The problem: promises that cannot fail

Listen to a typical promise: “We will improve healthcare.” Improve by what? For whom? By when? At what cost? A promise without those answers cannot fail — and therefore cannot succeed.

In serious systems, vagueness kills. p99 under 40 milliseconds at 10,000 requests per second, measured on production hardware, sustained for one hour. Then everyone knows what done means. He holds politics to the same cruelty toward fuzzy language.

The principle: hold language to load

No vagueness means every commitment passes a five-field test: what changes, by how much, by when, at what cost, how it is verified. No field may stay blank.

This discipline also governs failure. A missed target is not a communications problem. It is data. The miss is published within seven days, the cause named, the fix dated. A ledger that records misses is government.

If it cannot be measured at peak hour by a citizen with a phone, it is not a promise. It is a wish.

The program: three measurable commitments

  • 1. Five-field promises only. Every pledge states metric, baseline, target, deadline, budget, owner, and verification dataset. 100 percent carry all seven fields.
  • 2. Peak-hour verification. Services measured when citizens need them — 7 to 10am for clinics and transport. Monthly p50 and p90 waits for the 20 busiest service points. Averages alone are banned.
  • 3. A public miss log with fixes. Missed targets published within 7 days with cause, corrective budget, new date, and owner. Zero silent revisions, zero deleted baselines.

What this builds

Precision builds comparability. When every promise carries the same five fields, citizens can compare clinic against transport, this year against last year, one city against another.

Judge him by the numbers, never by his intentions. A figure who fears that table fears the reader. He intends to publish it every month, on time, whether it flatters him or not.