The Decade-by-Decade Codex Numbers Say Anime Hasn't Gotten Better. They Also Say It Hasn't Gotten Worse.

The "has anime gotten better over time" debate dies on contact with the rubric: every decade since 1980 averages between 7.00 and 7.41, and the gap is mostly catalogue bias.

"Anime peaked in the 80s" is a forever-argument, and so is "anime peaked in 2009," and so is "anime is in a golden age right now." The rubric won't settle it, but it will show what the numbers actually say versus the nostalgia. The numbers, it turns out, are boring. They are also more honest than the discourse that surrounds them.

The consensus everyone keeps recycling

The standard takes split three ways. The OVA-era loyalists treat the late 80s as a creative ceiling the industry has been sliding down ever since. The Bones/Madhouse romantics anchor the 2000s as the last decade of "real" craft before digital pipelines flattened everything. And the streaming-era boosters point at Ufotable compositing, MAPPA throughput, and Frieren and argue the medium has never been stronger. MyAnimeList encourages all three positions simultaneously, because its top 100 is a recency-biased popularity contest with a survivorship-biased classics tail glued on.

Anime Codex does not score on popularity. The rubric weights story, character, themes, world-building, animation, and cultural impact per genre, and the resulting decade averages — across the catalogue we've actually graded — collapse the argument into something the discourse refuses to accept: the medium has been remarkably consistent for forty-five years, and the variance between decades is smaller than the variance between any two shows airing in the same season.

What the decade averages actually say

Here are the numbers, full stop. 1980s: 7.41 average across ten entries. 1990s: 7.40 across twenty-three. 2000s: 7.00 across seventy-four. 2010s: 7.41 across fifty-five. 2020s: 7.34 across twenty-one. The 1980s, 2010s, and 2020s are functionally identical. The 1990s sit on the same line. The 2000s are the only decade that visibly dips, and that dip is the most interesting data point in the table — which we'll come back to.

If you came here expecting a clean upward curve to validate the "we live in a golden age" position, the rubric refuses you. If you came here expecting a clean downward slope to validate the "anime is cooked" position, it refuses you too. The averages are flat. The standouts are not.

The standouts tell a different story than the averages

The top of each decade does move, and it moves upward. Ashita no Joe leads the 1980s at 9.02 — a 47-episode Tokyo Movie Shinsha production whose Codex score is carried almost entirely by character (9.5), themes (9.4), and cultural impact (9.8), with animation pulling a 8.0 that reflects the limits of the era's cel work rather than any failure of direction. The 1990s peak with One Piece at 8.58, the 1999 Toei entry whose story (8.5), character (9.0), and themes (8.7) hold up even as the pacing discourse refuses to let it breathe.

Then the ceiling jumps. Fullmetal Alchemist: Brotherhood lands the 2000s at 9.25 — Bones, 64 episodes, a story score of 9.5 and an animation score of 8.8 that together make a case for it being underrated even at the top of MAL. Hunter × Hunter (2011) takes the 2010s at 9.23, Madhouse's 148-episode adaptation earning a 9.5 in world-building that no shonen of the prior decade can match. Frieren: Beyond Journey's End closes out the current decade at 9.03, with a themes score of 9.5 and animation of 9.2 from Madhouse's 28-episode run.

The peaks have risen from 9.02 to ~9.25 and held there. The averages haven't moved. Both things are true.

The catalogue bias nobody wants to talk about

Before anyone weaponizes the standouts into a "modern anime is better" narrative, look at the sample sizes. 1980s: n=10. 1990s: n=23. 2000s: n=74. 2010s: n=55. 2020s: n=21 — and the decade is half over.

This is selection bias and survivorship bias colliding in real time. The 1980s sample is tiny because we grade what's worth grading, and forty years of attrition have already filtered the decade down to its defensible canon — Ashita no Joe, the Miyazaki theatrical work, the OVA tier that survived the bubble. The 1990s sample doubles because more work survived in cultural memory and more of it was exportable. The 2000s explode to 74 because that's when the catalogue meets contemporary fandom — the decade where Anime Codex has the most coverage because it's the decade modern viewers actually litigate. The 2010s and 2020s entries are still being added.

Which means: the 1980s 7.41 is the score of a pre-filtered canon. The 2000s 7.00 is the score of a decade we've graded broadly, including the mid-tier and the failures. If we'd only graded ten 2000s shows, we'd have picked Brotherhood, Monster, Code Geass, Gurren Lagann, Death Note's first arc, and four others, and the average would be 8.5+. The 1980s average is not high because the 1980s were better. The 1980s average is high because we threw out the chaff before grading.

Why the 2000s dip is actually a compliment

The decade that looks worst on paper is the one with the most rigorous coverage, the one where we've graded the median and not just the canon. That's where you find shows like Death Note collapsing post-Yotsuba dragging an average down that nostalgia would otherwise float. The 7.00 average is not a verdict on the 2000s. It's a verdict on what an honest catalogue looks like when you don't pre-filter.

Apply the same honesty to any decade and the numbers converge. The 80s would drop. The 90s would drop. The 2010s and 2020s, once the catalogue fills in and the mid-tier MAPPA throughput and the forgotten seasonal isekai get graded, will drop too.

The counter-argument worth taking seriously

The strongest opposing read: averages are the wrong instrument. What matters is ceiling, and the ceiling has risen — from Joe's 9.02 to Brotherhood's 9.25 to HxH 2011's 9.23 to Frieren's 9.03. The argument continues: production capacity has expanded, more studios can hit the top tier, and a Madhouse working in 2023 has tools a Tokyo Movie Shinsha of 1980 didn't.

This is correct on its own terms and the rubric agrees up to a point. Animation scores have risen — Frieren's 9.2 is unreachable in 1980, full stop. But the rubric weights six criteria, and the other five do not show the same curve. Joe's 9.8 in cultural impact is a number Frieren cannot earn yet and may never earn. Ceilings rise on one axis and stay flat on others. That isn't "better." That's "different."

The verdict

The decade averages don't move because anime, graded honestly across six criteria, has been operating in the same quality band for forty-five years — the production stack has changed, the ceiling on animation has risen, and the rubric reflects both without pretending either constitutes progress. "Anime peaked in the [decade]" is a sentence that requires you to ignore five of the six criteria. The numbers don't care which decade you're loyal to.