PLACEHOLDER DATA. The scores, dates and review text in this build are not final. Set placeholderNotice: false in src/lib/site.ts once every score is genuinely yours.
How I Score
A number means nothing without the rubric behind it. This is the whole system, published so you can disagree with it precisely.
RUBRIC VERSION: v1 · LAST CHANGED: when it changed, not when this page was built
The five axes
Every series is scored on five axes, each out of 20. The total is the sum, so every review lands somewhere between 0 and 100. Five stats of 20 keeps the arithmetic honest and forces a finer judgement than a 1–10 scale, which matters when the whole point is noticing small shifts.
| Axis | What it measures | Max |
|---|
| Story | Plot, premise, escalation, worldbuilding. | 20 |
| Art | Linework, composition, panel flow, colour. | 20 |
| Characters | Depth, voice, growth, likeability. | 20 |
| Pacing | Momentum, filler, arc structure, readability. | 20 |
| Payoff | Do the setups land? Arc and ending quality. | 20 |
| Total | The number you see everywhere on the site. | 100 |
What the numbers actually mean
Without anchors a score is arbitrary, and the archive stops being comparable with itself. So these are the descriptions I score against — written to be checkable, not to sound considered. If you can't picture which row a series belongs in, the series hasn't made it clear yet.
| Axis | What the numbers look like |
|---|
| Story | - 5You couldn’t say what it’s about
- 10Functional. Gets from A to B, no surprises
- 15Setups land forty chapters after they were planted
- 20Every reveal feels inevitable in hindsight
|
| Art | - 5Action is unreadable, figures inconsistent
- 10Competent and invisible. Nothing wrong, nothing memorable
- 15Composition is doing storytelling work
- 20I would stop reading to just look at a panel
|
| Characters | - 5Interchangeable — swap two and nothing changes
- 10One well-drawn lead, the rest functional
- 15I can predict their choices and they can still surprise me
- 20I think about them when I’m not reading
|
| Pacing | - 5Reading out of obligation. The chapter ended and nothing advanced
- 10Steady and unremarkable. Arcs end roughly when they should
- 15I lose track of the chapter count
- 20I would not cut a single panel
|
| Payoff | - 5Setups abandoned, or resolved by accident
- 10Things resolve, but flatly. Nothing accumulates
- 15Moments land because of something planted forty chapters earlier
- 20The ending recontextualises everything before it
|
Verdict bands
The total maps to a band. Bands are fixed — they don't move because a series is popular.Solid Roll is the one that matters, because it's the band that means I'd recommend it, and it's the line the decision on every card is computed from.
| Score | Band |
|---|
| 90–99 | Crit Success |
| 75–89 | Solid Roll |
| 60–74 | Serviceable |
| 45–59 | Filler Arc |
| 30–44 | Nerfed |
| 0–29 | Crit Fail |
Build labels
The total tells you how good something is. The build tells you what kind of thing it is, from the shape of the stats rather than the sum of them.
- Carried by Art — art is 5+ ahead of story.
- Coasting on Premise — story is 5+ ahead of art.
- Slow Burn Payoff — payoff is 5+ ahead of pacing, or the series climbed hard.
- Relentless Pace — pacing at 16 or above.
- Slogging — pacing at 9 or below.
- Wasted Potential — the trajectory is Falling or Glass Cannon.
- Well Rounded — none of the above, which is a compliment.
Trajectory labels
This is the part that only works because I re-score as I read. Computed from the checkpoint history:
- Slow Burn — started at 58 or below, now 75 or above, gained at least 20.
- Glass Cannon — started at 70 or above, dropped 20+, now 55 or below.
- Rising / Falling — moved 10 or more in either direction.
- Volatile — an 18+ spread between best and worst checkpoint.
- Steady — none of the above.
- DNF — dropped, and living in The Basement.
The rules I hold myself to
- Old scores are never edited. If my opinion moves, a new checkpoint is added and the reason is written down. The movement is the content — a silently revised number would make every graph on this site a lie.
- Chapter counts come from one platform only. Platforms split episodes differently. If I switched mid-series the x-axis would stop meaning anything.
- I score what I actually read. A first impression at chapter 2 is labelled as a first impression, never dressed up as a verdict.
- Nothing is a trend unless it moves. A single-point change on a twenty-point axis is noise, and noise is never described as a trend. A movement has to clear three points, or go the same way across two consecutive checkpoints, before this site will call it a direction. (I caught myself breaking this rule in an early draft — a one-point dip was written up as the first warning sign of a collapse. It was noise.)
- Levels are positions; only movement is a claim. I will never tell you that a 78 is better than a 74. I will tell you that something moved from 58 to 88, and you can check that.
- Spoilers stay out of the note. Enough to explain the score, never enough to ruin a turn.
Calls: the part that can be wrong
A score is an opinion about now, and opinions can only ever be different. A call is a claim about later, and a claim can be wrong — which is the only thing on this site that can actually prove whether my read on a series is worth your time.
Every call names the future checkpoint that resolves it, and the resolution is computed from the data, never decided by me. Misses are displayed with exactly the same weight as hits.
Calibration
The weakest point of any long-running scoring project is the scorer. My taste will drift over the years, and a graph spanning three of them risks showing my movement rather than the work's.
The fix is a calibration set: a small number of named series pinned to fixed scores that define what a 70, an 80 and a 90 actually feel like. Whenever the rubric version changes, the anchors get re-read and re-checked. Until that set exists, treat any comparison between a series I scored this year and one I scored next year as provisional — and so will I.
What the score is not
It isn't objective and it isn't a ranking of quality in the abstract. It's a record of one reader's experience, timestamped, so that when I say "this one gets better" you can see whether I actually believed that at the time.
A static review tells you the conclusion. A checkpoint log tells you whether the conclusion was earned.