ESSENTIAL PATH Stop 17 / 18

Community building October 27, 2021 Updated: September 11, 2026 15 min read THY-REALITY

Transition Without Collapse: Building Parallel Institutions

How can new local capabilities be built without destroying existing ones first? This article develops a function-by-function transition: pilot, parallel running, thresholds, redundancy, fallback and only then reduced dependence on the old route.

“From Five People to the First Local Core” showed that a small group can complete one concrete project without a permanent ruler. But one successful project does not yet create an institution that can safely carry an important function. this article therefore does not ask how to destroy the existing system; it asks how to build another option well enough that part of a function can one day be transferred safely.

Here, “parallel” does not mean a secret state, an illegal order or a bid for total power. It means developing an additional capability beside the existing path: another mutual-aid network, another form of supply, a cooperative, a local service, a backup communication route or another clearly bounded way of performing a function. The old path remains available until the new one has demonstrated sufficient reliability.

That is the core of transition without collapse: capacity first, evidence second, transfer third. Critical functions must not become experiments on people. A transition needs bounded scope, success criteria, a fallback route, reversibility and an explicit point at which the pilot is expanded, modified or stopped.

A parallel institution is not a second miniature state

An institution can be small: a recurring agreement, a group of people, rules, responsibilities and resources that perform a function with sufficient reliability. A local mutual-aid fund, cooperative, community workshop, alternative supply route or local energy group can be an institution without becoming a new political centre.

This article therefore rejects the idea that an entire “alternative system” should be built at once. Such a project would quickly reproduce the concentration it is meant to reduce. Ostrom’s work matters here because complex problems are often handled by different, connected institutions at multiple levels, rather than by one universal structure.

Parallelism is therefore primarily plurality of capability. If the new solution works only by controlling every other function, it is no longer a fallback; it is becoming a new monopoly.

The safest unit of transition is not an ideology but a function. What exactly are we trying to improve or duplicate? Food access? Transport for older residents? Local repair capacity? Mutual aid? Energy for a community space? Communication during outages? Each function has different risks, legal boundaries and competence requirements.

This prevents the dangerous leap from criticism to total replacement. If a new group runs a community tool library well, it does not follow that it can run a health service, adjudication system or security function. Demonstrated capability is always bounded by what was actually tested.

the articles from “Mutual Aid, Insurance, and Social Security from the Bottom Up” through “How Do We Measure Whether Decentralisation Is Real?” therefore operate as a catalogue of functions and safeguards, not as a package one organisation should absorb in full. This article connects them only as far as needed for cautious transfer of a specific task.

Build first; reduce the old dependency only afterwards

The most dangerous strategy would be: cut off the old service first, then force people onto the new one. That is not a free transition; it creates a vacuum. For critical functions it is more responsible to keep old and new routes available in parallel until the new one reaches an agreed minimum.

Large digital migrations use a similar logic. UK government guidance recommends phased migration, temporary bridges between legacy and new systems and, where necessary, dual running so that defects are discovered before they become service failure.

The same principle applies to communities: dependency should be reduced on evidence, not on a calendar or enthusiasm. If the new route fails, rollback must remain possible without catastrophic consequences.

A pilot is not a miniature publicity campaign for a solution already chosen. It is a genuine test of an assumption. European institutions use regulatory sandboxes, test beds and living labs around the same basic idea: bounded experimentation in defined legal conditions so that evidence can be gathered before wider deployment.

UNDP similarly stresses that scaling is itself another experiment. A solution that works for twenty people may change at two hundred because coordination, cost, diversity of needs or power relations change. A pilot must therefore ask not only “does it work?” but for whom, under what conditions, at what cost and with what side effects?

Good early pilots are bounded in time, users, financial exposure and consequence. Higher-risk functions require a higher evidence threshold than low-risk ones.

Parallel running exposes the difference between promise and capacity

A solution that exists only on paper can promise almost anything. Parallel running allows direct comparison: response time, cost, access, errors, user experience, resilience and repair capacity. It turns an ideological dispute into an empirical question.

The new institution must not receive the privilege of being judged by its intentions while the old one is judged by its failures. Both routes should be measured against comparable functional criteria. If the new route is more decentralised but far less reliable or excludes vulnerable users, decentralisation alone is not enough.

The most useful discipline is to define a minimum service level in advance: what must be true before any transfer can even be considered.

Two organisations do not necessarily provide two independent routes. If both depend on the same supplier, bank account, server, warehouse, vehicle or person holding every key, they can fail together. UNDRR highlights exactly these interdependencies in critical infrastructure, where one failure can propagate across systems.

This article therefore asks whether redundancy is genuinely heterogeneous: another storage site, another communication channel, another funding source, another team, an alternative procedure or at least a manual mode. NIST contingency planning likewise uses alternate equipment, sites and even manual processing as distinct recovery options.

The goal is not to duplicate everything forever. It is to avoid removing one single point of failure only to recreate it under another name.

Bridges between old and new are part of transition, not a betrayal of principle

People often imagine a clean break: the old system on one side, the new one on the other. Real transitions move data, users, contracts, equipment and expertise gradually. Temporary bridges can therefore be a sign of responsibility rather than compromise.

A bridge can be a shared standard, temporary contract, data import, old and new payment channels, a joint on-call arrangement or an agreement that a larger system takes only cases beyond local capacity. the articles from “Subsidiarity in Practice: Which Level Should Decide?” through “When is larger scale genuinely necessary?” already established that subsidiarity does not mean isolation.

The bridge must not become a permanent hidden dependency. Every bridge therefore needs a purpose, owner, review date and plan for what happens if the connection fails.

Transfer only after the new capability passes predefined thresholds

Enthusiasm is not a transfer criterion. Before reducing the old route we need evidence of capability. The threshold may be modest for a simple local service; it must be much higher for a function affecting health, safety or essential access.

Thresholds should include at least sufficient capacity, stable funding, trained people, documented procedures, appeal, data protection, access for vulnerable users, fallback, error measurement and a successful period of real operation. WHO continuity planning similarly focuses on maintaining essential services through disruption, not merely possessing a plan.

If the threshold is not met, transfer is delayed. A failed pilot is cheaper than a failed institution when it was designed to generate learning without large harm.

Legality, safety and rights are not obstacles to alternatives

A parallel institution does not receive a moral exemption merely because it is local, voluntary or new. Health, finance, personal data, safety, construction, labour and other regulated domains have legitimate minimum standards, accountability requirements and protections for third parties.

This article therefore does not propose a secret legal order or evasion of rules that protect life, property and fundamental rights. Bad rules can be criticised and changed, but a pilot must not transfer the cost of its ideology to people who did not consent to it.

European experimentation frameworks are useful precisely as an analogy: innovation is tested inside defined legal boundaries, not in a zone without accountability.

The new institution needs safeguards against its own capture from the start

“Institutional Capture: When the System Starts Serving Itself” showed how institutions can begin serving themselves. Alternative organisations are not immune. When a pilot succeeds, money, prestige, jobs, information control and incentives to declare the organisation indispensable all appear.

Safeguards should therefore be present during the pilot: bounded mandate, transparent finances, separated roles, replaceable office-holders, public performance criteria, complaint mechanisms and the ability of users to leave or choose another route. “How Do We Measure Whether Decentralisation Is Real?” supplied the measures for checking whether decentralisation remains real as an organisation grows.

If a new institution needs monopoly, compulsory membership or concealed information in order to survive, the transition has gone in the wrong direction.

Once the new route has met its thresholds for long enough, dependence on the old one can be reduced in stages: first a small share of users or tasks, then a wider scope. At every step we watch for hidden dependencies, queues, staffing shortages or problems the pilot did not expose.

UK guidance on digital migration recommends phased introduction and dual running for the same reason: abrupt cutovers are risky in complex systems. The community-level rule is identical: until the new route has carried real load, the old one should not be removed irreversibly.

When the old route is genuinely no longer needed, it can be reduced or retired. Even then, documentation, data, knowledge or contractual options should be preserved where they make recovery possible if a serious defect later appears.

Twelve questions for a transition without collapse

Before transferring any function, complete a short transition sheet: (1) what function is moving, (2) who provides it now, (3) what is the alternative pilot, (4) who is accountable for it, (5) what minimum service level must be reached, (6) what legal and safety boundaries apply, (7) what dependencies the new route still has, (8) how long parallel running lasts, (9) how success and failure are measured, (10) what triggers expansion, (11) what triggers rollback and (12) when the old route can safely be reduced.

Without these answers we do not yet have a transition strategy; we mostly have a desire for change. The first this article action is therefore not to found five new institutions, but to choose one function from the “Local Resource Map: What Do We Already Have Around Us?” map, one core from “From Five People to the First Local Core” and one bounded pilot that can run for 30–90 days beside the existing route.

Stage 10 is now close to completion. “Reality, Freedom and Responsibility: The THY-REALITY 90-Day Plan” will turn the whole THY-REALITY curriculum into one 90-day plan with responsibilities, deadline, measurements and retrospective. The project will not end with a promise of a new world, but with one verifiable step in the real one.