Billions, Verifiably

How people take part in the Topsoil Commons

Every load bearing claim in this document is sourced from published research and production systems, cited in line, because a plan to involve billions that begins by inflating its own numbers has already failed.

Where we honestly stand

Today the Commons has one enrolled human: its founder. Every number in this document after that sentence is a rung on a ladder, a precedent from someone else's system, or a target with a gate. Nothing here is dressed as an achievement. The ladder only means anything because the counting rules below cannot be quietly loosened.

What involved means: five tiers, never blended

Involvement is counted in five tiers, published separately, always. Blending them into one big number is the oldest trick in platform reporting, and the Commons refuses it by construction.

• Reached: a person encountered the work and understood the invitation.
• First step: a person took one real action, a vote cast, a confusion reported, a classification made, sized in minutes.
• Contributor: a person has a verified entry in the public ledger.
• Recurring: a person contributed or voted in at least three separate months.
• Citizen: an enrolled member who votes in the Commons' governance.

The precedents justify the discipline. OpenStreetMap reports about 10.6 million registered accounts and roughly 40,000 to 50,000 monthly active mappers. Registered is not involved, and any system quoting the first number as participation is misleading its audience. iNaturalist, with more than 200 million observations behind more than 5,000 research papers, still sees its top 1 percent of accounts produce over 60 percent of observations. Concentration is a law of crowds, not a failure. The Commons designs for it honestly: the 1 percent are honored in the ledger, and the other 99 percent are counted in the tiers that describe what they actually did.

The road: institutions already hold the billions

No application reaches billions by being downloaded. Billions are reached through the institutions that already touch billions: schools, health systems, faith networks, employers, and public services. The Commons' road is the charter's road, adoption of basic tools by those institutions, plus tables anyone can open locally under the same constitution, with every local ledger public and the global count a sum that anyone can recompute.

The scale precedents are real. DHIS2, an open source health information platform under a BSD license, runs inside national health systems across more than 70 countries, not because patients chose an app but because ministries adopted a substrate. The humanitarian mapping surge after the 2015 Nepal earthquake put more than 6,000 volunteers onto OpenStreetMap, mapping 80 percent of the affected zones and over a million buildings in about a month (reported in Scientific Reports). The reach of the great membership networks dwarfs any platform's marketing: the Red Cross and Red Crescent movement counts more than 17 million volunteers, Rotary about 1.24 million members, Lions about 1.4 million. The Commons does not compete with these. It invites them, through the registry described in the companion document, and offers them basic tools worth adopting.

The loop that brings people back

Every basic tool carries the same participation loop, each step small enough to take today:

1. Use something genuinely useful, free, in your own language, on the cheapest phone, offline where the network demands it.
2. Take a first step sized in minutes: answer, classify, report, vote.
3. Receive a receipt that shows what your step did: the record it corrected, the count it joined, the decision it moved.
4. See the credit land in the public ledger, in proportion to what was verified.
5. Invite one named person, with the receipt as the message, because practical value forwarded to a specific person is how participation actually spreads.

Citizen science proves the loop at scale. Galaxy Zoo volunteers produced about 125 million classifications and dozens of papers. Foldit players solved the structure of a retroviral protease in about three weeks in 2011, a problem that had resisted scientists for years. Safecast volunteers built a radiation dataset of more than 120 million observations, published as open data and validated against official United States agency measurements, which is why governments could use it. Small steps, visibly aggregated, honestly validated: that is the entire trick, executed for years without cheating.

Decisions with consequences, or participation dies

The single clearest warning in the research is France. Its Citizens Convention for Climate seated 150 people by lot, produced serious proposals, and then saw only about 40 percent of them reach the eventual bill in recognizable form. Participation decoupled from power teaches people that participation is theater, and they do not come back. Taiwan teaches the opposite lesson: its Join platform, with more than 10 million unique visitors since 2015, binds the government to respond point by point when a proposal crosses 5,000 signatures. The response is owed, in writing, by a named ministry. Estonia shows the end state of trusted infrastructure, with 51.1 percent of all votes in its 2023 parliamentary election cast online.

So the Commons binds itself in advance. Every vote states, before it opens, exactly what it controls and when the result takes effect: roadmap priorities, featured work, phase gate reviews, and from the first revenue, a participatory budget. Results are published with a simple line that cannot be fudged: what changed because of this vote. A vote that changes nothing is never held.

Participatory budgeting: how votes become real change

When the question is whether mass participation moves real resources to real needs, one mechanism carries the strongest evidence on earth. Participatory budgeting began in Porto Alegre, Brazil, drawing more than 40,000 participants a year at its peak, and the measured results were physical: household sewer and water connections rose from about 75 percent in 1988 to about 98 percent in 1997, and studies across Brazilian municipalities found budget shares for health and sanitation rising measurably where budgeting opened to citizens (reported in World Development). Paris committed 500 million euros through its participatory budget between 2014 and 2020, with 127,880 voters in 2018. New York City's process put 38 million dollars to 67,000 voters in one cycle. Barcelona's Decidim based process counted 48,796 voters choosing 76 projects in 2025. Ireland shows the deliberative end of the same road: an assembly of 99 citizens chosen by lot led to the 2018 referendum that repealed the Eighth Amendment with 66.4 percent of the national vote, and Belgium's Ostbelgien made its lot drawn council permanent in 2019, with parliament required to respond.

The Commons uses participatory budgeting as its main way of turning votes into change, starting with its own first dollars and publishing every allocation. When people can watch a vote become a sewer, a tutor, or a meal, involvement stops being a metric and becomes a memory. Memories recruit.

Verification: how a stranger recounts us

Involvement is verifiable only if an outsider can check it without trusting the founder. The machinery:

• Public ledgers, federated. Every local table keeps its ledger in the open, and the global figures are sums over public records, recomputed by open code that anyone can run. Open Collective already operates transparent budgets in production, with communities allocating real money in public, and its codebase is MIT licensed and reusable.
• Verifiable counting. Ballot tallies use open, auditable methods, with risk limiting audits, the practice Colorado has run statewide since 2017, as the standard: check a random sample of the physical or digital trail until the reported outcome is statistically confirmed or corrected in public. Microsoft's ElectionGuard, MIT licensed, is the leading reusable toolkit for verifiable election technology.
• Fraud measured, not hidden. Gitcoin's quadratic funding rounds distributed more than 60 million dollars to over 3,700 projects, and its published audits are the honest price list of openness: one audited round removed 853 sybil accounts worth about 606,000 dollars and still carried a residual fraud tax of 0.61 percent, and a later round found roughly 23 percent of contributors were sybils. The Commons publishes its fraud estimate beside every count, with the method. A number with its error bar is credible. A number without one is advertising.
• Personhood without a wall. The design in the democratic foundation stands: enrollment by community vouching, no paperwork requirement. The commercial personhood systems were studied and are recorded here as options, not adoptions: Proof of Humanity registered roughly 19,000 people in its first version and, by its own FAQ, excludes people who cannot show a face or speak on camera, which this Commons will not do. World ID claims 10 million verified people and has met regulatory bans, fines, and a court ordered data erasure in Kenya, which is the price of biometrics as a foundation. One cautionary tale is named so it is never repeated here: a personhood project whose founder acknowledged claiming 9 million identities while fewer than 1 million were verifiably human. The Commons would rather report one honest human than nine million claims.

The shelf: what we may build on, and what we may only learn from

The Commons reuses code only under permissive licenses, MIT, Apache, or BSD, with provenance recorded. Confirmed reusable from the projects' own repositories: ElectionGuard (MIT), Open Collective (MIT), LiquidFeedback (MIT), DHIS2 (BSD), X-Road (MIT, with a license detection caveat to resolve before use), and Keycloak (Apache 2.0) for identity plumbing.

Studied with respect and never copied, because their licenses are copyleft: Decidim, Polis, CONSUL, and Loomio (AGPL), and DemocracyOS (GPL). Their ideas, especially Decidim's component model and Polis's consensus mapping, inform the design freely. Their code stays theirs.

Two items wait on decisions, flagged rather than buried: BrightID's application license is ISC, a permissive license outside the adopted MIT, Apache, BSD list, so adopting it needs an explicit policy call. Your Priorities states MIT but contains a nested AGPL file, so any use requires a per file audit first.

The ladder

• Rung 1, ten citizens: the founding circle enrolled and voting on real, reversible decisions. Phase 0, now.
• Rung 2, one thousand citizens: the Phase 0 gate, December 2026, with the first participatory budget executed and published.
• Rung 3, one hundred thousand: reached through the first institutional adoptions, a school network, a congregation network, a city process.
• Rung 4, ten million: substrate status in one sector, the DHIS2 shape of growth.
• Rung 5, one billion and beyond: only through federation, thousands of local tables and hundreds of institutions under one constitution and one rule, with the counts still recomputable by a stranger.

A rung is claimed only when the published tier numbers cross it for two consecutive quarters. Until then the ladder is labeled with the rung we are actually on. Hope is infrastructure here, and a ladder with its current rung marked honestly is what hope looks like in document form.

Real change, accounted

Participation is not the outcome. Outcomes are counted per layer of the Ascent, with method and data published and independent evaluation invited and funded, with results published either way: learning gains on the tutor's own eval pattern extended to real learners; belonging measured by whether people stay and whether they report being known; shared truth measured by corrections issued and stood behind; and when the open layers are entered, food and shelter counted in physical units from partner records. Failures are published with the same care. The change record sits beside the contribution ledger, so anyone can weigh what the Commons' people did against what the Commons' people changed, and judge whether the ratio is fair.

Colophon: written October 15, 2026, from a charge set by Groundskeeper, who advises this work: figure out how to get billions of people involved, verifiably, and have the collective projects drive real change.