Entities › How directed time is estimated

How directed time is estimated

Religious codes, laws, party rules and job descriptions are treated alike here: as instructions that direct some of their members’ waking time. The charts estimate that share and express it in neuron-equivalents. It's a share of time scaled by brain size, not a count of neurons in use. A brain that's following an instruction is still doing plenty else.

The estimate

Neuron-equivalents = members × 8.61 × 10¹⁰ × c. The neuron count is the measured human average. The share directed, c, is the fraction of a member’s waking hours that the organisation’s instructions decide. Each hour is weighted by how fully they decide it:

• Directed (1.0): the organisation sets the action. A shift at work, a prayer, a service, a lesson.
• Restricted (0.5): it rules out most options for a stretch of time. Examples are the Sabbath, and dietary laws that settle most of a meal.
• Ruled (0.25): it rules out one kind of option. Examples are a rule against one kind of food, or a community code on dress, media and technology.

c = Σ (share of members who keep the practice × hours a week × weight) ÷ 104.79 waking hours. The week has 104.79 waking hours because adults spend 9.03 hours a day in bed; meals take 8.68 hours a week; the waking part of a 25-hour Sabbath is 15.59 hours. US Bureau of Labor Statistics, American Time Use Survey (2024 eating; 2025 sleep).

The weights are judgement calls, not measurements. A rule that removes one kind of option governs less of an hour than one that removes most of them, and less than an instruction that fills the hour. Shares come from surveys of what members say they do, not from what doctrine requires. A code over “all other hours” covers only what the listed practices leave, so no hour is counted twice.

What it leaves out

• Influence without instruction: belief, identity, and the choices a member makes in the organisation’s spirit. We only count behaviour that takes a measurable amount of time.
• Children. Survey shares are for adults and are applied to all members.
• Overlap. Someone who belongs to a religion and works for a company is counted in both. The charts are separate, so nothing is added up.
• Shareholders and customers. They're influenced, not directed.

Religions

* The time a practice takes is our estimate; the share of members who keep it is from the survey. The chart covers every religion and branch counted at a million adherents or more in a standard reference (Pew Research Center, the World Religion Database, the bodies’ own counts, national censuses). Where no survey of a branch’s practice exists, it takes the figures of the tradition it belongs to; where none exists for a whole religion, it takes daily prayer at the median share of the surveyed religions. Both are marked in the chart. Branches and movements overlap (a Pentecostal may also be a Baptist and an Evangelical), so they are ranked apart from whole religions and never summed. Counts before 2015 are marked as weak data.

States

A state is counted by the time it directs that we can measure in every country. That's its employees at work, its soldiers off duty under military law, its pupils in class, and its prisoners, since it directs all of their waking time. Government employment is assumed to include the armed forces, so soldiers’ duty hours (48.1 a week, US active duty) aren't counted twice. Where a country doesn't publish its instruction time, we use the average of the countries that do (898 hours a year).

General law isn't counted. Most of it forbids things people wouldn't do anyway. Where it does bind an activity, such as traffic law at the wheel, nobody measures that time across countries. Taxation takes the proceeds of work rather than directing it. So each state’s share is a minimum. Where a country publishes no hours for public administration, we use its average for all employees. If that's missing too, we use the average of the other countries (41.6 hours).

Every state is included: the 193 members of the United Nations, Palestine, Kosovo and Taiwan. We researched thirty-three one by one. The rest use international data series. Public employment is the median public share of employment across a country’s ILOSTAT surveys since 2010, applied to ILO modelled employment for 2025; armed forces are the World Bank’s series from the IISS (2020); pupils are UNESCO enrolment less the private share. Where a figure is missing, we estimate it from the median of the country’s region and mark it *. India’s public employment is the latest official count (2012) and excludes public companies; North Korea’s is all employment, since nearly all work is in state units. Left out: Vatican City, whose few hundred residents are almost all its own staff, so the measure does not apply.

Shares are of residents. * estimated from the regional median, or an estimate in the source. Notes on each state appear in the chart’s details.

Animal societies

We treat a typical group of a social species as an organisation. Its share directed is the time spent in coordinated group activity, with the same weights. Coordinated group action (a joint hunt, travel the group decides, colony-set work) is directed, 1.0. Time spent with the group but not in coordinated action is restricted, 0.5. Time alone counts for nothing. Neurons are the cortex or pallium, or the whole brain for insects. So this table isn't on the same scale as the organisation charts, which count whole human brains. Only the Bigg’s orca and the sperm whale have measured time budgets. Every other share is our estimate from the field studies. Species with no neuron count take the figure for the nearest relative that has one (dog for wolf and wild dog, banded mongoose for meerkat, chimpanzee for bonobo, striped for spotted hyena, Eurasian jay for pinyon jay). A Hadza hunter-gatherer camp is included for scale.

* estimated. Group sizes, time budgets and call repertoires: research/animal-groups-research.json in the site’s data, with the field studies each figure comes from.

Political parties

A party is counted by the time it sets (the meetings and study its rules require), or by the time its members say they give it. Ruling communist parties set meetings by rule, and we estimate how long they last. Surveys of members’ time exist for the British parties and, pooled, for Canada’s. Every other party takes the average of those surveys, about 2.4 hours a month. That's a stand-in, marked as such, for a member of a mass party with no attendance rule. US parties have no members, so they are counted by registered voters and the share of all adults who worked on a campaign (3.5%) or went to a rally (5.5%) in 2020. The chart covers every party that reports at least a million members, plus smaller parties in Germany, Britain, Canada, Brazil, Cuba and South Africa. Parties without a published membership (Nigeria’s PDP, Bangladesh’s Awami League and BNP, Ghana’s NPP and NDC, Kenya’s parties) are left out, as are those whose latest figure is more than ten years old: Aam Aadmi Party (India; latest figure 2014); Pakistan Tehreek-e-Insaf (Pakistan; latest figure 2012); MPLA (Angola; latest figure 2008).

Corporations

Paid hours are directed time, with weight 1.0. No company publishes its employees’ hours. So each company uses the average hours actually worked per employee in its industry, in the country where most of its workforce is. That average includes part-time workers. We don't count conduct rules outside work, or dormitory rules where workers live on site.

The chart covers every company with at least 300,000 employees. A company that belongs to a larger one in the list (PetroChina within CNPC) is counted only in the parent; staffing firms’ placed workers are not their employees and are not counted. Hours: ILOSTAT, mean weekly hours actually worked per employee by economic activity, latest year. China has no series by industry, so its companies use the National Bureau of Statistics mean for enterprise employees (2024). Where a company does not publish where its staff work, the country is inferred from its operations (†). Employees: company annual reports and the Fortune Global 500 (2025).

Data version 2026.10.02. Download (JSON).