Reviewer Panel BETA

Where the scores come from. Each reviewer’s published video grade is converted to a 10-point scale: stars ×2, letter grades from A+ = 10 in half-point steps, scores out of 100 ÷10. These are opinions from published reviews, not measurements of the disc.

Strict or generous. Each reviewer is compared with the others who reviewed the same discs, after allowing for how well each disc is rated in general (a “two-way model”: every score is the disc’s general standing plus the reviewer’s usual habit). A reviewer at −0.5 pts typically scores half a point below the others on the same disc. “Give or take” and the line through each point show the likely range (95%), found by re-sampling the discs. When two reviewers’ likely ranks overlap, they are marked “too close to call”. A simpler comparison against the plain average of the others is in “The precise figures”.

Upscales. For each reviewer we compare how far they sit from the other reviewers on upscaled discs with how far they sit on native 4K discs. That shows whether they treat upscales differently from their peers. It describes a pattern in scoring, not its cause.

Eras and capture. These use the reviewers’ average score, before UHD Reference adds its +0.2 native 4K bonus, so a scoring rule never leaks into a comparison of eras. Film and digital capture are compared only where both exist: digital capture appears from 2010.

Genres. Genres come from TMDB. A film with several genres counts in each, so genre figures overlap and aren’t independent of each other. Like the eras, they use the reviewers’ average score before the native 4K bonus. “Divided” counts discs with three or more reviews where one reviewer is far from the rest or the reviewers split into two camps.

Disagreements. The gap is the highest minus the lowest reviewer score. “One reviewer vs the rest” means one score sits at least 2 points from every other while the others agree (within 1 point with three reviews, within 2 points with four or more). “Reviewers divided” means a gap of 3 points or more with no single outlier. Smaller differences are normal variation between reviewers and are not listed.

Matches. A match shows whether two reviewers tend to rate the same discs higher or lower (correlation: strong at 0.6 or more, some at 0.4 to 0.6, weak below). Reviewers who use only a few distinct scores look like a weaker match even when their scores are close, so the share of discs within half a point is shown beside it.

Samples. Every figure states how many discs it uses. Reviewer figures need 30 discs, pairs 30 shared discs, upscale comparisons 20 discs on each side, and era figures 10 discs; below that a figure reads “Too few discs”. Headline figures use one decimal because the likely ranges are about ±0.1; the precise figures keep two. In Pipeline discs (a single review) are left out.

Filters, Compare and Ask. Filters recalculate every figure for the discs you choose, and “leave out a reviewer” recalculates as if that site’s scores didn’t exist. Compare sets two reviewers against each other on the discs both reviewed. Ask turns a plain-English question into the same filters and shows how it was read. Questions outside its vocabulary get an honest “can’t answer yet”, never a guess.