When correction exceeds 20% of weekly capacity, shift investment from faster fixes to upstream prevention

Why This Is a Rule

Below 20% correction overhead, errors are manageable noise — they consume some resources but don't significantly constrain productive output. Above 20%, correction has become a structural tax on the system. One-fifth of capacity diverted to fixing errors means the system is spending more time maintaining itself than it should, and the correction load is likely growing as the error-generating conditions remain unaddressed.

The strategic shift from correction to prevention is counterintuitive because correction feels productive — you're fixing problems, producing visible results, maintaining output. Prevention feels like overhead — you're investing time in changes that won't produce immediate visible output. But prevention reduces the error generation rate, which compounds: a 50% reduction in error rate produces 50% fewer corrections every week, indefinitely. Faster correction only reduces the time per correction instance while leaving the generation rate unchanged.

The 20% threshold is the inflection point where the correction tax is large enough to justify prevention investment. Below 20%, individual corrections are the efficient response. Above 20%, the volume of corrections signals a systemic problem that individual fixes can't resolve — you need upstream intervention that reduces how many errors are generated in the first place.

When This Fires

• When tracking correction time (Multiply direct correction time by 3x for true cost — context-switching, opportunity cost, and verification overhead are invisible) reveals that correction exceeds 20% of domain capacity
• When "firefighting" has become a significant part of your work routine
• When process improvement feels less urgent than the corrections demanding attention right now
• When the same error types keep recurring despite effective correction procedures

Common Failure Mode

Investing in faster correction instead of prevention: "Let's build a tool to fix this error faster." If the error occurs 10 times a week, reducing fix time from 30 to 15 minutes saves 2.5 hours weekly. Preventing 7 of the 10 occurrences saves 3.5 hours weekly (at the original fix time) and the savings grow if the prevention improves further. At >20% correction load, prevention ROI exceeds correction-speed ROI in almost all cases.

The Protocol

(1) Track correction time for one month using Multiply direct correction time by 3x for true cost — context-switching, opportunity cost, and verification overhead are invisible's 3x multiplier. Calculate: what percentage of weekly capacity does correction consume? (2) If <20% → continue with correction-based approaches. Optimize individual correction procedures. (3) If ≥20% → shift investment to prevention. For the top 3 error types by correction time: identify the upstream cause (Classify errors as execution, knowledge, or judgment failures before correcting — each type needs a fundamentally different fix, A true root cause, eliminated, makes the error impossible — if it only reduces frequency, keep digging) and design structural prevention (Recurring errors with the same root cause need structural fixes, not more effort — process changes beat discipline every time). (4) The prevention investment will temporarily increase total time spent (correction continues while prevention is being built). Accept this short-term cost — the long-term reduction in error generation rate will more than compensate. (5) After prevention deployment, re-measure correction percentage. If still ≥20% → the prevention didn't address the right causes. Re-diagnose.