Pay for what you use.
Share in what you build.

Your cost share + your benefit share.

Calculate each app separately for one calendar month, then combine the results into one platform statement. All examples use US dollars, before any applicable tax.

dᵢ · Your direct cost

Verified, attributable service costs from your intentional activity: database work, storage operations, or an explicitly requested AI task.

S · Shared operating cost

The app’s approved share of hosting and shared services, after provider credits. Shared invoices are allocated once using a published usage rule, never charged in full to every app.

bᵢ · Benefit units

A published proxy for useful activity. Launch default: one unit per UTC day with at least one qualifying, intentional foreground action; at most one per day. Inactivity, background jobs, notifications, and open tabs earn zero.

wᵢ · Your relative share

Your benefit units divided by everyone’s eligible benefit units in that app. Include sponsored and capped participants before applying discounts so their shortfall is not shifted to others.

R · Reward pool

Proposed default m = 1: rewards equal 100% of operating cost. Before caps and rebates, users collectively fund 2 × C: half costs, half rewards. This is a 100% markup on cost, not a 100% profit margin.

Benefit is not objectively measurable. Active days are an understandable starting proxy, not a claim that we know someone’s happiness, income, or relationship success. Each app must publish its qualifying actions before charging. For dating, this could be an intentional browse, like, or message action; message content is not scored.

Zero use means zero charge. If bᵢ = 0, that person’s charge is zero. Their dormant storage and unsolicited background work are sponsor-funded, outside C. If nobody uses the app, all costs are sponsor-funded and there is no user-funded reward pool. A free launch is an explicit subsidy with a published budget, not a debt to recover from future users.

One pool. Three visible shares.

These percentages apply to the reward portion actually collected, after caps, refunds and settlement—not to gross revenue. Operating costs are reimbursed separately.

With $100 of costs and 1,000 equally active users with equal direct costs, the uncapped charge is $0.20 each. If all rewards are eligible, the $200 collected funds $100 of costs, $10 for the platform, $10 for founders, and $80 shared by developers and promoters.

If a reward has no eligible recipient, that share is rebated to the users who funded it. It is never silently absorbed by the platform or another pool. The 80% pool is shared by eligible contributors across the app, including developers and promoters. An individual user having no referrer does not trigger a rebate; if the entire shared pool has no eligible recipients, its 80% is rebated.

One contribution pool for building and growing.

Paying directly per raw line encourages bloated code and penalizes a small, excellent fix. We include code volume as a limited signal alongside reviewed contribution points for development, non-code work, and verified direct promotion.

Pⱼ combines accepted work points from the trailing 12 months with qualified referral points for the current statement month. For development and other reviewed work, each work item gets a public scope and a reviewer-agreed score before work begins: 1 for a small bounded improvement, 3 for a substantial fix or feature, 5 for a major milestone. Completion evidence is required. Non-code work uses the same scale. Work cannot be split into artificial tickets to increase points. Proposed promotion scoring: one point per distinct directly referred user with qualifying activity and a settled positive payment in the statement month, during their first 12 months. Referral points are recalculated each month, not accumulated again over the trailing year. Both roles enter the same points denominator. This starting calibration must be validated and published before launch; point units are a policy choice, not an objective measure of value.

Lⱼ is eligible changed lines from accepted work in the same window: additions plus useful deletions, excluding generated code, vendored files, lockfiles, formatting-only changes, duplicates, and reverted work. A replacement counts once, not once for deletion and again for addition. Squashed diffs and semantic review prevent repeated churn from earning credit. AI-generated volume alone earns no lines credit; reviewed integration, tests and outcomes can earn points.

The sums include all eligible developers, other contributors, and promoters. A promoter can earn points without writing code. There is no fixed developer/promoter split; it follows the combined weights. The square root reduces the advantage of volume. If the only two contributors have equal points but 900 versus 100 eligible lines, their shares are 56.25% and 43.75%, not 90% and 10%. If no eligible lines exist, points receive 100% of the weight. Only people with accepted points can receive weight, including code weight. If there are no accepted points, the entire 80% shared pool is rebated. For example, aggregate final weights of 3 for development and 1 for promotion divide an $80 pool into $60 and $20; equal weights divide it into $40 each.

Named reviewers cannot approve their own rewards. Co-authors agree the point and line split once, before acceptance. Founders can also earn contributor rewards for distinct accepted work; all roles are disclosed. The same expense or work item cannot be reimbursed as a direct cost and rewarded again as a contribution.

A direct referral, not a recruitment chain.

Promoter reward = 80% × collected reward pool × the promoter’s contribution weight. Promotion earns points in the same pool as development, rather than a fixed commission on one user’s payment. One direct promoter per user per app, confirmed by that user before their first qualifying use. Attribution lasts 12 months from that first use; it cannot be restarted with another account.

Sign-ups alone earn nothing. Payments start only after genuine activity produces a settled payment. No self-referrals, duplicate identities, bots, unsolicited bulk invitations, or rewards on someone else’s recruits. Refunds remove the related referral points and require the affected monthly allocation to be recalculated and corrected before settlement. Users without a referrer still support the shared contribution pool; no personal referral rebate applies. After 12 months, that referral no longer earns promotion points.

Promoters must disclose their reward relationship when sharing an invitation. Personal identities and private activity are not published in the public ledger.

Rules that make the formula accountable.