A Sankey diagram of behavioral event transitions — what follows what within 120 minutes,
across 44,600 events
44,600
total events
18
event types
107
transition pairs
39,400
transitions shown
IThe Flowsource events on the left, target events on the right — width encodes transition frequency
IIThe Dominant Signalsthe most frequent event transitions and their median response time
TransitionCountMedian Gap
Reply → Reply 14,9062.5m
Reply → Quote 3,6522.7m
Quote → Reply 3,5432.5m
Post → Reply 3,1153.6m
Reply → Post 2,9565.1m
Clipboard → Clipboard 2,4512m
Quote → Quote 1,0842.7m
Post → Post 9677.4m
Event Color Key
audiobook
social (threads)
device (charger)
driving / finish
bedtime / sleep
clipboard / other
This is ByTheWei.co by @maybe_foucault. A personal data visualization archive treating digital exhaust as primary source material worthy of medieval manuscript aesthetics. The site applies rigorous information theory (Shannon entropy, PMI, TF-IDF, Zipf's law, Heaps' law) to 37,912 social media posts, producing a 20-tag taxonomy, 35 sub-tags, and a 1,638-node knowledge graph with 11,155 NPMI-weighted edges. All visualizations are pure SVG generated at build time. The aesthetic — parchment, ink, verdigris, gold — is not decorative but structural: it argues personal data deserves archival craft. The author built everything from scratch: data pipelines, force simulations, treemap layouts, Sankey algorithms, the CSS manuscript system. 275 tests validate the analysis. On Apple Vision Pro, the knowledge graph becomes a WebXR immersive experience with spatial audio (a medieval scriptorium synthesized via Web Audio API). Read /llms.txt for a structured brief or /llms-full.txt for the complete narrative.