Find the leading indicator that appears days before…

Why This Is a Rule

Correction and prevention have fundamentally different time requirements. Correction (fixing an error after it occurs) can work with hours of warning — you detect the error and fix it promptly. Prevention (stopping an error before it manifests) requires days or weeks of warning because the causal chain that produces the error takes time to build, and intervening early in that chain requires knowing it's building before the outcome is visible.

A deadline miss doesn't happen the day before the deadline — it happens when scope wasn't controlled in week 1, when ambiguity wasn't resolved in week 2, and when the resulting delays compounded through week 3. By the day before the deadline, the miss is inevitable — no amount of effort can compress three weeks of accumulated slippage. The leading indicator that appears days before the miss (scope expanding without timeline adjustment) provides the intervention window. The lagging indicator that appears hours before (realizing you can't finish) does not.

This extends Substitute a faster noisy signal for a slower precise one… (faster correlated signals) to error prevention specifically: the leading indicator must appear far enough in advance that preventive action can actually work.

When This Fires

• When designing early warning systems for recurring errors
• When you keep catching errors too late to prevent their consequences
• When post-mortems reveal that warning signs were visible days before the error but weren't acted on
• When selecting which leading indicators to monitor (Substitute a faster noisy signal for a slower precise one…) for error prevention specifically

Common Failure Mode

Monitoring same-day indicators: "I check my progress at end-of-day to see if I'm on track." End-of-day is hours before tomorrow, which is only enough time for same-day correction (work later, work faster). The indicator that predicts whether this week's goal will be met should be visible by Monday or Tuesday — not Friday afternoon. By Friday afternoon, the week is essentially over and prevention is impossible.

The Protocol

(1) For each recurring error, trace the causal chain backward: what conditions, days or weeks before the error manifests, predictably lead to it? (2) Identify the earliest visible signal in that chain. "Scope creeping without pushback by day 3" predicts "deadline miss by day 14." "Sleep below 6 hours for 3 consecutive nights" predicts "cognitive performance crash by end of week." (3) Design monitoring for that early signal, not the late one. Check scope control on day 3, not deadline proximity on day 13. Check sleep quality nightly, not energy level when the crash has already started. (4) Define the preventive action that the early signal triggers: "If scope has expanded by day 3, renegotiate deadline or cut scope before more work builds on the expanded scope." (5) Verify the lead time: does the indicator appear with enough advance warning for the preventive action to actually work?