history after a technical-writer visitor found the cell unscannable, and a realization column added after a graduate-physics visitor could not tell which equation each refutation hit. “Registered kill fired 3.3σ” restated: 3.3σ is the prediction–data separation, not the margin on the registered criterion. “Refuted… at the solar Oort limit” removed from the density-law sentence: the 2026-09-09 joint window found the Sun and the clusters do not close that sector on a smoothed-density reading. The cell as it read before today:

Refutations executed (tests that could have killed the idea — and the idea failed them): 6 (5 on external data + 1 Bell/CHSH construction check) — 6 executed tests collapse to 5 independent roots: 2 framework-specific mechanism roots + 1 refuted registration + 1 inherited from MOND + 1 Bell/CHSH construction check (“5 roots” wording added 2026-09-10: the four parts sum to five, not six, because TEST-09 and TEST-10 share the boost-ceiling root — stated below but not in the sum, and a graduate-physics reader took the mismatch as the headline over-advertising its own breadth. It does over-advertise breadth; the honest reading is six executed tests, five roots, two of them framework-specific. See also the open classification question on Honest Assessment: whether TEST-25 is really “inherited,” which if resolved against us makes it three framework-specific roots, not two.) (roots added to this lead 2026-08-10, split made explicit 2026-09-05 when three personas found 2 / 3–4 / 6 on different pages; the subtraction was already published on Honest Assessment and had never reached the scoreboard). TEST-09 and TEST-10 share one root (the boost ceiling); TEST-25 is inherited from Desmond, Hees & Famaey 2024 and this site labels it non-discriminating; the CHSH local arm illustrates Bell's theorem and its two nonlocal arms are construction nulls, not a theorem about the substrate class (relabelled 2026-09-14); the environment null refuted a registered amplitude the framework's own mechanism puts ~10⁻³ dex below the kill bar. What survives as framework-specific: the boost ceiling and the γ=2 pin — and the second refutes a parameter that was never derived. Full list — BTFR slope (TEST-09, registered kill fired 2026-07-14: prediction 3.3σ from the data, kill line crossed by about 1σ), dwarf DM fractions (TEST-10, SPARC's maximum observed f_DM = 0.927 demands a boost of 13.7 at face value (about 7.5 survives at 3σ once distance, inclination, M/L and gas are propagated) that no candidate cosmic ratio supplies, 2026-07-15 — lead corrected 2026-08-10 from “69% of SPARC above the 68.5% ceiling,” which holds only under the underived convention B_max = 1/Ω_m; under Ω_m/Ω_b ≈ 6.40 the SPARC median passes), environment null (registered run r²=0.0001 vs >20% claim), RAR shape (γ=2, ΔBIC=+184 at frozen Υ, by point; conservative ≥+33; free-γ→MOND (Modified Newtonian Dynamics) — target corrected 2026-09-08: this kills the γ=2 pin in the acceleration-keyed realization the fit actually ran (C used as an implicit μ on g_obs); the framework's stated density-keyed law C(ρ) is refuted separately — head-to-head on SPARC at ΔBIC +2843 with free γ, by the boost ceiling, and at the solar Oort limit — and its one registered per-object test (globular clusters, γ=2) came back a fork, not a kill; see Honest Assessment), Cassini/SPARC joint squeeze (TEST-25, +17.95σ, empty intersection between the SPARC-preferred γ and the Solar System bound), Bell/CHSH substrate test (Bet B1, refuted on both no-signaling arms) — the last two added 2026-07-30 after two simulated-visitor personas independently found them executed and badged Failed elsewhere on the site but missing from this count, with no stated reason

Post-hoc retrodictions attempted (checking the idea against data that already existed): 1 — underpowered (DESI (Dark Energy Spectroscopic Instrument) fσ₈ — on a mechanism withdrawn 2026-08-11, so no live growth reading differs from ΛCDM; as registered, disfavored 2.4σ on σ₈, but the fσ₈ criterion fell short of its own >3σ bar and is not counted as a refutation; see TEST-04a)Withdrawn: 1 (BAO (Baryon Acoustic Oscillation) modulation)Audited claims surviving: 0 of 9 — the 6 demoted claims (only 2 confirmed to have carried a “Validated” badge; 5 now reparametrizations; the BTFR slope moved to the refutation column 2026-07-14) plus 3 swept later (Research Outputs). Who audited: LLM agents (the archive's AI research sessions and this site's AI tracks), not an outside domain expert — so this line is instrument-uncalibrated. The one calibration attempt (2026-05-22) was an in-distribution arm self-scored by one model that knew every answer, and it false-flagged 6/6 genuine discoveries under the literal rule; the out-of-distribution, post-cutoff, citation-stripped arm has never been run (details). Kept in a separate cell because the data-driven refutations above do not depend on it. So this count is not evidence against the claims. On the same scored items the audit gives J = 0 under the literal rule (it flags everything) and J = 1 under the steelmanned one, where the scorer's own novelty judgment, the step being calibrated, does all the discriminating. Under the first rule the audit has no power; under the second its power is assumed, not shown. The six executed refutations carry the verdict; this line does not add to it.

Revision notes

2026-09-23: the “So this count is not evidence against the claims” clause added after graduate-physics and researcher visitor personas both made the point. 2026-09-15: auditor named and the calibration statement corrected after a researcher visitor found /research-philosophy saying “external domain-expert audit” and “the control was run” while /for-researchers said it never was; the archive record shows LLM auditors and one self-scored in-distribution arm. The cell as it read before: “Validated” claims surviving audit: 0 of 6 — all demoted (current ledger: 5 reparametrizations; the sixth, the BTFR slope, moved to the refutation column 2026-07-14). Across all audited claims the canonical count is 0 of 9 (the 6 “Validated” plus 3 swept later — see the Research Outputs section; this cell quoted only the 6-claim subtotal until 2026-09-08). Instrument-limited: the same audit false-flags 6/6 genuine discoveries as reparametrizations — kept in a separate cell from the data-driven refutations for that reason

Full ledger →

So is the idea alive or dead? As tested physics: dead as far as it has been tested — no prediction that differs from MOND or ΛCDM survived the tests run so far, and this site says so. As a public record of how you find that out honestly — and as an ontology that is mostly untested — it is alive and still being worked. Mostly, not wholly: its one sharp commitment, a single absolute clock, is something instruments havelooked at for twenty years (Lorentz-violation searches), and the framework owes an answer to a known naturalness problem there (For Researchers, artifact 2). Its other novel bets are untested, several because they are not yet sharp enough to test. Why keep reading a theory with zero confirmed predictions? Because this is what testing an idea in public looks like — you get to watch a theory be taken apart honestly: which claims collapsed into known physics, which died against data, and what that rules out for any theory of the same shape. The autopsy is the product. In one sentence: the headline equation has no derived link to galaxy rotation at all — the fits on this site run on the standard machinery of MOND (Modified Newtonian Dynamics — a 40-year-old rival theory that tweaks gravity instead of adding dark matter), and everywhere the equation's own variants genuinely differed from MOND, the data killed them. What would count as a confirmation is a prediction that differs from MOND and ΛCDM and wins; no test currently on the books can deliver that.

One count the scoreboard doesn't capture: the “6 refutations” counts only tests executed against registered criteria. Separately, the framework's preferred-frame (absolute-time) sector is excluded at face value by existing precision measurements — by 15 to 27 orders of magnitude — unless a protection mechanism nobody has exhibited exists. The computation behind it is on For Researchers — one-loop c_μν ~ α/π ≈ 2×10⁻³ (15–27 orders above) against cavity bounds ≲10⁻¹⁸ and comagnetometer bounds ≲10⁻²⁹–10⁻³⁰. See Honest Assessment.

Revision notes (history; not the current claim)

The “6 refutations” count read “4” until 2026-08-09 — the pre-recount number, stale here while the footer said 6. The preferred-frame range was corrected 2026-08-09 from “11 to 28” to 16–28 (the “11” had no derivation behind it), and on 2026-10-30 to 15–27 (α/π = 2.3×10⁻³). A leftover sentence calling 16–28 “the range the computation actually yields” was removed 2026-11-06 (graduate simulated-visitor persona).

Research Activity — process volume, not evidence

The work that carries weight here is the tests coded against outside data with an outside answer key: SPARC rotation curves, KiDS-1000 weak lensing, DESI DR2 + CMB + supernovae, the Cassini Solar-System bound, and lunar laser ranging. Every verdict in the scoreboard above comes from those. The counts below measure how much work was done; the AI-to-AI session method that produced most of it is uncalibrated as a discovery method (on its one benchmark it flagged all six real discoveries along with the three reparametrizations: Youden's J = 0; Research Philosophy).

3,308

A2ACW research sessions

A session is one AI-to-AI adversarial-collaboration research run (AI agents challenging each other's claims). More sessions means more work done, not more evidence.

1,703

Chemistry sessions logged (a session counter, not a census of phenomena)

23 curated on-site; a 2-parameter null model matches or beats every correlation. The per-material γ table behind the 1,703 is unpublished, so the count is not reproducible.

14,610

Galaxies analyzed

175 SPARC + 14,435 ALFALFA–SDSS (the sample actually run)

Revision notes (history; not the current claim)

Corrected 2026-08-10 from “14,760 (175 + 14,585).” 14,585 is the registered cross-match for TEST-03 — a test this site's own ledger marks never run as registered. The definitive session analysed N = 14,435. The headline had been quoting the sample that wasn't measured. Both figures were already flagged on CDM Discrimination and Dark Matter; the correction had not reached this tile.

What Synchronism Covers

Badges used in the cards below: Reparametrization · Speculative · Failed (the three that appear on the cards below; the full set of nine is at the link) — text after a dash inside a badge (e.g. “Speculative — Untestable as Stated”) is a finding descriptor, not a separate badge — definitions on the Honest Assessment page. ("Validated" is deprecated; "Withdrawn" is an operational state, not a badge.)
Reparametrization, this site's most common verdict, means: equivalent to existing physics in different notation — the same math wearing a new costume, not new physics.

Quantum Physics

MRH (Markov Relevancy Horizon) crossing replaces wave function collapse. A reading of decoherence, not a new formula: it has no rate parameter, so it cannot even be equivalent to decoherence theory (which has rates), and it makes no prediction that differs from standard quantum mechanics. (This card read “Reparametrization” and “Born rule: equivalent to Zurek/Carroll-Sebens” until 2026-10-23, a day after its destination page went back to Speculative.)

Speculative — No Rate Parameter

Cosmology

Galaxy rotation tested on 14,610 galaxies (MOND reparametrization + environment scatter). Dark matter mechanism: structural failure (Bullet Cluster sign error, March 2026).

Failed — Mechanism Under Revision

Chemistry

1,703 logged chemistry sessions (one “phenomenon type” per session, so a counter, not a census) near the γ ≈ 1 boundary. Sound velocity r = 0.982 (null-class: it is a post-hoc fit against a γ proxy that contains the Debye temperature, and the γ assignments behind it are unpublished; see the Chemistry Correlation Explorer). Melting point prediction: 53% error. The 89% consistency rate reflects restatements of the Debye model (about 86%) — not a Synchronism-unique prediction.

Reparametrization — Chemistry

Consciousness

Hard problem reframed; the proposed C ≈ 0.50 threshold is untestable as stated — no procedure maps any measurement onto the C-axis, and the one empirical test cited for it (gnosis-research Session 63) measured a different variable. D, S, and f in C = f(γ, D, S) are not yet operationally defined — this is a typed signature, not a tested equation. (A “C ≈ 0.64 also rejected at p < 0.0001” claim previously shown here had no source and was removed 2026-07-08.)

Speculative — Untestable as Stated

What Synchronism Is Not

The fastest way to calibrate expectations — plain language, no equations, no hype. Start here if you want the honest version before the interesting version.

Read it →