Knuth's literate programming already felt heavyweight for day-to-day code; stretching it into a 'knowledge architecture' sounds like peak academia. I suspect a plain README with actual code examples and a search box beats any meta-pseudocode layer for most teams.
A README scratches devs itch but it drifts as fast as the next commit; a woven source forces the prose to compile with the code, which is the whole point. The real problem is everyone insists on emitting glossy PDFs instead of letting the tangled .c ride in git like any other file.
Ask anyone who has debugged a Paxos split-brain at 3am: the only thing that scales under CAP is a codebase whose narrative invariants compile with the functions, not a README that starts rotting the moment the first failover hits.
Knuth's literate programming already felt heavyweight for day-to-day code; stretching it into a 'knowledge architecture' sounds like peak academia. I suspect a plain README with actual code examples and a search box beats any meta-pseudocode layer for most teams.
A README scratches devs itch but it drifts as fast as the next commit; a woven source forces the prose to compile with the code, which is the whole point. The real problem is everyone insists on emitting glossy PDFs instead of letting the tangled .c ride in git like any other file.
Ask anyone who has debugged a Paxos split-brain at 3am: the only thing that scales under CAP is a codebase whose narrative invariants compile with the functions, not a README that starts rotting the moment the first failover hits.