A Field Test Designed to Fail Honestly

Part 10 of 10 in the 10,000 People Economy series.

The 10,000 People Economy is ready to be questioned in the field. It is not ready to be declared proven, and the distinction matters.

Development pilots are good at producing visible activity. People attend training, equipment arrives, organisations are registered, funds are disbursed and photographs are taken.

None of those facts proves that a customer bought the output, the cash arrived, wages were sustained or the work was additional.

A serious pilot must be designed so that the model can fail.

Start With the Causal Chain

The model's proposition is not simply that support creates jobs.

It is:

verified demand leads to executable production; production receives the required assets, inputs and working capital; people perform paid work; capability improves; customers pay; and the resulting income and production persist.

Every arrow can break.

A buyer may express interest without ordering. An enterprise may invoice without collecting. Equipment may arrive without electricity. A learner may complete training without performing the occupation. A new firm may take an existing contract and create no net work.

The field test must measure each break.

Do Not Begin With 6,460 Jobs

The model calculates 6,460 provisional placements from a 10,000-resident community and a planning participation parameter of 64.6 per cent.

That number is a design allocation. It should not become a promise to a selected community.

The first field phase should ask what demand, production, infrastructure, skills and physical resources actually exist. Only then can the model calculate how many roles are demand-backed, revenue-supported and affordable in cash.

The accepted placement count should remain zero until those gates pass.

Test Mechanisms Before Testing the Whole System

The model combines many ideas: buyer-backed production, paid apprenticeships, community finance, local value chains, external sales and shared federation services.

Launching all of them at once creates a dramatic pilot and weak evidence. If the result improves, we do not know why. If it fails, every component can blame another.

Start with bounded tests:

• one buyer-backed product pathway;
• one paid apprenticeship contract;
• one cash-funded working-capital instrument;
• one shared maintenance or market service.

Each test should have a defined mechanism, delivery standard and outcome.

Measure Outcomes That Resist Inflation

The primary outcome should not be enrolment, registration or planned employment.

It should be paid, productive participation sustained for at least twelve months.

A counted role should have:

• a named person;
• verified work performed;
• payment under the approved rule;
• output linked to accepted demand or an explicit public-purpose mandate; and
• evidence that the enterprise can continue paying.

Other primary measures should include real earned income, sales collected in cash, wage-payment reliability and net new work after displacement.

Invoice value is not cash. A person trained is not necessarily a person employed. A supported job is not necessarily an additional job.

Look for Harm

A model can produce visible output while making some participants worse off.

The test should track household debt, unpaid family labour, delayed wages, unsafe work, business displacement, elite capture, environmental pressure, community conflict and dependency on one customer.

Results should be reported by gender, age, disability, location and prior labour-market status. An average gain can hide unequal access or concentrated losses.

Community participation in governance is essential. It does not replace individual consent or independent protection of participant data.

Use Comparison Where It Is Credible

If several eligible sites exist and rollout must be staggered, the order can create a comparison between sites receiving the intervention earlier and later. Where that is not appropriate, matched communities and strong baseline data can support a more cautious comparison.

The design should fit the number of sites and the realities of implementation. A complex statistical method cannot rescue two selected communities and a weak baseline.

Qualitative research should explain how institutions operated and why people responded. It should sit beside outcome measurement, not replace it.

Decide the Failure Rules in Advance

A pilot becomes unfalsifiable when every result is interpreted as progress.

The test should pause for wage arrears, unsafe conditions, serious data failures, lost legal authority, a critical physical shortage or liquidity below the protected minimum.

The model should be rejected or narrowed when:

• demand-backed roles remain far below the predeclared threshold;
• enterprises cannot convert revenue into enough cash to operate;
• apparent job gains disappear after displacement;
• physical resources cannot support production;
• shared federation services increase cost or risk; or
• governance concentrates benefits in a small group.

Changing the model after evidence is legitimate. Changing the success definition after evidence is not.

Publish the Null Result

An independent evaluation function should hold the preregistered protocol and reproduce the primary analysis. The implementer should not be able to reclassify outcomes alone.

The intervention version, data definitions, deviations and all preregistered outcomes should be published. Personal and commercial data must remain protected, but methodological transparency is non-negotiable.

If the pilot stops early, publish what it learned. Perhaps buyers would not commit. Perhaps working capital was too large. Perhaps the training pathway lacked authorised assessment. Perhaps the physical account exposed a binding constraint.

Each of those findings would be valuable because it narrows the conditions under which the model can work.

The purpose of the field test is not to defend the 10,000 People Economy. It is to discover the conditions under which the model produces sustained, additional and defensible outcomes-and the conditions under which it does not.

That is how an ambitious idea earns the right to scale.

Further reading

• Ostrom, E. (2010). Polycentric systems for coping with collective action and global environmental change. Global Environmental Change, 20(4), 550-557.
• Tcherneva, P. R. (2018). The Job Guarantee: Design, Jobs, and Implementation.
• Todes, A. and Turok, I. (2018). Spatial inequalities and policies in South Africa. Progress in Planning, 123, 1-33.
• United Nations et al. (2009). System of National Accounts 2008.

---

Previous: Accounting for a Community Economy
Series: Return to the 10,000 People Economy series