How scores work

Each section has its own criteria — movies get Story, Acting, Visuals, Sound and Rewatchability; tech gets Design, Performance, Battery and friends. Every review scores each criterion from 0.0 to 10.0, and each criterion has a weight that says how much it matters. The critic score is the weighted average, rounded to one decimal.

Weights can be tuned per review: Sound weighs more for a musical, Battery weighs more for a phone you carry all day. The breakdown on every review page always shows the effective weight actually used — if a criterion doesn't apply, it's simply left out of the average, never counted as a zero.

Anyone can rate any review 0–10 — no account needed, one rating per person, editable any time. But a plain average is fragile when few people have voted: one troll (or one superfan) would own the number. So we use the same idea as IMDb's Top 250 — a Bayesian average.

The score behaves as if 10 phantom ratings at an average score already existed. With a handful of real ratings the prior dominates and the score stays cautious; by around fifty ratings the crowd has fully drowned it out. Skepticism at low numbers is the feature.

That average is per section, not site-wide. People rate films, series and hardware differently, and smoothing a film toward a number that mostly describes how readers rate headphones would be the wrong kind of caution. A section needs 50 ratings before it is trusted to describe itself; under that it borrows the site-wide mean.

Sometimes a review attracts a sudden burst of coordinated 0s (or 10s) that has nothing to do with the thing being reviewed. An automatic check runs every hour and looks for two things happening at once in the last 24 hours:

When both signals fire, ratings from that 24-hour window are de-weighted to ×0.25 — not deleted, not hidden. Legitimate raters caught in the window still count, just less. The adjustment is logged publicly, and the review immediately shows a notice like this one: