The Correlation Between Popularity and Quality in Anime Is 0.33. That's Not Nothing. It's Also Not What You Think.

We scored 187 shows on the Codex rubric and ran the numbers against MAL membership counts — popularity predicts about a tenth of the variance in quality, and the tails are where the medium actually lives.

Fairy Tail has 1,848,185 members on MyAnimeList. Ashita no Joe has 68,883. The first scores 6.25 on the Codex rubric. The second scores 9.02. If you want a single data point to break the spell of vote-average aggregation, that's it — a 27-to-1 popularity ratio inverting cleanly into a nearly three-point quality gap. The question isn't whether popularity tracks quality. It's how loosely, and where the slippage shows up.

Are Popular Anime Actually Good? The Number Is 0.33.

The honest answer to "are popular anime actually good" is: weakly correlated, often, in aggregate, with enormous exceptions on both ends. Across 187 catalogue entries, the Pearson correlation between log MAL membership and Codex score is r = 0.33. That's a real signal — popularity is not random noise against quality — but it explains roughly eleven percent of the variance. The other eighty-nine percent is everything vote-average sites flatten: rubric weight by genre, the cultural moment a show aired into, whether a studio's compositing carried a thin script, whether a quiet seinen ever found its audience.

The consensus reflex on anime forums is to treat MAL's top-rated list as a proxy for the medium's canon. It isn't. It's a proxy for which shows got watched by enough people to generate a stable score above 8.0, weighted by recency bias and the demographic skew of the site's userbase. That's a real measurement of something. It just isn't a measurement of quality. Vote-average sites conflate "how many liked it" with "how good it is." The correlation is real but loose — and the gaps are where it gets interesting.

The Underseen Tail: Five Shows the Algorithm Buried

Start with the top of the underseen list. Ashita no Joe (1980, Tokyo Movie Shinsha, 47 episodes) scores 9.02 on the Codex and 8.83 on MAL, with under 69,000 members. The rubric weights it on character (9.5) and cultural impact (9.8) — Joe Yabuki is the spine of the entire delinquent-protagonist tradition that runs through forty years of shonen after him, and the show's animation (8.0) is held down only by the technical limits of 1980 cel work. If you haven't sat through it, our full breakdown of the 47 episodes is the entry point. The point here is structural: a show this foundational with this small a member count is the cleanest possible demonstration that the popularity signal decays with age and accessibility, not with merit.

Pluto (2023, Studio M2, 8 episodes) sits at 8.96 Codex against 8.44 MAL with 254,812 members — a Netflix release of an Urasawa adaptation that should have detonated and instead got politely watched. The rubric reads it on themes (9.3) and story (9.2); the cultural impact score (7.8) is the honest discount for a show that vanished into the streaming-recommendation void. We've mapped where it sits on the seinen axis in more detail, but the headline is that an eight-episode prestige adaptation by a boutique studio cannot, by structural design, generate the membership numbers a 175-episode Satelight shonen can — and the rubric doesn't pretend otherwise.

Honey and Clover (2005, J.C.Staff, 24 episodes, 8.66 Codex, 7.98 MAL, 285,979 members) is the josei case. Character 9.2, themes 9.0, and a fanbase smaller than three Black Clover arcs. Hikaru no Go (2001, Studio Pierrot, 75 episodes, 8.65 Codex, 8.08 MAL, 150,411 members) is the genre-niche case — a Go anime, of all things, with a 9.0 cultural impact score because it triggered a measurable national surge in youth Go participation. Tongari Boushi no Atelier (2026, BUG FILMS, 13 episodes, 8.50 Codex, 8.69 MAL, 339,760 members) is the recency case: a show that hasn't had time to accrue an audience but has already cleared the rubric ceiling on world-building (9.2). Different reasons in each case. Same pattern: the popularity signal misses them.

The Overpopular Tail: Five Shows the Numbers Refuse to Round Up

Now invert. The five most popular shows in the catalogue that the Codex scores below 7.0 are all long-running Pierrot, Satelight, or A-1 shonen. Bleach (2004, Studio Pierrot, 366 episodes, 6.55 Codex, 8.00 MAL, 2,212,907 members) — the structural problems are documented in our standalone breakdown, but the short version is story 6.5, character 6.0, a show whose membership count reflects two decades of cumulative reach rather than per-episode quality.

The Seven Deadly Sins (2014, A-1 Pictures, 24 episodes, 6.60 Codex, 7.60 MAL, 2,214,380 members) gets a 7.2 on animation and 7.0 on cultural impact and then bleeds out on themes (6.0). Black Clover (2017, Studio Pierrot, 170 episodes, 6.52 Codex, 8.14 MAL, 1,907,373 members) scores in the mid-6s across every criterion — character 6.8, story 6.5, world 6.4, animation 6.0 — which is what genuine genre-average looks like when the rubric stops grading on enthusiasm. Soul Eater (2008, Bones, 51 episodes, 6.38 Codex, 7.86 MAL, 1,775,279 members) is the most defensible of the five — animation 8.0, world 7.0 — but its 5.5 story and 6.0 character scores reflect an anime-original ending the rubric cannot un-see. Fairy Tail (2009, Satelight, 175 episodes, 6.25 Codex, 7.57 MAL, 1,848,185 members) anchors the bottom.

These shows are not unpopular. They are not under-watched. They simply do not score what their membership counts would predict if popularity were a quality signal. The MAL scores themselves hint at this — every one of the five sits between 7.57 and 8.14 on MAL, a compressed band that the Codex spreads across a much wider range because the rubric weights character, themes, and story independently rather than averaging gut reactions.

The Steelman: Maybe Popularity Is Measuring Something Real

The strongest counter-argument is that popularity measures sustained engagement, and sustained engagement over 175 or 366 episodes is itself an achievement the rubric undervalues. A show that holds two million viewers through a decade of weekly releases is doing something a 24-episode josei masterpiece is not. That's true. It's also not a quality claim — it's a market-fit claim. The Codex rubric isn't designed to grade market fit. It's designed to grade story, character, themes, world-building, animation, and cultural impact, weighted by genre. When Fairy Tail scores 6.25, the rubric is not claiming nobody enjoyed it. It's claiming that on six specific axes, the show performs at a mid-6 level, and the 1.8 million members don't move those axes.

The r = 0.33 is the honest reconciliation. Popularity and quality are positively correlated. They are not the same variable. Anyone telling you otherwise is selling an aggregator.

Eleven percent shared variance is the actual relationship between how many people watched an anime and how good it is. The other eighty-nine percent is where the editorial work happens — and where every interesting show in the catalogue, in both directions, actually lives.