Our methodology
A shared language for how fantasy feels to read, what it asks from you, and why two books belong together.
We compare books dimension by dimension. Two books can have the same sum and still sit far apart because their strengths are distributed differently.
Taste
Weighted similarity across sixteen dimensions
Context
Subgenre and guide intent adjust importance
Fit
Audience and commitment checks applied
Matrix 1.0.0
Taste = 100 × Σ(weight × similarity) ÷ Σ(weight)
Fit = clamp(Taste - penalties, 0, 100)
Inside the percentage
Every contribution remains visible, so a recommendation can name both its strongest matches and its biggest differences.
Base weights are stable. A source book's subgenres increase the dimensions most likely to define that reading experience: political fantasy raises politics and complexity; romantic fantasy raises romance and character focus; cozy fantasy raises tone and low darkness. A guide can then state additional intent explicitly.
Audience and maximum commitment can reduce a score. A requirement for a complete series, or content the reader has explicitly excluded, makes a candidate ineligible rather than merely less similar.
Compatibility, not taste
Audience describes who a book was primarily written for. It does not make a book darker, simpler, or less character-driven, so it never changes the meaning of a taste value.
Audience compatibility is asymmetric. An adult reader can reasonably be shown a Middle Grade crossover book. A younger reader should not automatically be shown an adult book simply because its politics, magic, or darkness score is numerically close. Generic taste comparisons omit this penalty; guides and future personalized recommendations apply it when reader context is known.
Darkness describes the overall reading experience. Reader advisories identify particular kinds of material, so two equally dark books can still require very different guidance.
Advisories do not change taste similarity. They support filtering and can make a candidate ineligible when a reader explicitly excludes that content. They are practical guidance rather than an exhaustive scene-by-scene warning system.
These use a 1-5 scale because the difference between low and high intensity remains useful across fantasy subgenres.
How much the book depends on power games, factions, dynasties, court politics, strategy, and betrayal.
How brutal, bleak, cynical, violent, or emotionally heavy the book feels.
How much the book depends on character voice, interiority, relationships, and personal change.
How many threads, reveals, factions, dependencies, and long-range payoffs the reader tracks.
How far cultures, institutions, history, geography, and systems are developed beyond the immediate plot.
Categories avoid false precision. They are easier to understand and more honest about the judgment involved.
A close taste match is less useful if it quietly leads into a long unfinished series.
Profile scope
A long series can change tone, darkness, politics, and complexity as it develops. One flat average can hide the exact information a reader needs.
Each volume can carry its own Matrix values, audience, and advisories. These book-level values are the source of truth.
A series page summarizes the included books. When the experience changes materially, it shows the direction of travel instead of pretending the whole series has one stable profile.
For a series such as Harry Potter, early and late volumes would therefore be scored separately. The series overview could explain that the story becomes darker, more political, and more demanding over time.