A yes-or-no meter for scientific claims comes with no error rate

The setup

Consensus is scoped narrowly: a search engine over peer-reviewed papers, built for the task of checking whether a claim has support across studies rather than reading one paper at a time. Structured data on the vendor's own homepage describes it as searching '220M+ scientific papers' with a 'Consensus Meter showing scientific agreement on yes/no questions' and 'Study Snapshots with key findings and methodology,' and lists a free tier alongside paid plans. The company's about page states Consensus was founded in 2021 by Christian Salem and Eric Olson, positioning the product around making research 'easier to find, understand, and trust.'

What the documents show

A features page describing its literature-review use states that 'every source' is 'linked to a real peer-reviewed paper' and that Consensus 'extracts structured snapshots from every study — methods, outcomes, populations, sample sizes — and links every claim directly to the paper.' The same page markets the product as helping 'do literature reviews in hours, not weeks' through search, screening, extraction, and synthesis steps, framed as speeding up a reviewer's own process rather than delivering a finished systematic review to that standard. The Consensus Meter itself is described only as surfacing agreement on yes-or-no research questions, not as a replacement for reading the underlying studies.

The friction

The same features page states plainly that the product produces 'no hallucinations,' a claim asserted rather than measured with any published error rate, in contrast to competitors that publish their own tallied failure counts. Consensus's own materials do not state what share of relevant literature its 220-million-paper index actually covers for a given field, nor do the pages retrieved here specify how disputed or minority findings are weighted when the meter renders a single agreement score. A reader has no way, from the vendor's own pages, to check the 'no hallucinations' claim against a documented test.

What changed in the work

For someone checking whether a claim holds up, Consensus's own framing is that it compiles a cited answer across many studies faster than manually opening a database and reading abstracts one by one, which the company states as its purpose rather than something independently confirmed here. That still leaves the reader responsible for opening the linked snapshots and judging whether the studies the meter drew on are the right ones to weigh, particularly since the vendor's own pages do not describe the meter as a systematic review.

• Does the search tool state which papers it excluded, or only which ones it surfaced?
• Is an 'agreement' score built from a handful of studies or from the field's full evidence base?
• Would the answer change if a dissenting or minority study were weighted more heavily?

A consensus meter is a useful shortcut for locating disagreement, not a verdict, and the vendor's own pages describe it as a screening aid rather than a substitute for reading the papers it cites.