Part three. The Checker
The fourth condition
The checker must still be able to do the work themselves. Otherwise they are not checking, they are agreeing.
Read the transcript
Three conditions came earlier: repeatable, no customisation, no thought. The fourth is about who verifies the result, and it is the one this volume is named for. Whoever checks the output must still be capable of producing that output themselves. A person who cannot do the work can confirm that output exists, that it is formatted correctly, and that it looks like output they have seen before. They cannot detect that it is wrong in a way that looks right, which is the only failure mode that matters. This constrains the benefit, and that is uncomfortable. The practical form is a sampling regime performed by someone competent. Not the person who configured the system, for the same reason you do not mark your own work. And competence decays if unused.
Three conditions were in part one: repeatable, no customisation, no thought. The fourth is about who verifies the result, and it is the one this whole volume is named for.
Whoever checks the output must still be capable of producing that output themselves.
A person who cannot do the work can confirm that output exists, that it is formatted correctly, and that it looks like the output they have seen before. They cannot detect that it is wrong in a way that looks right, which is the only failure mode that matters.
This is uncomfortable because it constrains the benefit. The saving from automation is usually calculated as removing the skill from the operation entirely, and the fourth condition says you must retain enough of it to check.
The practical form is a sampling regime performed by someone competent: a fixed percentage of output, verified independently against the standard, by a person who could have produced it. Not the person who configured the system, for the same reason you do not mark your own work.
And competence decays if unused. A checker who has not done the work in two years is a checker in name. Keeping a small share of the work manual is not inefficiency, it is the maintenance cost of being able to detect failure.
Two things worth doing, neither of which costs anything
Read the standard in full, free and in the open, before you pay anything or give anyone an address.