What the rubric saw

Every paper below was evaluated blind, by the same rubric. No journal name, no citation count, no author prestige fed into the score. We've grouped them by what the evaluation revealed, because the point isn't the score itself. It's whether the score changes the answer you'd otherwise have given.

When prestige would have misled you

These papers carry the signals researchers lean on — high citations, recognizable venues, established authors. Read on those signals alone, you'd trust them. The rubric read the work instead.

When the work was better than its journal

Citation count and journal prestige would have filtered these out. The rubric, reading only the work, rates them well above where their visibility would place them.

When the paper was later retracted

Each of these was evaluated blind, with no retraction information available to the models. Each was flagged in the lowest reliability tier before — or independently of — the public record catching up.

The rubric on landmark work

It isn't a hatchet. Shown genuinely strong work, the rubric says so — and shows why.

The rubric on the literature about peer review

We pointed the rubric at the research on research evaluation itself — the studies on reviewer bias, error detection, and inter-rater reliability. It reads them the same way it reads everything else.