Working spec · v0.2
The Human Ledger is a public reporting standard for any organisation using AI on people’s work. It asks them to publish the conditions their people work under, year after year, in numbers anyone can check, and it never grades whether the work is any good.
The essay is the full argument: Part I on how we came to count the wrong things, Part II on the last job left when AI does the rest.
“It measures everything, in short, except that which makes life worthwhile.”
Robert F. Kennedy · University of Kansas · 18 March 1968
He was talking about GDP. Nearly sixty years later, the way organisations measure AI has the same blind spot: it counts the output and the savings, and has no category at all for the people doing the work.
What it is
A disclosure standard, from the same family as working-time rules and health and safety reporting. Organisations already publish output, adoption and headcount. This adds what those figures leave out: how much time people have, whether anyone is still being trained and hired, who did the work, what had to be redone, and who got hurt.
How it works
Three building blocks, borrowed from the way carbon reporting is built. Five principles set the rules. Six conditions are what every organisation discloses, identical in every sector. Sector annexes show what those conditions look like in one industry’s own numbers.
Why we need it
What gets counted gets protected, and right now AI’s ledger counts only the gains. The costs land on people (less time, fewer junior roles, sign-offs nobody has time to check) and stay invisible until they are too big to fix. Insurers and rating agencies already measure some of this, and keep the results to themselves.
How it eases the disruption
Disruption hurts most when its costs show up late. A graduate intake that quietly stopped, or a review step that became a rubber stamp, is cheap to fix in year one and very expensive in year five. Published every year, the same six numbers give workers, boards, investors and regulators an early warning, and minimum floors stop the race to the bottom before it starts.
Start here · the full thesis
The standard is the short version. The essay is the whole argument: why we keep counting the wrong things, what AI does to that mistake, and why the last job left may be the person who signs off.
There are no comments on this site, on purpose. Every argument about the Ledger, from a principle you’d kill to an annex you could write, happens in the LinkedIn group, where the people who would have to file these numbers already are.
The list
Get new principles, revisions and the arguments that killed them, roughly once a month and never more often.