The Canon in aggregate

Books are more than half of it, which says less about the three media than about how much more a reader gets through in a life. Click a segment to open that medium in the browse grid.

Scroll to zoom · drag to pan · double-click to reset

Each medium as a share of itself rather than a raw count, because books outnumber films more than two to one and anime nearly five to one, and raw bars would draw the sizes of the three collections instead of the shape of each. This is the comparison the merged list exists for: anime peaks a year below film and books, and each medium owns one end of the range. Only books reach down to six, only films reach up to seventeen, and anime is the narrowest band of the three, bunched hard around thirteen.

Scroll to zoom · drag to pan · double-click to reset

A steep hill centred on thirteen and fourteen. The shape is a curatorial fact rather than a fact about the media: this is the range the curator reads and watches in, and the thin tails at six and seventeen are where the collection has the least to say. Click a bar to open that age in the browse grid.

The canon by genre, proportioned by how many works carry each group. Science fiction is on almost half the works, which is an honest record of one curator's appetite rather than a claim about what is worth reading. Click a tile to open that group in the browse grid.

How often two groups land on the same work, drawn for every pairing that reaches five works. Thick chords are habitual pairings; hover a node or a chord to isolate one genre's relationships and push the rest back.

Scroll to zoom · drag to pan · double-click to reset

Four works from antiquity, a Shakespeare cluster around 1600, then a climb that starts in the 1950s and crests in the 1990s. Part of that is film and anime not existing before the twentieth century, and part is survivorship: the further back you go, the harder a work has had to work to still be here.

One bubble per decade and complexity age, sized and lit by how many works sit there. Reads as a band rather than a trend, which is the finding: the canon's complexity spread has not moved much across seventy years of it.

Alphabetical round the dial, so the shape is a reading and not an artefact of sorting. History and war set the highest bar, with dystopia and literary fiction just behind; children's and young adult sits two whole years below anything else, which is the one place on this page where the number behaves exactly as you would guess.

Scroll to zoom · drag to pan · double-click to reset

Each bubble is a genre group: across is how many works it holds, up is what it asks on average, and the bubble's size is how far its works spread either side of that average. A wide bubble low on the axis is a genre the canon carries at every level; a small tight one high up is a genre it only visits at the deep end.

Sized by how many titles carry each word, with the ordinary machinery of English stripped out. What is left is a fair summary of what this collection is about: worlds, children, kings, ghosts, space, life. Click a word to search the canon for it.

Scroll to zoom · drag the background to pan · drag a node to pin it · click a creator for their card · hover an edge for the genre tags two creators share

Directors, studios and authors in one graph, because a creator is the one field all three source sites carry. Anyone with at least 2 works in the canon, linked wherever two of them share 3 or more raw genre tags; the count beside the Detail slider says how many creators and links that leaves, and most of the — creators in the canon never reach it, because they are here for exactly one work. Nobody in this graph collaborates with anybody, so an edge means these two make the same kind of thing, and its weight is how many tags their shelves have in common. Node size is work count and node colour is the medium, the same three colours as the doughnut at the top of the page and the badge on every card in the browse grid; the shaded hulls behind them are the groups of similar creators the layout settles into. Everything drops the floor to 2 shared tags; Tight raises it to 3 works and 4 tags. Use the three medium swatches to mute a medium and see whether the directors really do sit apart from the novelists.

Scroll to zoom · drag the background to pan · click a genre for its card · hover an edge for the works that carry both

All — genre groups, linked wherever at least 5 works carry both; the count above the graph gives the connections that leaves. Node size is the number of works in the group and link thickness is how often the two appear together. Ring lays the same groups out on a circle so the edge weights can be read against the chord diagram further up. This graph covers all — works, because every work carries genres whichever site it came from.

— of — works. Themes come from the film and anime sites, which record them; the books site never has, so all — books are absent from this graph. Read it as a map of the films and anime, not of the canon.

Scroll to zoom · drag the background to pan · click a theme for its card · hover an edge for the works behind a pairing

The themes that occur on at least 4 of those — works, out of — in the data, linked wherever two of them land on the same work at least twice; the count beside the Detail slider gives the themes and links that leaves. Themes are far more specific than the genre groups above, so where the genre network shows one science fiction node this shows artificial intelligence, technological alienation and body horror as three separate neighbourhoods. Node size is how often a theme occurs, colour depth is how many other themes it pairs with, and the hulls are the communities the pairings fall into, each captioned with its best-connected theme. Everything drops the floor to 3 occurrences; Tight raises it to 6.