Science, verified. For everyone.

How sure are we, really?

A claim on the record ·

Verdict

Unsupported
Contested
Supported
0% agree 100%

Evidence Grade

DATOM checks whether scientific claims hold up, by someone with no stake in the answer, and records the result in one place anyone can read.

The unit

What is a datom?

Study

One paper.

Datom

One specific finding from it.

Datomer

All the datoms that answer one question.

A datom is one finding, small enough that breaking it down further would lose the meaning. A datomer is all the findings that speak to one question.

The Problem

The reproducibility crisis is a trust problem.

A claim doesn't need to be true to spread on social media. It needs to sound scientific. Misinformation spreads by accident; disinformation spreads on purpose; both move faster than the evidence that would settle them. Health myths, wellness marketing, and viral "studies" all travel on the authority of science itself, and by the time anyone checks whether the evidence holds up, the claim has already shaped what people believe, what they buy, or what they refuse to do.

70%

of researchers have failed to reproduce another scientist's experiments. Findings that can't survive a second look don't stay in journals. They end up in headlines, product claims, and social feeds, repeated as fact long after the evidence behind them has collapsed.

Baker, Nature 2016 (survey of 1,576 researchers)

$28B

wasted annually in the U.S. on irreproducible preclinical research. The same unverified findings fueling that waste are the ones spreading online with no way to check them, and no authoritative record either side of a dispute can point to.

Freedman et al., PLOS Biology 2015

6×

faster is how quickly false news spreads than the truth on social media, reaching more people at every depth of a sharing cascade. The claim wins the race long before the evidence that would settle it arrives, which is exactly the gap a public record has to close.

Vosoughi et al., Science 2018

Claim
→
Independent Verification
→
Public Record
Updates as evidence changes

The Ledger

One record. Open to everyone.

Independent.

DATOM does not produce the claims it verifies, does not employ the authors, and does not profit from the answer. Verification happens at arm's length, every time.

Recorded.

Every verification enters the ledger with the claim, the evidence examined, the result, and a timestamp. Entries can be added to, never rewritten: the record persists and updates as new evidence arrives, and a correction is appended, not erased.

Public.

Every verified claim's verdict, its Evidence Grade, and when it was last confirmed, free to read. The full evidence opens with a free account. See what each tier includes →

Two ways in

Read the record, or get a claim verified.

Check a claim you've seen

Look up what the evidence actually says about a claim. Every verdict is free to read, with the full evidence one click down.

See a claim on the record →

Get a claim verified

Commission an independent Confidence Report on a claim that matters to your work, or set up ongoing access.

For organizations →

Questions

Frequently asked questions

Is a verdict an AI opinion? ⌄
No. The verdict and grade are computed from the underlying evidence, not asserted by a human editor. AI does help read the studies and draft the summaries, and we would rather say so plainly than let you assume otherwise. What keeps that honest is that nothing rests on it: every finding is tied to a real, named study you can open and check yourself.
Does DATOM decide what is true? ⌄
No. DATOM makes the evidence inspectable and computable and records where it points. The decision remains with the reader.
What prevents misuse or cherry-picking? ⌄
DATOM makes cherry-picking visible. The full bibliography is public, so omitting evidence reduces credibility rather than hiding inside a summary. The domain classification and grading basis are printed on every record and are auditable.
Who pays for verification, and doesn't that bias the result? ⌄
Payment moves priority, never outcome. Public demand triggers verification whether or not the claim's subject pays. Commissioning and sponsorship are disclosed on every record. The ledger is append-only, so an unfavorable result can never be withdrawn, only appended to. And public sentiment carries zero weight in the verdict or grades, enforced by a permanent build gate. Read the full firewall policy →
What happens when DATOM gets one wrong? ⌄
The correction is public and the history is preserved. The record is append-only: a correction is added as a new entry with the evidence that triggered it, never rewritten or deleted. You can see when and why a grade or verdict changed, and the version history stays on the record.
Why does this claim have no verdict yet? ⌄
The record is open to every claim, so a claim that has been asked but not yet verified still has an entry: it is on the record and pending, not rejected. Entries advance through continuous prioritization across editorial selection, reader demand, and sponsorship.
How is a claim graded and given a verdict? ⌄
The verdict is the direction of the evidence: how many independent sources support versus contradict the claim. The Evidence Grade is the material quality of that evidence base, read as a letter, alongside a field-relative grade against what is achievable in the claim's domain. Both are computed from the evidence structure, not authority, and both update as evidence changes.
Why are contradictions included? ⌄
Contradictions are information. They move the verdict toward Contested and narrow applicability without erasing supporting evidence or forcing false consensus.
Who is this for? ⌄
Anyone who wants to know whether a claim holds up: readers checking something they saw online, researchers, investors performing diligence, and institutions making high-stakes decisions.
Is DATOM free to use? ⌄
Yes. Every record's verdict, grades, and summary evidence are free to read. A free account opens the full evidence tree. Paid tiers add sponsored verification and commissioned Confidence Reports.
How can my organization work with DATOM? ⌄
Companies, research labs, investors, institutions, government agencies, and universities can commission independent verification or integrate the record directly through DATOM Pro.
Is this replacing peer review? ⌄
No. It complements peer review by making underlying evidence explicit and reusable.
How is this different from a meta-analysis? ⌄
Meta-analyses aggregate results. DATOM preserves structure, provenance, and disagreement. Each contribution remains visible and comparable.
How are new studies handled? ⌄
New evidence is added as new entries. The verdict and grades update transparently as the record evolves, and the change is logged in the version history.
Does DATOM work with proprietary data? ⌄
Yes. DATOM supports private records and internal grading while maintaining the same structural logic as the public ledger.
Is DATOM a database, a product, or a standard? ⌄
Infrastructure: a data model, a grading framework, and a trust primitive.
What is a datom? ⌄
One specific finding from one study: a measurement, a method, a dataset, an assumption, a constraint, or a replication condition. It is small enough that breaking it down further would lose the meaning, which is why datoms cannot be merged or simplified. See the diagram →
What is a datomer? ⌄
All the datoms that answer one question, kept as parts you can open rather than flattened into a summary. When a record names its Anchor, that is the datomer its verdict was read from. See the diagram →