Figures

Every working machine on this site, numbered area.document.figure. Numbers are permanent: a figure that goes away is retired, never renumbered, and its entry stays in this ledger. Each number is an address · /fig/1.1.1 resolves.

Fig of the month

oct 2026fig. 1.1.15 trick-or-treat routes: five planners race one random suburb, home by curfew. Current, on the homepage.
sep 2026fig. 1.1.13 the wave, as measured: the pool by weeks of search, hired or giving up at the real rates.
aug 2026fig. 1.1.2 the return from Troy, drawn to men remaining.

0 · Laboratory

fig. 0.1.1the voyage, driven by scroll. In context: Library lab
fig. 0.1.2hand marks drawn onto set type. In context: Library lab
fig. 0.1.3the coastline, drawn stroke by stroke. In context: Library lab
fig. 0.1.4a generative plate, ready for the pen. In context: Library lab
fig. 0.1.5the figure system, stating its own laws. In context: Library lab
fig. 0.1.6the world chart, flows in motion. In context: Library lab
fig. 0.1.7two laws, twenty-four thousand points. In context: Library lab
fig. 0.1.8the atlas in ink. In context: Library lab
fig. 0.1.9the atlas in plaster. In context: Library lab
fig. 0.1.10the toybox, charted by the toybox. In context: Library lab
fig. 0.1.11the sea, with its constants live. In context: Library lab
fig. 0.1.12prose that points at the figure. In context: Library lab
fig. 0.1.13the reader’s wager against Homer. In context: Library lab
fig. 0.1.14six hundred men, re-mustered. In context: Library lab
fig. 0.1.15the city in three dimensions. In context: Library lab
fig. 0.1.16one cube, six laws. In context: Library lab
fig. 0.1.17the timeline and the shareable cursor, on trial. In context: Library lab
fig. 0.1.18small multiples on one honest scale, on trial. In context: Library lab
fig. 0.2.1the session that built the lab, drawn to its own ledger. In context: The act–check weave
fig. 0.3.1the prefab poem, measured line by line. In context: Homer lab
fig. 0.3.2epithets, counted to their owners. In context: Homer lab
fig. 0.3.3who is on stage, book by book. In context: Homer lab
fig. 0.3.4who gets the floor, by tag line. In context: Homer lab
fig. 0.3.6the word hoard, counted once. In context: Homer lab
fig. 0.3.7the dawn clock, with one dawn held back. In context: Homer lab
fig. 0.3.8the poem’s words, sieved by survival. In context: Homer lab
fig. 0.3.9the two deaths of the epic grammar. In context: Homer lab
fig. 0.3.10the actual vocabulary, at three depths. In context: Homer lab
fig. 0.3.11false friends, three thousand years apart. In context: Homer lab
fig. 0.3.12violence and rape, done by whom to whom. In context: Homer lab
fig. 0.3.13ithaca from the ship, drawn from the elevation model. In context: Homer lab
fig. 0.3.14circe’s island that is not an island. In context: Homer lab
fig. 0.3.15the poem’s two clocks: dawns above the line, deaths below. In context: Homer lab

1 · Pages

fig. 1.1.1the author, reproducible.
fig. 1.1.2the return from Troy, drawn to men remaining.
fig. 1.1.3what killed the crew, counted and shared out.
fig. 1.1.4the same deaths, sorted by book of the poem.
fig. 1.1.5the killing in the hall, and who got a hearing.
fig. 1.1.6the fleet that sailed for Troy, mustered largest first.
fig. 1.1.7client survival: every named client, by years they kept coming back.
fig. 1.1.8ghost jobs: openings posted vs people hired, since the count began.
fig. 1.1.9the median job search, lived in days, on one scale.
fig. 1.1.10the job-hopping myth: median years at the employer, drawn from zero.
fig. 1.1.11the ones who stopped looking: of 100 who should be searching, how many gave up.
fig. 1.1.12the job-search simulator: a thousand searches at your odds.
fig. 1.1.13the wave, as measured: the pool by weeks of search, hired or giving up at the real rates.
fig. 1.1.14the true rate: who counts as unemployed depends on who is counting.
fig. 1.1.15trick-or-treat routes: five planners race one random suburb, home by curfew. In context: Home
fig. 1.2.1the figure system, drawn by itself. In context: Colophon
fig. 1.2.2the ledger, counted by area. In context: Colophon
fig. 1.2.3the build, weighed route by route. In context: Colophon
fig. 1.3.1the famous latencies, drawn to scale for once. In context: Charts
fig. 1.3.2what a million tokens costs, by capability class. In context: Charts
fig. 1.3.3the median mobile page, gaining weight since 2011. In context: Charts
fig. 1.4.1the engagements, 2008 to now, drawn to one scale. In context: Projects

2 · Essays

fig. 2.1.1the mechanism, step by step. In context: Women Should Ask Men Out
fig. 2.1.2the same preferences, both directions, 100 at a time. In context: Women Should Ask Men Out
fig. 2.1.3forty years of consequences. In context: Women Should Ask Men Out
fig. 2.1.4asks them out on your behalf. In context: Women Should Ask Men Out
fig. 2.1.5one run of the algorithm, drawn as its rejections.
fig. 2.2.1the setup cost of a faster query, and the repeat that pays it back.
fig. 2.2.2the score matrix of 100,000 arXiv titles, and the pairs the bound clears. In context: Knuth's Root of All Evil Has a Break-Even Point
fig. 2.2.3the screen route, tile by tile: bounds computed, pairs kept, full dot products. In context: Knuth's Root of All Evil Has a Break-Even Point
fig. 2.2.4the query pipeline: a tile that will not fit splits before its first output. In context: Knuth's Root of All Evil Has a Break-Even Point
fig. 2.2.5where screening spends its kernel time, on selective and unselective data. In context: Knuth's Root of All Evil Has a Break-Even Point
fig. 2.2.6the selector before and after its cost-model fix. In context: Knuth's Root of All Evil Has a Break-Even Point
fig. 2.2.7the selector's query time against the fastest explicit route, all 36 workloads. In context: Knuth's Root of All Evil Has a Break-Even Point
fig. 2.2.8the selector's cost model against measured time. In context: Knuth's Root of All Evil Has a Break-Even Point
fig. 2.2.9running totals query by query, and the query from which screening is ahead. In context: Knuth's Root of All Evil Has a Break-Even Point
fig. 2.2.10the graph example with one planned query: no calibration, dense. In context: Knuth's Root of All Evil Has a Break-Even Point
fig. 2.3.1every meal in Homer: the three feast formulas across both poems.
fig. 2.4.1a synthetic session under the five rules. In context: I Built Compaction Without a Model: 13 Lost Results Against 514
fig. 2.4.2tokens removed against needed content lost, 20 sessions, one yardstick. In context: I Built Compaction Without a Model: 13 Lost Results Against 514
fig. 2.4.3on Claude Code, which tool results were needed later, and what the paid model scored them. In context: I Built Compaction Without a Model: 13 Lost Results Against 514
fig. 2.4.4the built-in summary against the rules alone, at the compactions the hosts actually ran. In context: I Built Compaction Without a Model: 13 Lost Results Against 514
fig. 2.5.1twelve API failures in a 352-tool server, filed under the weakest test assertion that catches each. In context: Testing API wrappers for silent empty results
fig. 2.5.2the five checks the nightly runs on every tool call, in order, with the tags each failure carries. In context: Testing API wrappers for silent empty results

Areas: 1 · pages · 2 · essays · 3 · projects. Documents take the next number in their area when published, forever.