Paper № II · Transit, variance · May 2026 · 6 pages
Some NYC Subway Transfers Are Mathematically Worse
A variance-aware argument for better routing.
Sample entry Seeded from real work while the archive is under construction. The published paper replaces it.
The MTA’s route planner treats a transfer as an average wait. Riders experience it as a gamble. This paper makes the gamble precise: when arrival times carry real variance, some transfers (Court Square among them) are measurably worse than the averages say, and a variance-aware router would send you a different way.
Every paper gets a behind-the-paper essay on Substack: the thinking, the dead ends, the data and code.