MIRAmodular interoperable research attribution Try it

Reference · the unlimited tier

Schema, prior art, receipts.

The supporting material for everything on the site: what the schema contains, where MIRA sits in the lineage, and the source of every quote we used.

Schema at a glance

One source of truth; everything else generated.

The canonical grammar is mira.yaml (LinkML). From it, LinkML generates mira.shacl (machine-checkable conformance), mira.ttl (RDF), and mira.jsonld (the JSON-LD context that is the wire format). Hand-editing the generated artifacts is a bug; edit the source and regenerate.

mira.yamlgen-shaclmira.shacl+mira.jsonld+mira.ttl

Node types: Question · Claim · Evidence · Study · Protocol · Request · SourceDocument. Core slots used on this site: observationBase · observationStatement · is_grounded_in/grounds · follows · sourceDocument · addresses · supports/opposes/informs · request_for/request_target. Canonical source: github.com/MIRA-science/schema.

Prior art

Extending a lineage, not reinventing it.

Prior artShared ideaWhat MIRA adds
NanopublicationsThe reified assertion — a statement that carries its own provenance and author.A question layer — open unknowns as durable, first-class nodes — and the specific relation this observation is about that claim, carried as its own relation slot (the observationStatement / observationBase split).
AIF (Argument Interchange Format)Argument structure — supports and attacks between claims.
PROV-OProvenance — who produced what, from what, when.

MIRA is compatible with all three; the additions are the question layer and the observation-about-claim relation. Most systems that emit claims-with-source-spans already produce, in effect, an Evidence + observationStatement + sourceDocument triple — the mapping is short, which is the whole bet of a shared schema.

Receipts index

Every quote, with its source.

The verbatim quotes and hard numbers used across the site, each with the research file and line where it is recorded. A judge who clicks through will find the caveats we cite first; the two research files carry the full unverified lists.

COVID-19 origins

there does not exist any publicly available list of the earliest known covid cases… which cannot be independently verified or duplicated.Judge Eric Stansifer, Rootclaim debate decision · research/04-covid-findings.md:16–20
No line list of early COVID-19 cases is available · extraction introduces up to about 50m of noise; 155 of 164 points recovered.Worobey et al. 2022, Science 377:951 · research/04-covid-findings.md:26
I was not able to evaluate several lines of inquiry which I thought could have been valuable for both sides.Judge Will Van Treuren, Rootclaim debate decision · research/04-covid-findings.md:128–129
A cursory search suggested some (more credible) reports of early seropositivity in retrospective studies of US blood banks that would have been hard to explain under the ZO theory.Van Treuren, his unexamined crux · research/04-covid-findings.md:124–126
The necessary model does not exist; neither their analysis nor Worobey’s could be used to support or reject the zoonosis hypothesis.Stoyan & Chiu 2024, JRSS-A · research/04-covid-findings.md:146–149

Are eggs healthy?

a sensitive and specific biomarker of egg consumption has not yet been reported in the literature.Münger et al. 2018, Genes & Nutrition 13:26 (635-paper review) · research/05-eggs-findings.md:180–181
Fasting TMAO, pooled across chronic feeding trials: WMD −0.08 µmol/L (−0.47 to 0.32), p = 0.69, I² = 23% — a consistent null.Aryaeian 2025, Nutr J 24:165 · research/05-eggs-findings.md:234–236
Plasma choline in free-living people explains ~4% of its own variance (r ≈ 0.20); correlates with egg intake at r = 0.06 (largest study), better with alcohol.research/05-eggs-findings.md:199–201

The Request→Study handoff (one lab, Roam + DG plugin)

445 issues over 40 months · 29% claimed · 15% by someone other than the creator · 38% of claimed issues produced a result · median 12 days claim→first result.Matt Akamatsu’s published lab updates · DRAFT.md:337–343

On what we left out. Numbers that did not survive checking are not on the site: the COVID “23 orders of magnitude” framing, and the eggs CCA 40%/28% figure (still to be confirmed against the open-access Formisano 2025 full text). The self-audit baseline is 347 violations, not the stale 661 in an older validation note. Full unverified lists: research/04-covid-findings.md §4 and research/05-eggs-findings.md §5.