Snapshot 3 · 2026-05-08 · archived by choice
1 month · what existed before the reset.
This graph captures the structure on the day before we decided to do the radical reset. 1,057 files, 3,154 nodes, 331 communities. Look at the size. Compare with Day 1. It becomes visible why wiping cost less than carrying on.
- 3,154 nodes
- 5,109 edges
- 331 communities
- 1,057 files
- ~3M estimated generation tokens
Interactive · heavy · may take a few seconds
The graph
This is the largest of the three (~3 MB of data). On a slow connection, loading takes a moment. If it stalls, refresh and try again.
How to read · three symptoms of overgrowth
Why we chose to wipe
1 · 331 communities, 101 of them so thin they were left out of the report. When a system produces more clusters than you can navigate, structure has turned into noise — small isolated corners with no bridge to the whole.
2 · 346 isolated nodes (≤1 connection). More than 10% of the total. Documentation no one referenced, orphan files, overlap. The cost of maintaining was higher than that of starting over with the lessons preserved.
3 · The lessons did survive. What was worth keeping stayed: base ADRs, pipeline patterns, governance decisions. The rest was scaffolding that had finished its job and never got removed.
What is not in here
Radical honesty about this snapshot
This graph is an archaeological photograph. It does not represent farpa today and does not guide current decisions. It is here for one purpose: to make the reasoning behind the reset visible. Paths sanitized, no personal data, no secrets.