Why the Welles database refuses my field notes

The most important schema decision in the Welles database is what it refuses to store: my field notes.

Welles is an ethnography first. The fieldwork side runs on a margin-coded notebook system (D for discipline, F for family, $ for economics, H for hierarchy, C for childhood, R for risk, T for training ritual), expanded into digital summaries the same evening. That material is interpretive by nature and it belongs in the research vault.

The database is the other instrument: source-backed ecosystem facts only. Gyms, events, fights, development pathways, all traceable to public records, promotion pages, and posted fight cards. The structured map so far covers 113 gyms and 90-plus documented fights across 37 events in Northern Thailand, from village grassroots programs up to ONE Lumpinee.

The decision: a hard boundary between the two. Fieldwork can inform what I look for in the ecosystem data, but interviews and observations never populate rows. Mixing them would quietly convert impressions into facts, which is exactly the failure mode ethnography gets accused of.

The tradeoff: the database grows slower, because everything needs a source. I am also deliberately early on the SQL itself, working through a gated curriculum before any production schema exists, so right now the "database" is one learning table and a rule book.

The thing that broke: nothing yet, in code. What broke was my assumption that one project needed one data store.