The membrane splits into two hemispheres: the corpus (sporeprint � 127 papers, search index, RSS) and the cortex (everything else � feeds, JS, HTML, data, images). The corpus is dense and heavy; the cortex is wide and light.
At each bend number (1, 3, 7, 14, 22, 42, 69, 137, 314, 420, 666, 1000, 1618), we measure how much of the total is captured by the top N files. The curve tells us the compression shape of the organism � a power law with a long tail.
| bend # | files | MB | % of total | mean size | ? key |
|---|
Cumulative bytes captured � the S-curve of compression. The steep rise at left = the 7 giants. The long tail at right = hundreds of tiny feeds.
How bytes are allocated across file types. HTML and JSON dominate by count; the search index (JS) and atom.xml dominate by bytes.
| type | files | % files | bytes est. |
|---|
7 files = 33.1% of all bytes. This is the Pareto peak � the nucleus of the organism.
| # | file | size |
|---|
Each category as a proportion of the whole. Tufte's small multiples let the eye compare without legend.
894 files, 59 MB, 252K lines. The minimum viable membrane � remove any file and something breaks. This is 37.5% of total bytes but 51.3% of total files. The core is many small files; the periphery is few large ones.
At file #314 (p � 100), the mean file size is 420,149 bytes. 420 � the sourdough bend number. p talks to itself through the data. The numbers are not assigned; they are measured.