Why adding a review gate buys less than it costs
Sorting finished work at the end does not change what arrives at the end, and a gate is sorting.
The second reason, and it is older than this problem by decades. A review gate sorts what arrives at it. It does not change what arrives at it.
Inspection at the end of a line sorts what arrives; it does not change what arrivesDeming, 1982unverified. That argument was made and won in manufacturing long before anybody had this problem, and it transfers cleanly: you cannot sort your way to a stream that needs less sorting.
It is worth being careful about what is and is not being claimed. Gates are useful. Catching a bad thing before it ships is better than not catching it. The claim is narrower: a gate moves work between two piles at the end, and the quantity this volume is about — how much gets produced without anybody judging it — is decided upstream of the gate and is not affected by it.
| what a gate does | what it does not do |
|---|---|
| catches some of what is wrong before it ships | reduce how much wrong work is produced |
| gives you a count of what it caught | tell you about what it did not catch, which is the remainder |
| makes the work that passes it more trustworthy | make the work that bypasses it — because the gate has finite capacity — any more trustworthy |
And a gate with finite capacity creates a bypass. That is the part people miss. When the gate cannot keep up, what actually happens is not that everything waits — it is that some things go around, chosen by whatever pressure is highest that week. So the gate's own capacity limit decides which work goes unexamined, and nobody designed that rule.