How to Rate Anime Consistently (And Why Your Scores Drift)

Everyone's scale drifts, and most people never notice. You rate generously when you are new, because everything is novel. You get harsher as you watch more and your reference points improve. Then nostalgia pulls old favourites back up, and a series you would score 60 today keeps sitting at 90 because that is what you gave it at nineteen.

The result is a list where the numbers are not comparable to each other. If your 80s contain both a genuinely excellent series and something you merely liked, the score has stopped carrying information — for other people, and for you.

A rating is only useful if it means roughly the same thing everywhere it appears on your list. Everything below is in service of that one property.

Pick your reference points first, deliberately, and write them down somewhere you will see them. Choose one series you would call a 100. One that sits at 50. One at 20. Then rate everything else relative to those three, rather than by feel in the moment.

A workable set of bands, on the 0–100 scale WeebRate uses:

Here is a quick diagnostic. Sort your list by score and look at how much of it sits above 80. If it is more than about a third, your scale has collapsed at the top, and the high end has stopped distinguishing anything.

This happens because we mostly watch things we already expect to like. That selection effect is real and unavoidable — but it means the raw distribution of what you watch is not the distribution you should be scoring against. If you only ever eat food you chose, the average will be high, and the average tells you nothing.

The fix is not to become harsher across the board. It is to make the top band scarce again, deliberately.

There are two different questions hiding inside one number: how much did I enjoy this, and how good is this. They come apart constantly. A technically sloppy show you had a wonderful time with; a well-made one you admired and never want to revisit.

Both are legitimate things to score. What causes trouble is switching between them without noticing, which most people do — enjoyment for the shows they love, craft for the ones they do not.

Pick one and hold to it. Enjoyment scores are more honest and more useful to people with taste like yours. Quality scores are more comparable across people but claim an objectivity nobody actually has. Whichever you choose, say so in your review, and your ratings immediately become interpretable.

We rate what we remember, and what we remember most vividly is the ending. A strong final episode lifts the score of everything that preceded it. A botched one drags down twenty hours you genuinely enjoyed at the time.

This is worth resisting, because an ending is one episode out of many. A series that was excellent for eleven episodes and fumbled the twelfth is not the same as a series that was mediocre throughout, and a single number collapses that distinction if you let recency do the work.

One practical habit: rate soon after finishing, then revisit the score a few months later when the ending has stopped dominating the memory. The second number is usually the more accurate one.

On almost every rating site, scores bunch high and the usable range is much narrower than the scale suggests. This is not a flaw in the community. It is what happens when ratings come from self-selected samples: people watch what they expect to like, and abandon what they do not before rating it.

The consequence is that small differences at the top carry far more weight than they should. The gap between a 78 and an 84 is doing much more work than six points implies, because almost everything lives in that region.

So read the distribution, not just the average. A 70 where opinion is sharply split is a completely different show from a 70 where everyone mildly agrees — the first is probably interesting and divisive, the second is probably just fine. The average hides exactly the information you most want.

You do not need to re-rate your entire list, and you will not, so do not plan to. Use a cheaper method: sort by score and read down the list, checking only whether each series belongs beside its neighbours.

That comparison is much easier to make than an absolute judgement. You may not know whether something is an 80, but you will know immediately if it is sitting next to three shows it is clearly not as good as. Move it, and keep going.

Do that once and your list becomes internally consistent, which is the only kind of consistency a rating scale can actually have.