When Recipients Become Providers: The Scaling Logic Behind ARC

TL;DR for operators

How do you scale a high-touch expert-support service when each central team can serve only so many locations and only within limited geographic reach? ARC tests an architecture in which central organizations seed nearby programs, while sufficiently mature supported sites can later become local providers themselves.¹

In the paper’s sixth of twelve ordered assumption settings, central college-based hubs alone produce about 323 school programs after 40 years. Keeping the initial central hub while allowing supported schools to become local hubs raises that to about 551. Full ARC, which combines expansion of college hubs with recursive school-hub formation, reaches about 992. These are conditional simulation outputs, not forecasts. :chatgpt-content-reference{index=“0”}

The operational distinction is architectural: adding central delivery capacity helps, but the model’s largest expansion appears when successful recipients can become durable providers that support additional nearby sites. Sensitivity analysis points in the same direction, with supported secondary-hub formation, support capacity, and service radius among the strongest tested long-run drivers.

The real deployment evidence is much narrower: one Indiana university prepared undergraduates to support three newly formed rural FIRST LEGO League teams, with favorable but small, retrospective, uncontrolled survey results. The practical question is therefore not whether the simulation predicts statewide adoption, but whether a scaling strategy invests only in more central capacity or also in a credible pathway for successful local recipients to become providers.

Central capacity alone does not generate the model’s largest expansion

A geographically dispersed support service has two constraints before demand even enters the picture. Experts can only serve so many recipients, and they can only reach so far.

The straightforward response is to add more central providers. ARC includes that route: colleges with relevant technical capacity can become what the paper calls primary hubs, training undergraduate mentors and supporting nearby K–12 robotics programs.

The simulation suggests that this helps, but it does not reproduce the larger expansion of the complete architecture.

At the paper’s sixth of twelve ordered assumption settings, the primary-hubs-only scenario produces about 323 programs after 40 years. A scenario that retains the initial primary hub while allowing schools to become secondary hubs produces about 551. Full ARC produces about 992.

The twelve levels should not be read as estimated probabilities running from pessimistic to likely to optimistic. They deliberately vary ARC-specific assumptions whose real values are not yet known. Natural FIRST program dynamics remain fixed while assumptions about ARC activation, support, maturation, hub formation, persistence, capacity, and related transitions become progressively more favorable.

The comparison therefore answers a structural question: given the assumed transition rates, which pieces of the architecture change the growth trajectory?

The answer is that distributed seeding and recursive local propagation perform different jobs.

Secondary hubs change who is allowed to create the next node

ARC’s distinctive mechanism begins after a supported school becomes mature enough to operate more independently.

Some mature schools can become secondary hubs. Instead of remaining permanent consumers of university support, they can mentor other nearby schools. Those newly supported schools can mature in turn, and some can later become secondary hubs themselves.

This changes the system from a collection of central service territories into a recursively expanding network.

The spatial Markov model formalizes that process at the school level. Each year, schools can move among program states, become reachable by active hubs, request or receive support, mature, close, backslide, or—under specified conditions—become secondary hubs. College hubs also activate or cease operation.

That recursive structure explains why favorable scenarios can show accelerating and sometimes super-linear growth before slowing as Indiana’s finite school population becomes saturated. The paper has not observed that statewide growth pattern in practice. It emerges from the simulated mechanism when hub formation and persistence are sufficiently favorable.

This distinction matters because a secondary hub is not simply another unit of central capacity. It is a change in the system’s reproduction rule: a successful endpoint can become a new origin for service delivery.

The ablations tell us what each scaling mechanism contributes

The paper’s scenario comparisons work as ablations rather than as real-world treatment comparisons.

Full ARC also reaches about 1,415 of Indiana’s 1,925 modeled public K–12 schools after 40 years at this setting. Reach does not mean that all of those schools receive direct support or have programs. It means at least one active hub lies within the model’s service relation.

That separation between reachable, supported, and active program is operationally valuable. Organizations often collapse all three into a single growth metric. ARC instead treats them as different states with different constraints.

A regional operation can have nominal coverage while lacking delivery capacity. It can provide support without achieving recipient independence. And it can generate successful local programs without converting any of them into new service nodes.

Those distinctions are more informative for capacity planning than a top-line participation count.

Sensitivity analysis identifies what needs real measurement next

Because many ARC-specific transition probabilities cannot yet be estimated from field data, the sensitivity analysis is more consequential than the large year-40 totals.

Across the tested ranges, the biggest long-run change comes from the probability that a supported mature school forms a secondary hub. Varying that factor changes the projected year-40 program count by roughly 430 schools. Direct-support capacity changes it by about 338, while reach radius changes it by about 302.

For operators adapting this architecture elsewhere, those quantities translate into measurable questions:

Cognaptus inference: an organization testing a distributed-service model should instrument these transitions early rather than waiting for aggregate growth numbers. If local-provider conversion is the mechanism expected to generate nonlinear expansion, then conversion rate, time-to-conversion, retention, and downstream service capacity belong in the operating dashboard from the pilot stage.

That inference extends beyond education; the numerical results do not.

The pilot establishes feasibility, not the long-run scaling mechanism

ARC also includes a real deployment at one Indiana university. Nine undergraduates enrolled in a 16-week, three-credit course combining mentor preparation, robotics and AI topics, and recurring work with three new rural FIRST LEGO League teams.

Seven undergraduates completed the retrospective survey. Mean ratings increased from 3.00 to 4.29 for confidence teaching technical concepts, from 3.29 to 4.43 for adapting explanations, and from 1.86 to 4.00 for connection to the surrounding community.

Four parents provided proxy ratings for participating children. Mean robot-programming knowledge increased from 2.00 to 4.00, access to programming resources from 2.00 to 4.25, and programming or problem-solving practice from 3.00 to 4.25.

These results are descriptive. The surveys were collected retrospectively after the program, there was no randomized or matched comparison group, the respondent counts were small, and children were not surveyed directly.

What the trial demonstrates more credibly is operational feasibility: a university could assemble the course, prepare undergraduate mentors, run recurring workshops, and support three newly created rural teams.

It does not yet test the system’s defining long-run mechanism. The study has not followed multiple cohorts long enough to establish how often supported schools mature into durable secondary hubs, how much support those hubs can provide, or whether the recursive expansion seen in the model appears across regions.

The scale decision is architectural before it is predictive

ARC offers a useful way to think about distributed expert services because it separates several decisions that are often bundled together.

Adding central hubs determines where expertise originates. Increasing capacity determines how many recipients each hub can handle. Extending reach determines which recipients can plausibly be served. Building a secondary-hub pathway determines whether recipients can eventually expand the provider network themselves.

The paper’s simulation says those choices can generate very different growth paths under the specified assumptions. Its pilot says the first university-to-school link can be implemented. Neither result establishes how many programs ARC would actually create across Indiana over four decades.

For an operator, that is enough to change the pilot question. Instead of asking only whether a central team can serve more locations, test whether the service can transfer enough capability and ownership that some locations eventually cease to be endpoints.

A distributed program begins to scale differently when success creates another provider.

Cognaptus: Automate the Present, Incubate the Future.