The Decade Nostalgia Says Anime Peaked In Is Not the Decade the Rubric Scores Highest

Walking the Anime Codex averages decade by decade shows that "has anime gotten better over time" is a worse question than the fandom thinks — and the answer is duller than either side wants.

The nineties won. Not by the margin the Bebop-poster crowd insists on, not by the landslide the Evangelion sermon demands, and not in the criteria anyone brings up at 2 a.m. on a subreddit. By nine-hundredths of a point over the 2010s, and half a point over the 2000s, on a sample of twenty-four shows against catalogues two and three times larger. That is the entire empirical case for the golden-age argument, and it deserves to be handled like the fragile artifact it is.

"Anime peaked in the [decade]" is a forever-argument. The rubric won't settle it, but it will show what the numbers actually say versus the nostalgia. MyAnimeList's top 100 is a survivorship museum — the 1998 shows that made it there are the ones that survived twenty-six years of re-watching, and the 2023 shows on it haven't been tested by anything yet. Anime Codex weights six criteria (story, character, themes, world-building, animation, cultural impact) and reports what the current catalogue looks like when you stop letting nostalgia do the grading. What follows is not a defense of any decade. It is a defense of reading the numbers before writing the eulogy.

The 1990s Post the Highest Average, and It Means Less Than You Think

The 1990s Codex average is 7.46 across nine shows. That is the winning number. It is also nine shows. The catalogue for that decade is dominated by titles that had to be great to still be discussed — Cowboy Bebop's 8.71 sits at the top, carried by a 9.5 on cultural impact and an 8.2 world that Sunrise built out of jazz cues and Watanabe's Western-noir grammar. Nothing about that score is wrong. What is wrong is treating a nine-title sample as proof of a decade's quality when the selection was pre-filtered by two and a half decades of canonization.

Ask what the 1990s Codex catalogue would look like if you forced it to include every OVA nobody remembers, every twelve-episode romantic comedy that aired on a Tuesday, every Toei filler-machine that ran for a hundred episodes and got cancelled. It would look like the 2000s. Which is to say: it would average around 7.03.

The 2000s Average Is the Lowest, and That Is a Compliment

The 2000s post 7.03 across seventy-two shows. That is the lowest decade average in the catalogue, and it is also the decade that produced Fullmetal Alchemist: Brotherhood — a 9.25 whose 9.5 story, 9.4 character, and 9.5 cultural marks put it above every 1990s title in the catalogue including Bebop. Bones delivered sixty-four episodes without a structural sag, and the rubric rewards that with the highest Codex score in the entire five-decade sample.

The 2000s look worse on average because the sample is honest. Seventy-two shows is enough to include the Naruto filler catastrophe, the Bakuman romance-vow drag, the School Rumble parody detour, and every mid-tier harem the era produced. Higher n means more middle. More middle means lower average. That is arithmetic, not decline.

The 2010s Are the Decade the Rubric Actually Likes Most, Once You Adjust

The 2010s average 7.40 across fifty-five shows and produce Hunter × Hunter (2011), Madhouse's 148-episode adaptation that scores 9.23 with a 9.5 world and a 9.5 across story and character. That world score is the highest in the decade's standouts — higher than Brotherhood's 9.0, higher than Bebop's 8.2 — and it is earned across arcs (Chimera Ant, Election) that the medium had not previously proven it could execute at that length without collapsing.

A 7.40 average on n=55 is quietly the strongest performance in the catalogue. It is closer to the 1990s (7.46) than the 1990s is to the 2000s (7.03), and it survives a much larger sample. If the question is which decade the rubric grades most generously per unit of catalogue coverage, it is not the nineties. It is the 2010s, by a whisker, in a photo finish nobody wants to call.

The 2020s Are a Fragment, and Frieren Is Not a Trend

The 2020s post 7.30 across thirty-three shows. The decade is four years in when you read this, and the standout — Madhouse's Frieren: Beyond Journey's End, a 9.03 with a 9.5 themes and a 9.2 animation — is a twenty-eight-episode fantasy that succeeds by being slower than its era. That is not evidence the decade is better. It is evidence one show is very good and hasn't lost points to any of the failure modes that drag longer catalogues down.

Ask again in 2029. The 2020s will have their Gachiakuta-style back-half collapses averaged in, their idol franchises padding the middle, their Spy × Family sequels asking whether cultural footprint can carry a rubric that character work alone won't. The 7.30 will move. It will probably not move up.

The 1980s Are a Rounding Error the Discourse Treats as a Foundation

The 1980s post 7.35 on n=9 with Ashita no Joe's 1980 entry (9.02 Codex, 9.5 character, 9.4 themes) as the standout. Nine shows. That is not a decade. That is a shelf. Tokyo Movie Shinsha's forty-seven-episode boxing epic is doing most of the lifting, and the rubric agrees with the lift — character work that high, on a shonen weighting, will carry a scorecard. But treating a nine-title sample as a decade grade is the same category error as reading a Golden Age from the survivors of a purge.

The Counter-Argument: Nostalgia Isn't Wrong, It's Just Not Grading

The strongest opposing case is that the nineties felt different because they were different — Bebop, Utena, Evangelion, Trigun, Escaflowne existed inside a production culture that hadn't yet been streaming-optimized, and the rubric can't grade a vibe. Fine. The rubric doesn't try to. It grades six things, weights them by genre, and returns a number. What it exposes is that the 1990s' 7.46 average and the 2010s' 7.40 average are separated by less than one round of Codex scoring noise, and the 2010s produced two shows (Hunter × Hunter, and others outside this piece) that hit the same 9.2+ tier as anything the nineties has to offer.

The vibe argument is a real argument. It is not a scoring argument. When the cultural criterion pulls a scorecard around — as it does for many older titles the fandom still cites — the rubric captures that. It just doesn't let it overrun story, character, themes, world, and animation on its own.

Anime did not get better. It did not get worse. It got larger, and larger catalogues regress toward their middle. The rubric says the nineties edge the field by an interval smaller than the confidence you should place in a nine-title sample, and it says the decade that actually delivered the two highest single scores in the catalogue is the 2000s and 2010s. Peak is a story people tell. Average is what the numbers show, and the numbers show a flat line with spikes.