Integrity findings
Paper query · 2606.10992Detector result on named evidence, not a paper verdict.
The query succeeded; no auto-published findings are recorded for this paper. As of 2026-08-05.
INDEX 18,064 integrity findings 696,671 claim nodes 7,620 open flags
Pith is building a public, inspectable history of what a paper claimed, what was checked, what was challenged, and what was later corrected or superseded. It will not add the record up into a score.
Pith will not tell you whether a paper is true. It shows the record and its limits so you can judge the evidence.
Every source stays separate. Silence is never turned into a clean bill of health.
Pith is an open preprint site with review built in, alongside arXiv rather than on top of it.
Enter an arXiv ID. Pith resolves the paper and queries each live source separately.
Machine-readable twins: /mesh.json · /evidence.json2606.10992
Surface Crevasse Evolution Observed Using Matched Field Processing and Source Relocation at Hansbreen, Svalbard
Resolved against the papers table. Legacy arxiv_id remains authoritative.
Paper · Integrity · Claims · Citation record
Showing records for 2606.10992. Each source was queried independently.
Live The public source is available now.
Shadow The capability is not a public evidence record.
Empty The query succeeded and found no matching records.
Unavailable No result could be established.
Not collected The declared collection does not contain it.
Not requested The source was not queried. Neither state means zero.
Detector result on named evidence, not a paper verdict.
The query succeeded; no auto-published findings are recorded for this paper. As of 2026-08-05.
Extracted claim text and dependency edges; not settled truth.
Showing the 4 latest records (maximum 8).
Two-step workflow of matched field processing followed by arrival-time relocation for meter-scale icequake source localization under sparse array constraints
Independent measurements of the same crevasse opening episodes, for example via repeated drone surveys or time-lapse cameras over the study period, that produce opening rates or diffusion values outsi
The relocation step after matched field processing truly removes systematic location errors and delivers unbiased meter-scale accuracy with the available sparse instruments.
We propose a two-step workflow integrating Matched Field Processing and discrete arrival times relocation to map glacial seismic activity at meter-scale resolution under limited instrumentation. Apply
Resolved reference and quoted context; not dependence.
299 stored references found for the latest stored version.
Neutral one-hop delivery of an external event notice.
Showing the 3 latest records (maximum 8).
Event source title: <b>To: “Results of the downhole microseismic monitoring at a pilot hydraulic fracturing site in Poland — Part 1: Event location and stimulat
Event source title: Automatic microseismic event picking via unsupervised machine learning
Event source title: Automatic microseismic event picking via unsupervised machine learning
Publication freezes a citable version, not the world around it.
Sources move. Errors are found. Claims are challenged. Analyses are rerun. Papers are revised or superseded. When those events are scattered across databases and private correspondence, readers cannot reconstruct what changed or why.
Pith’s answer is a dated record whose sources, limits, and disagreements remain visible, without collapsing them into one score.
The record keeps kinds of evidence separate because they answer different questions.
Evidence type
What a named source contained at a stated time.
Evidence type
A named program produced a repeatable result from fixed inputs.
Evidence type
A model produced an assessment. It remains a fallible judgment.
Evidence type
An attributed person made a review, objection, or challenge.
Evidence type
A small formal proof checker accepted an exact machine statement under listed assumptions. Whether the statement represents the scientific prose or the world remains challengeable.
History rule
Challenges, corrections, and superseding records point to their target without deleting what came before.
Every view must name the paper identity, collected records, observation time, and rules used to produce it. A generated paper history is a dated reconstruction from declared inputs. It is never stored as canonical truth.
Pith can query integrity findings from named detectors; extracted claim text and dependency edges; stored citations, with quoted context available on citation surfaces; and neutral one-hop flags about external events involving cited works.
Each source keeps its own coverage and limits. They are not merged into a paper verdict or a single timeline.
Pith is building typed, content-addressed records around those sources. Content-addressed means the identifier changes when the record’s bytes change.
The generated paper biography and replay by separately controlled operators remain shadow until their preregistered tests pass.
Scientists
Inspect exact records around a paper, distinguish silence from missing coverage, and challenge a specific judgment.
Readers
Follow what changed without relying on a score or treating Pith as a truth authority.
Funders
See what is live, what remains contingent, and which published tests can stop the next phase.
Prospective hires
See the actual work: identity, replay, hostile testing, privacy, and independent operation. To help build it, write to hello@pith.science.
Disagree with a finding or extracted claim? Challenge that item and cite the evidence.
The original record and the challenge remain public. Disagreement does not silently rewrite history.
Findings · Challenges · Protocol · Flags
/mesh.json and /evidence.json expose the public dossier schema used by this page. Responses include the schema version, observation time, source states, and a context_digest, a hash clients can use to detect a changed rendering context.
Separately timed requests are not promised to be identical. Compare digests before treating two responses as the same observation. Preserve evidence types; never promote a capture or delivery notice into an accepted check.
Schema pith.mesh-hub.v1.
curl -sS 'https://pith.science/mesh.json' curl -sS 'https://pith.science/mesh.json?paper_id=0704.0001' curl -sS --get 'https://pith.science/mesh.json' --data-urlencode 'paper_id=math/0211159'
| Surface | Responsibility | What it can establish | Mode | Human | Machine |
|---|---|---|---|---|---|
| Evidence Mesh | Content-addressed objects, checks, chronology, and a receipt-backed paper biography fold. | Object type controls effect. Custody is not correctness. | shadow | /evidence |
/mesh.json |
| Integrity | Continuous detectors, signed findings, challenges, Pith Number bundles. | A finding is a named detector result, not a paper verdict. | live | /integrity |
/pith/{id}/integrity.json |
| Tools | Point the same review machinery at a bibliography, claim, or falsifier you bring. | Instrument coverage is local; silence is not a certificate. | live | /tools |
/tools/reference-audit.json |
| Citations | Resolved references, quoted contexts, foundational works. | A citation edge is evidence of reference, not dependence or fault. | live | /citations |
/citations/search.json |
| Flags | Events-only one-hop notices near cited works. | Delivery only. Never adverse inheritance through a bare citation. | live | /flags |
/flags |
SPECIMEN: schema-true, not a live event
Projection profile pith.paper-biography.v0. This is a schema-true specimen, not a live generated history.
{
"as_of": "\u003cRFC3339\u003e",
"collection_frontier_hash": "sha256:\u003cmanifest\u003e",
"coverage": {
"receipt_ids": [
"..."
],
"state": "complete_within_manifest | partial"
},
"event_frontier_hash": "sha256:\u003coperator-checkpoints\u003e",
"identity_snapshot_digest": "sha256:\u003caliases-and-resolutions\u003e",
"note": "Schema-true specimen only. The paper-biography fold is not a public default read.",
"output_hash": "sha256:\u003crender-model\u003e",
"policy_fingerprint": "sha256:\u003cbehavioral-policy\u003e",
"profile": "pith.paper-biography.v0",
"reducer_digest": "git:\u003ccommit\u003e",
"subject_id": "paper:arxiv:\u003cid\u003e",
"subject_version_id": "paper-version:arxiv:\u003cid\u003e:v\u003cn\u003e"
}
Preregistered Mesh gates live in the architecture decision; this page does not paint gate status by hand.
Read the architecture and preregistered failure gates. Gate results come from test artifacts, never hand-edited labels on this page.
This page is itself a dated observation. Its digest is printed below.
context_digest=sha256:58506f8b9b8c66f26df5ff8b46d599974b6c113ddee5bfbaee3ffec996a97b09 · template_digest=mesh_hub.html · rendered_at=2026-08-05T11:51:00Z