The first Saved Hours question was deliberately small:
If a machine saves labour time, who receives the time?
It was not a forecast about mass unemployment, and it was not a promise that automation would liberate everyone. It was an accounting question placed before the saved work disappeared into a new normal. A tool can remove ten hours from a process. Those hours can become shorter working time, better care, lower prices, public capacity, repair, profit, denser targets, fewer jobs, more tickets, more monitoring or more output.
They can also disappear without being named.
That was the point of The Saved Hours Doctrine: a saved hour is not returned to a person merely because a technical system made it unnecessary. The hour has to pass through contracts, management, budgets, bargaining power, public policy and measurement. In that passage, the hour can change owners.
The second question is physical.
If the saved hour is actually returned, what makes it usable?
A person cannot inhabit returned time in the abstract. Returned time may need heat, light, cooling, connection, transport, tools, public buildings, reliable services, local production, networks and compute. It may need electricity at the right time, in the right place, under conditions ordinary people can afford and command.
That is the argument behind Post-Work Requires Power.
The new working paper does not replace The Saved Hours Doctrine. It extends it. The doctrine asks who receives the time and value created when labour is made technically unnecessary for a defined output. Post-Work Requires Power asks what electricity and wider energy-system access are required if some of that time is genuinely returned to human activity.
Automation is often imagined as both a labour saver and a source of abundance, while electricity planning usually counts only the machine side.
When a data centre, semiconductor plant or automated system needs electricity, that demand can be named as infrastructure. It appears as future load, industrial requirement, national capacity, strategic investment or energy security. When a person uses electricity during time that work no longer claims, the demand is easier to describe as discretionary consumption, rebound or leakage.
That may be an accounting convention. It is not a neutral moral fact.
If machines are built in the name of freeing human time, the life inside that freed time should not become the residual claimant on the system.
This is where the new paper connects to The Missing Mechanism. A warning is not an allocation mechanism. A principle is not a labour inspector. A public statement is not a budget. Measurement is not yet protection. The same is true for electricity. It is not enough to say that future systems will be efficient, that automation will save work, or that growth, markets and planners will handle the rest.
The mechanism still has to be named.
Who forecasts the returned human side of demand? Who decides whether new capacity goes first to factories, platforms, data-centres, public facilities, homes, local production, personal compute or care? Who can challenge the allocation? Who receives access in practice? And what happens when the system has already been built before the human demand was allowed to appear?
Japan makes the question inspectable.
Post-Work Requires Power uses Japan as its first field edition because the official FY2026 to FY2035 electricity forecast explicitly names historically novel demand from data-centres and semiconductor factories. That does not make the forecast morally wrong. It does not show that planners are choosing machines over people. It shows something narrower and more useful: novel machine-side demand can be written into the future before it exists at scale.
If that can be done for industrial and compute loads, then possible human-side demand from reduced compulsory work can also be modelled explicitly.
The paper does not claim to discover time-use rebound. Literature already exists on saved time, shorter work, spare-time rebound, energy sufficiency and activity substitution. Its contribution is the planning boundary: automation-added energy, changed human energy from time actually returned, and genuinely avoided work-system energy have to be placed inside the same annual accounting frame.
The last term is important. A shorter workday does not automatically retire a building. It does not automatically remove a train path, shut down equipment, reduce floor area, change peak capacity or decommission an old process. Energy is avoided only where the old work-system is actually retired, rescheduled, downsized, removed or materially reorganised.
A saved hour is not an energy saving by itself. It is a social possibility with a material surface.
That is why The Afternoon Was Not in the Forecast worked first as fiction. It made the problem feel ordinary: a returned afternoon, a child repairing an arcade cabinet, and a power system whose priorities had been written before the returned afternoon had a category. The fiction was not proof. It was a doorway. The working paper is the slower version of the same question.
There is a bad version of this argument. It would say that every person should use more power because work has been reduced. That is not the claim. Some demand is wasteful. Some demand is extractive. Some demand should lose in a fair allocation. Ecology, land, minerals, water, grid constraints and affordability remain real.
The better version is simpler:
Current demand is not a moral ceiling.
Observed demand is shaped by prices, income, housing, climate, schedules, exhaustion, infrastructure, appliance ownership, public services, social norms and access to tools. It shows what people do under present constraints. It does not reveal the complete activity set of a society organised around less compulsory work.
The old question was: who gets the saved hours?
The new question is: who gets the new electricity?
Those are not separate questions for an automated society. They are the same allocation problem seen through two ledgers. One ledger records time. The other records material access. If either ledger is missing, the promise can be captured before it reaches the person it was supposedly built to free.
The saved hour does not end when work disappears.
It begins when a person can actually use it.
For that, the saved hour also needs power.
