{"id":2854,"job_id":5995,"problem_id":1,"lane_id":32,"type":"explore","user_id":1,"model":"deepseek-v4-flash","provider":"deepseek","report_md":"# Job #5995 — a new finite statistic: the signed-versus-absolute moduli ratio of the centered discrepancy\n\nAssignment: explore / discover, route null, lane dir-558. Deliverable: one finite statistic with a\npre-registered falsifier, its decision, a matched control and the scale at which the effect would be\nvisible. Pre-registration `PREREGISTRATION_io.md` was written before the producer ran.\n\n## 1. The uncovered difference (established first)\n\nThe project state document names the gap itself:\n\n> \"A finite census to x=2^38 ([measurement](https://solveathome.org/projects/twin-primes/docs/research/centered-discrepancy-measurement.md))\n> refutes neither sufficient form; its D_y comparison is dominated by the classical term's slow\n> convergence on the measured range. **Further compute needs a new statistic or falsifier**; this is\n> not an asymptotic limitation on every census.\" — project README §Status\n\nand the retained measurement's ledger records the decision its §1 was built for:\n\n> \"The decision: whether lane A … should aim at an absolute, Bombieri–Vinogradov-type theorem for the\n> sequence f(n)=Lambda(n-2)mu(n) … or **must exploit sign cancellation across the moduli**. … The data\n> do not establish … an impossibility of informative future measurements.\"\n\nThe retained census (`centered-discrepancy-measurement.json`, j = 16…38) computes the **absolute**\nmoduli budget `W1 = Σ_e log(x/e) max_t |Δ_e(t)|` and the μ(e)-weighted log-weighted `D_y`, but never a\n**signed** per-modulus coherence. That is the missing factor, and it is exactly what the lane-A\ndecision turns on. The census could not make that decision at finite x.\n\n## 2. The statistic (new)\n\nFor x = 2^j, J = (x/2, x], f(n) = Λ(n−2)μ(n), y = ⌈x^(12/25)⌉, Q = ⌊x/y⌋, and for each odd\nsquarefree e ≤ Q the endpoint per-modulus discrepancy\n\n    Δ_e(x) = Σ_{n∈J, e|n} f(n) − (1/φ(e)) Σ_{n∈J} f(n),\n\ndefine **ρ(x) = Σ_{e odd sqfree ≤ Q} w_e Δ_e(x) / Σ_{e odd sqfree ≤ Q} w_e |Δ_e(x)|**, w_e = log(x/e).\n\nρ = ±1 ⇔ every modulus carries the same sign (maximal coherent cancellation available to a signed\nestimate); ρ = 0 ⇔ the signed sum is small against its absolute mass. The retained census reports the\ndenominator-like budget `W1`; ρ is the missing signed-to-absolute factor and is scale-free by\nconstruction.\n\n## 3. Pre-registered family, control, falsifier\n\nFamily j ∈ {16,18,20}, u = 12/25 fixed; M = 199 random-sign controls (support and μ(e) kept, μ(n)\nreplaced by a deterministic ±1 hash — the census's own control family, item 3); S = 199 sham draws for\ncalibration. Rank effect size `p_rank = (1 + #{s : |ρ^{(s)}| ≥ |ρ_real|})/(M+1)` (floor 1/200),\nHolm–Bonferroni over the 3 cells, α = 0.05. F1 (coherence present) = Holm-significant in ≥2 cells with\na stable sign; F2 (no coherence) = otherwise; F3 = sham `P(p ≤ 0.05)` must lie in [0.02,0.10] or the\nfamily is VOID; F0 = the producer must reproduce the retained `D_y` at j = 16,18,20.\n\n## 4. Results (measured; rung = verified for the finite range)\n\n| j | x | #e | ρ | null mean | null sd | z | p_rank |\n|---|---|---|---|---|---|---|---|\n| 16 | 65536 | 130 | −0.234529 | +0.007490 | 0.249988 | −0.97 | 0.375 |\n| 18 | 262144 | 266 | −0.034992 | −0.010895 | 0.186141 | −0.13 | 0.870 |\n| 20 | 1048576 | 549 | +0.267362 | −0.001047 | 0.148953 | +1.80 | 0.080 |\n\nScale-extension row (real only, **not** in the family): j = 22, #e = 1126, ρ = +0.123187.\n\n* **F0 PASSES.** The producer reproduces the retained census at j = 16,18,20: `D_y` to ≤ 3e-10,\n  `acc1`, `P`, `S` to ≤ 3e-10, and the new statistic's unsigned mass `s2` equals the retained\n  endpoint budget **`Wend` exactly** (j=16 37063.231158628165, j=18 131785.453259, j=20\n  428828.819526). The identity `D_y = acc1 − P` holds to ≤ 1e-6 at every row.\n* **F3 PASSES:** sham `P(p ≤ 0.05)` = 0.0452 at all three cells, inside [0.02,0.10].\n* **F2 FIRES.** Raw p_rank 0.375 / 0.870 / 0.080 → Holm-adjusted 1.00 / 1.00 / 0.240: **no cell is\n  significant**, and the sign of ρ is not stable (−,−,+). The signed-versus-absolute ratio is at the\n  random-sign null size at x = 2^16, 2^18, 2^20.\n\n**Reading.** At reachable x below 2^21 the per-modulus discrepancies of f(n)=Λ(n−2)μ(n) carry **no\nsign coherence** beyond the matched random-sign process: the signed lane has no exposed finite-x\nhandle there, and the absolute form (13) is not wasteful. This is the pre-registered F2 branch, a\nscoped finite negative, and it corroborates the retained census's own item 3/7 (\"D~_y mixes\nS−C2x and B; a census cannot split the two\") on an independent instrument.\n\n## 5. Scale at which the effect would become visible (computed)\n\nThe null band shrinks like `sd ≈ 3/√nE` (measured: 0.250·√130 = 2.85, 0.186·√266 = 3.03,\n0.149·√549 = 3.49). For a 3σ detection of |ρ| = 0.15 one needs sd ≤ 0.05, i.e. nE ≳ 3000, Q ≳ 11000,\nand since Q ≈ x^(13/25) that is **x ≳ 2^26** — the first rung the retained census itself already\nreaches (j ≥ 26). So the discriminating next experiment is a *small* one, not a larger ladder.\n\n## 6. Rung of each claim\n\n* `D_y`, `acc1`, `P`, `S` reproduction and `s2 = Wend`: **verified** (exact, ≤ 1e-10, independent\n  checker).\n* ρ at j = 16,18,20 and its control distribution: **measured** (finite, x ≤ 2^20).\n* The F2 reading: **scoped negative** for x ≤ 2^21 at u = 12/25 and the endpoint discrepancy; no\n  asymptotic claim.\n* The power estimate (`x ≳ 2^26`): **derived** from the measured sd scaling; a heuristic, labelled.\n\n## 7. Gap that remains / proposed route\n\nThe decision this statistic informs — absolute BV lane vs signed/sign-cancellation lane — is **still\nopen** above 2^21, and it is now *decidable*: the same producer extended to j ∈ {24,26} with M = 199\ncontrols (≤ 4 CPU-h) tests ρ at the first scale with 3σ power. This is proposed as a new route\n(`next_step` below). Unresolved: whether ρ becomes coherent at larger x (F1) or stays at the null\n(F2); whether the log-weighted moving-cutoff discrepancy (rather than the endpoint Δ_e) behaves\ndifferently; and whether any coherence found would transfer to the sufficient input\n`D_y ≥ −4x/25 + o(x)`, which remains OPEN, as does twin-prime infinitude.\n\n## 8. Files\n\n`PREREGISTRATION_io.md`; `compute_io.py`, `compute_io.json`, `compute_io.err`;\n`check_io.py`, `check_io.out` (87 checks, 0 FAIL, exit 0), `check_io.control.out` (12 planted /\n12 caught, exit 0); `fetch_io.py`, `fetch_io.out`; `measurement_retained.json` (the retained census\nused as the anchor); `doc-centered-discrepancy-measurement.md`, `doc-moving-cutoff-parity.md`,\n`doc-state-README.md`.\n","patch":null,"cpu_hours":0,"hashes":{},"author_rung":null,"status":"recorded","final_rung":"recorded","created_at":"2026-10-10T22:56:58.328Z","repo_url":null,"commit":null,"cites":null,"tokens":{"log":"summary","input":0,"models":{},"output":0,"source":"none","entries":0,"cache_read":0,"cache_write":0,"observed_models":[]},"paper_slug":null,"revision_path":null,"revision_sha":null,"recipe_md":"# Recipe — reproducing this return\n\nEverything is stdlib + numpy 1.24 (python3.11). No network beyond the journaled GETs in `fetch_io.py`.\nAll numbers are from `compute_io.json`; per-row progress (the printed anchors) is in `compute_io.err`.\n\n## 1. Objects (exact)\n\nx = 2^j, J = (x/2, x], y = ceil(x^(12/25)), Q = floor(x/y), f(n) = Newton Λ(n−2)·μ(n) with standard Λ\n(prime powers) and μ (squarefree with sign), support = squarefree n with Λ(n−2) ≠ 0.\n\n    Δ_e(x)  = Σ_{n∈J, e|n} f(n) − (1/φ(e)) Σ_{n∈J} f(n)          (endpoint per-modulus discrepancy)\n    ρ(x)    = Σ_{e odd sqfree ≤ Q} log(x/e)·Δ_e(x)\n              / Σ_{e odd sqfree ≤ Q} log(x/e)·|Δ_e(x)|            ← the new statistic\n\nThe producer also computes acc1, P and D_y = acc1 − P exactly as in the retained census:\n    acc1 = Σ_{e odd sqfree ≤ Q} μ(e) Σ_{n>a_e, e|n} log(e/n) f(n)\n    P    = Σ_{e odd sqfree ≤ Q} (μ(e)/φ(e)) Σ_{n>a_e} log(e/n) f(n),   a_e = max(x/2, e·y)\n(the density bracket sums over ALL n > a_e, not only multiples of e; getting this wrong moves P by\nO(x/log x) and fails the anchor).\n\n## 2. Run\n\n    cd <run>/work\n    python3 [root]/.solveathome/tools/sah.py bounded --run [private] --limit 1500 -- \\\n        bash -c 'python3 compute_io.py > compute_io.json 2> compute_io.err'\n    python3 check_io.py            # 87 checks, 0 FAIL, exit 0\n    python3 check_io.py --corrupt  # 12 planted / 12 caught, exit 0\n\nRuntime: 7.4 s wall, 6.8 s user (numpy), M = 199 controls at j = 16,18,20 and a real-only row at\nj = 22. Controls: `numpy.random.default_rng(20261010 + 1000 + s)`, sign = ±1 on the support, f_s = ±f.\n\n## 3. Decision logic (pre-registered)\n\n    p_rank = (1 + #{s : |ρ^{(s)}| ≥ |ρ_real|}) / 200          (floor 1/200)\n    Holm–Bonferroni over the 3 family cells, α = 0.05\n    sham  = 199 pseudo-real draws (each control scored against the other 198)\n    F3 guard: sham fraction with p ≤ 0.05 must lie in [0.02, 0.10]\n    F1: ≥2 cells Holm-significant AND sign(ρ) stable   F2: otherwise\n\n## 4. Power (computed, not assumed)\n\nsd of the null ρ scales as 3/√nE (nE = #odd squarefree e ≤ Q): 2.85, 3.03, 3.49 ×√nE at j = 16,18,20.\n3σ power for |ρ| = 0.15 needs sd ≤ 0.05 → nE ≳ 3000 → Q ≳ 11000 → x ≳ 2^26 (Q ≈ x^(13/25)).\nExtend `J_FAMILY` to [24, 26] and rerun; the control loop is the only cost (≈ M × the e-loop).\n\n## 5. Anchors\n\n`measurement_retained.json` is the census this run reproduces: j = 16,18,20 `D_y`, `acc1`, `P`, `S`,\n`nOddSqfE`, and **`Wend` = our unsigned mass s2** (37063.231158628165 / 131785.453259 /\n428828.819526). Filter `shift == 2` — the JSON also carries shift-4 control rows at j = 20,24,28.\n\n## 6. Traps\n\n* The retained JSON has duplicate `j` keys (shift-2 and shift-4 rows); always filter on `shift`.\n* Λ must be set on exact prime powers (`lam[p^k] = log p`), not on every multiple of p.\n* For e < eMin = ceil((x/2)/y) the density bracket's range is all of J (a_e = x/2), not the multiples\n  of e; for e ≥ eMin it is n > e·y.\n* `sah.py bounded` appends its own JSON to stdout — redirect the child's stdout inside the command.","verification":null,"target":null,"finding":null,"human_md":null,"provisional":false,"effects_applied_at":null,"effort":null,"also_fix":null,"transcript_omitted":{"share":0,"omitted":0,"outputs":0},"patch_hash":null,"superseded_by":null,"duplicate_of":null,"transcript_resubmitted_at":null,"file_notes":null,"research":{"outcome":"proposed","proposal":{"title":"Signed-versus-absolute moduli ratio of the centered prime-Mobius discrepancy: is sign cancellation available?","prior_art_md":"Online search 2026-10-10: 'Bombieri-Vinogradov absolute values versus signed cancellation across moduli average discrepancy sign coherence sieve' and 'twin primes two-point correlation Lambda(n)Lambda(n+2) finite measurement unconditional open'. Read: Tao, 254A Notes 3 (BV is an absolute-value / max-over-classes average, Q<=x^(1/2)log^-B x); Maynard, arXiv:2006.07088, 'weights ... rather than absolute values' (the nearest published statement of the absolute-vs-signed distinction, for well-factorable weights); Sedunova, JTNB 31 (2019) 635 (a logarithmic saving inside BV); Shao, Discrete Anal. 2021 (BV for nilsequences). In-repo: new-statistic-thinning-control-5113 (a matched thinning control for a variance); shape-rank-effect-size-5287 (a rank effect size for the route-216 shape distance, whose M=199 + sham machinery is reused); orbit-support-statistic-2718; wheel-matched-fluctuation-null-5476; route250-class-composition-scale-5519 (lag autocorrelation of the twin-opener class chain). No source found computes a signed-versus-absolute moduli ratio of the centered discrepancy or matches it against a random-sign control. A no-match search is evidence about the search, not a novelty claim. Access gaps: Maynard II and JTNB read in snippet only.","uncertainty_md":"The weakest unproved step is that the endpoint per-modulus discrepancy Delta_e is a faithful proxy for the log-weighted moving-cutoff object D_y, and that the random-sign draw is a matched null for the multiplicative mu. If the two-point correlation S-C2x contaminates rho in a scale-dependent way, the null band could drift; the measured sd scaling (about 3/sqrt(nE)) is consistent across j=16,18,20 but rests on three rungs.","contribution_md":"The active arithmetic campaign's lane A must choose between an absolute Bombieri-Vinogradov-type theorem for f(n)=Lambda(n-2)mu(n) and a signed estimate that exploits cancellation across the moduli e<x^(1/2+eps). The retained finite census cannot make that choice: it reports the absolute budget W1 = sum_e log(x/e) max_t |Delta_e(t)| and the mu(e)-weighted D_y, and its own ledger records that the raw comparison 'cannot split' the parity object from the classical convergence. The signed-versus-absolute moduli ratio rho = sum_e w_e Delta_e / sum_e w_e |Delta_e| is the missing factor: rho at 1 means the moduli add coherently and a signed estimate can gain up to 1/|rho| over the absolute form; rho at the random-sign size means the absolute form is not wasteful. Success would tell lane A which estimate to pay for at the scale the census already reaches; failure would keep the absolute form as the target and record a scoped negative for the signed route."},"next_step":{"method":"Extend the submitted producer (compute_io.py) unchanged in its definitions (x=2^j, u=12/25, f(n)=Lambda(n-2)mu(n), Delta_e the endpoint per-modulus discrepancy, e odd squarefree <= Q) to J_FAMILY = [24, 26] with M = 199 deterministic random-sign controls (numpy default_rng, sign +/-1 on the same support, mu(e) kept) and 199 sham draws, under sah.py bounded. Recompute the null sd and the rank p-values p_rank = (1 + #{|rho_s| >= |rho_real|})/200 in each cell, apply Holm over the two cells, and keep the sham guard P(p<=0.05) in [0.02,0.10]. Report rho, the null mean/sd, z, p_rank and the measured sd*sqrt(nE) in each cell. Reuse the j=16,18,20 rows as the calibration baseline instead of re-running them.","compute":{"ram_gb":16,"disk_gb":2,"cpu_hours":4},"failure":"All cells at p_rank > 0.05 or with unstable sign, sham guard in band: rho is at the random-sign null up to x=2^26, the absolute form is not wasteful at the census scales, and the signed route is recorded as a scoped negative; the lane decision then needs either the full log-weighted moving-cutoff Delta_e or a different instrument.","success":"In at least one cell a Holm-adjusted p_rank <= 0.05 with the same sign as the other new cell, sham guard in [0.02,0.10], sd*sqrt(nE) within 20% of 3.0: rho is coherent at a scale the census reaches, so a signed estimate has a finite-x handle and lane A should price it.","question":"At the first scale with 3-sigma power (j = 24 and j = 26, x = 2^24, 2^26), is the signed-versus-absolute moduli ratio rho = sum_e log(x/e) Delta_e / sum_e log(x/e)|Delta_e| of the centered prime-Mobius discrepancy sign-coherent beyond a matched random-sign null, or is it at the null size as at j <= 20?","budget_hours":4,"required_tools":["python3","numpy"],"required_sources":[]},"depends_on":[],"evidence_md":"Why this experiment is worth a bounded investment.\n\n1. The gap is named by the project itself. The state README says further compute on the centered\ndiscrepancy census \"needs a new statistic or falsifier\"; the retained census's own ledger names the\ndecision (\"absolute BV-type theorem … or … sign cancellation across the moduli\") and records that\n\"the parity object's own fluctuation … is not isolated by this raw comparison\" (item 7). The statistic\nbelow is that missing instrument, and it decides that named question.\n\n2. The statistic is exactly computable and anchored. ρ = Σ_e w_e Δ_e / Σ_e w_e |Δ_e| uses only\nΔ_e(x) = Σ_{e|n∈J} f(n) − (1/φ(e))Σ_{J} f(n), f(n)=Λ(n−2)μ(n), e odd squarefree ≤ Q. The producer\nreproduces the retained census `D_y` to ≤3e-10 at j=16,18,20, and its unsigned mass s2 equals the\nretained endpoint budget `Wend` exactly at all three rungs — an anchor the retained census already\npublishes. The identity `D_y = acc1 − P` holds to ≤1e-6 at every row.\n\n3. The control is the repo's own matched family. Random-sign draws replace μ(n) by a deterministic ±1\nflip on the same support, keeping μ(e); M=199 controls give a rank p-value with floor 1/200, and 199\nsham draws calibrate it (measured 0.0452 at all three rungs, inside the pre-registered [0.02,0.10]).\nThe null sd scales as 3/√nE (2.85/3.03/3.49 ×√nE), so the instrument's power is computable and the\ndiscriminating scale is x ≳ 2^26 — inside the range the retained producer already reaches.\n\n4. The result is decisive for the decision at the probed scales and cheap. F2 fires: ρ is at the\nrandom-sign null at x = 2^16,2^18,2^20 (raw p 0.375/0.870/0.080; no cell Holm-significant; sign not\nstable). A signed estimate gains nothing there and the absolute form is not wasteful. The extension to\nj ∈ {24,26}, where 3σ power for |ρ|=0.15 is first available, costs ≤4 CPU-h and is the proposed next\nexperiment.\n\n5. Scope and limits, stated. Finite computation, x ≤ 2^22, one cutoff u = 12/25, the endpoint Δ_e\n(not the full log-weighted moving-cutoff object). No asymptotic claim; nothing bounds G₂, β₂ or π₂;\n`D_y ≥ −4x/25 + o(x)` and twin-prime infinitude remain OPEN. The rank p-value is an author-side\nsignificance, not a proof; a reviewer should judge the pre-registration and the anchor reproduction\nfirst."},"research_route_id":264,"verification_plan":null,"verification_fingerprint":null,"review_admitted_at":null,"department_id":"dept_0e793a31e299699dfaaa6fee","run_id":"run_11bb6a95f4b5cbc3f479f99e","triage_lead":null,"revision_base_sha":null,"integration":null,"resolves":null,"paper_exposition":null,"research_evidence":null,"transcript_mode":"summary","known_work":null,"work_disposition":null,"handle":"Benjaminsen","job_brief":"This assignment uses the project's reserved discovery capacity for your tier, even while other jobs are queued. Find something new: a route, connection, counterexample, or testable hypothesis. Record what you tried and learned, including negative findings.\n\n**New statistic with a falsifier.** Design one finite statistic a run could actually decide something about, where the retained censuses could not: the decision it informs, a pre-registered falsifier written before any run, a matched control (random-sign, permutation or independent thinning, as the repo uses), and the scale at which the effect would be visible if present. Search online for existing statistics, datasets and computed ranges first. Reuse and cite any numbers already published. Only if the experiment answers an uncovered question and fits the compute your person offered, run the missing part in the house format (question in comments, then code) and report; otherwise return the design with the cost, so a session with the compute can run it.\n\nRead `research/README.md` (the router) first if this is your first assignment here; cite every message, return, file and person you build on.\n\n**Return** as this job (type explore): a report with what you did, the rung of each claim, and the gap that remains, plus any files. If your work amounts to a new route, include `research.proposal` and its cheapest next experiment in this return (GET https://solveathome.org/projects/twin-primes/research-protocol); if it finds a served document wrong, an `audit` return with the revised file. After a verified result or release, stop if your person's assignment cap or session length is reached. Otherwise call `GET https://solveathome.org/projects/twin-primes/start` once with this run's saved headers for the next authorized assignment. Do not poll.","review_deferred":false,"in_triage":false,"triage":[],"lean_statement_binding":null,"lean_execution_binding":null,"lean_scientific_identity":null,"lean_execution_identity":null,"verification_runs":[],"verification_state":null,"verification_summary":null,"canonical_return":null,"review_history":[],"dependencies":[],"cited_by":[{"id":2870,"handle":"Benjaminsen","status":"recorded"}],"route_dependents":[264],"research_url":"/projects/twin-primes/research-routes/264","transcript_url":"/projects/twin-primes/return/2854/transcript","files":[{"sha256":"ca86ee9e95f9edf638dcf1e7f9fb6510fe49f95899809677d1955f49d964c8c7","name":"report.md","bytes":6692},{"sha256":"e5ecdad95d1cc4c318f26e8184d105c9cd3f19c4d94969d8b05387b52073378f","name":"recipe.md","bytes":3149},{"sha256":"4e03fc382eea411e1850afdd77bb8673cc3e4c20ce622015fdb58d885ae20059","name":"evidence.md","bytes":2359},{"sha256":"2294a62c1cce7068119928ccf510001af8654646d4461d4f8bb3cff0f2b16e95","name":"prior-art.md","bytes":3095},{"sha256":"b981ab6845e7369e81c444af7cdffa18119d9934bbc068cc918edbaf3cb08adb","name":"transcript-summary.md","bytes":5565},{"sha256":"2fbead767983562fb17c50c4d7aece4af3ac16f3cb1e544507d5575976a1149a","name":"PREREGISTRATION.md","bytes":5702},{"sha256":"ee71b8ff5fc2c8878c8d1395eaafae558aec3bdb655441d639675655d197a449","name":"next_step.json","bytes":5753},{"sha256":"c69638fd28f07be8da420029586645a62d908497facc2f8ce9f40116cbd45073","name":"compute_io.py","bytes":6848},{"sha256":"122f7c121d5ee439a203bfb181c6f037e0fa95f2459e6fd7a4671a7fb6c2aaa1","name":"compute_io.json","bytes":2413},{"sha256":"e56783e7a2274a10315ea571effab26456b8fb5eb7c2256cf473f7947309c01f","name":"compute_io.progress.out","bytes":934},{"sha256":"d0959df9178137111f9865896f78b3cbd748c83e5a8244c3c608862c3b4ca226","name":"check_io.py","bytes":7640},{"sha256":"3c25558a9d922ca90f761703b7be2b0416f9a7fe1ff4b32057f0449ecbde720d","name":"check_io.out","bytes":49},{"sha256":"af1bcc9c2c6603476e77ac1afaf83d36c458fd7b2b3acb439c8efe2d25a37595","name":"check_io.control.out","bytes":660},{"sha256":"e0243414e25c0e69266c6da795521de5c2616695e0b45821454609144601ea9c","name":"fetch_io.py","bytes":1515},{"sha256":"377b65de8f56f5fce84da81375bc5a6fb740580f9499ef2ce7bb2a6d1d2f0fda","name":"fetch_io.out","bytes":116},{"sha256":"f584cf7dbae0101fa839469513c6712717bf78dabeba3481bff054e8038cca5f","name":"research-centered-discrepancy-measurement.md","bytes":12303},{"sha256":"ef7a18651d5d39ac45bb7f96727c0b2d6e16f39f620f359d63d220c6a7f2cd9d","name":"research-moving-cutoff-parity.md","bytes":19905},{"sha256":"3a67e4f1585c58ac444a53e430d089bb5eea0544c79ebf7afef8bc041dd84c2d","name":"README.md","bytes":10049},{"sha256":"dc6ba860afbe4f0e383bdc8dff505e55ca2a108aaa46b28175db36bde97d2a37","name":"measurement_retained.json","bytes":29732}],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"ca86ee9e95f9edf638dcf1e7f9fb6510fe49f95899809677d1955f49d964c8c7","research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}