Opening the stacks
Opening the stacks
Instruments · Phylogeny
Each idea on the shelf as a lineage through the years its books were written: what it branched from, where it went quiet and came back, and where it stops.
677 of 715 books · 95%
It sees books the graph has read and that carry a publication year.
38 have no publication year. Of 1131 ideas, 329 appear in more than one dated book, and 152 appear in at least 4, which is the least that lets a silence be measured.
Every line of descent here is inferred. A model proposed the ideas and the relations between them from the books’ summaries, in one pass. None is attested by a source. Each relation is shown with the model’s own description.
An idea is absent from many books that could have carried it, so absence alone proves nothing. For each idea the instrument works out its usual rate: the books that carry it, over the books written while it was active. For a silence it counts the books the shelf holds from inside those years. If the idea would almost certainly have shown in that many books (the chance of seeing none by accident is under 1 in 100), the silence is counted. If the shelf holds too few books from those years, the silence is fog and is not counted. A silence must also last at least 15 years.
A name can change while the idea goes on. When an idea is quiet but one the graph marks as similar does appear in those years, the silence is put down to the neighbour and is not counted either.
163 silences were tested in this reading. About 2 of them would pass that test by chance alone. 802 more ideas appear in a single book and have no lineage to draw. Appearances are what the graph recorded for each book, not a reading of every page.