# PRIOR-ART.md — RETURN in Context ## How RETURN relates to, differs from, and learns from its predecessors **Version:** v0.2 companion document **Date:** 2026-07-04 **Author:** Thomas Schüller · independent researcher · Vienna, Austria **ORCID:** 0009-0003-9799-6747 · forschung@tschueller.com **License:** CC BY-SA 4.0 --- ## 0. Why this document exists Anyone who has followed open-source licensing for the past decade will read the RETURN Protocol and ask, correctly: *hasn't this been tried?* The honest answer: **parts of it have, repeatedly, and most attempts either failed, stayed marginal, or triggered community backlash.** This document names every relevant predecessor, describes what each actually did, states honestly which of their problems RETURN inherits unchanged, and identifies precisely which mechanism properties are new. This is not marketing. Where a criticism of a predecessor applies equally to RETURN, it is stated as such. --- ## 1. The two established families — and the thin third one Prior approaches fall into two well-populated families and one nearly empty third: **Family A — Source-available licenses** regulate commercial use through the license itself, almost always with a *binary* mechanism: commercial vs. non-commercial (PolyForm Noncommercial, Prosperity), competing vs. non-competing (FSL, Elastic License, Confluent), over vs. under a user limit (Fair Source 25), before vs. after a time delay (BSL, FSL), or inside vs. outside an identity category (Peer Production License). **Family B — Non-license funding mechanisms** leave the software fully free and collect money through voluntary, subscription, or state channels: GitHub Sponsors, Open Collective, Tidelift, thanks.dev, the Open Source Pledge, the Sovereign Tech Fund. **Family C — mandatory, revenue-proportional return with full usage freedom** — RETURN's category — is nearly empty. It contains one prominent draft (Bruce Perens' Post-Open, 2024, not yet operational) and one marginal ideological approach (the Peer Production License, which is closer to Family A in mechanism). There is no established, successful precedent in this category. That is both the opportunity and the warning. --- ## 2. The predecessors, one by one ### 2.1 License Zero / Parity / Prosperity (Kyle E. Mitchell, 2017–2020) A dual-licensing "vending machine": developers licensed work under Parity (strong copyleft) or Prosperity (non-commercial with a 30-day commercial trial), and License Zero sold private commercial licenses at fixed price tiers through Stripe, taking a commission. **Status:** shut down in 2020. Mitchell himself cited platform complexity and the difficulty of the subscription model; the license texts survive, the marketplace does not. **How RETURN differs:** no time trial, no per-seat license tiers, no commercial sales agent. RETURN grants immediate, permanent freedom and replaces license *purchase* with a revenue-proportional *obligation* administered by a non-profit with no sales interest. **What RETURN inherits:** the core question License Zero could not answer at scale — will commercial users actually engage with a payment mechanism they can simply ignore? — applies to RETURN identically. ### 2.2 PolyForm Project (Mitchell + Heather Meeker, 2019–) A family of standardized source-available templates (Noncommercial, Small Business, Internal Use, Free Trial, and others), intended as the "missing licenses" analogous to Creative Commons. The Small Business license uses a threshold aligned with the EU SME definition. **Status:** active, moderate niche adoption (best-known adopter: EPPlus). Community discussions document persistent ambiguity around "permitted purpose" and category boundaries. **How RETURN differs:** PolyForm uses static *category* thresholds with individually negotiated commercial licenses above them. RETURN replaces category jumps with a continuous formula and negotiation with automatic computation. **What RETURN inherits:** the boundary-definition problem. PolyForm's "is this use commercial?" becomes RETURN's "what revenue is attributable?" — a different question with the same legal fuzziness. ### 2.3 Business Source License — BSL/BUSL (MariaDB 2013; HashiCorp 2023) Time-delayed open source: code is always visible, non-production use is always free, production use requires either an Additional Use Grant or a commercial license, and after a Change Date (maximum four years) the code automatically converts to a genuine open-source license. **Status:** active and widespread (MariaDB, Sentry formerly, CockroachDB, Couchbase, HashiCorp). The HashiCorp switch in August 2023 triggered the largest license backlash in recent memory: the OpenTF manifesto gathered over 30,000 GitHub stars within a month, and the OpenTofu fork now lives under the Linux Foundation. **How RETURN differs:** no time delay and no competition clause. RETURN-licensed work is free for everyone from day one, permanently; the obligation is financial/in-kind, never a usage prohibition. **What RETURN inherits:** the OpenTofu episode is the strongest recent evidence that large communities will *fork rather than pay* when a license change feels extractive. RETURN's answer — that its terms are fixed from the start, not changed midstream — is real but untested. ### 2.4 SSPL (MongoDB 2018; Elastic 2021) AGPL with a radicalized Section 13: anyone offering the software as a service must open-source their entire service stack. Submitted to the OSI in 2018, withdrawn in 2019 after sustained objections; the OSI board later stated flatly that the SSPL is not an open-source license, citing discrimination against a field of endeavor (OSD 6) and encumbrance of separate programs (OSD 9). Debian, Red Hat, and Fedora dropped MongoDB. **Notably:** Elastic returned to open source by adding AGPLv3 in September 2024, and Redis returned via AGPLv3 with Redis 8 in May 2025 — the first major *reversals* of the source-available wave, signaling that restrictive relicensing costs more community trust than it captures revenue. **How RETURN differs:** no disclosure obligation of any kind. RETURN never requires a user to open their stack; it prices use instead of punishing a business model. **What RETURN inherits:** the OSD 6 conflict, fully. RETURN conditions obligations on commercial use and is therefore **not open source under the OSI definition** — see § 4 below. RETURN says this itself before others say it. ### 2.5 Fair Source (Sourcegraph 2015/16) and FSL / Fair Source movement (Sentry 2023/24) Two generations. The original Fair Source License (drafted by Heather Meeker for Sourcegraph) used a **per-organization user limit** — free below N users, paid above. Sourcegraph itself later abandoned it. The Functional Source License (Sentry, late 2023) is a simplified BSL: free for any non-competing use, with a fixed two-year conversion to Apache 2.0 or MIT. In 2024 Sentry launched fair.io as an umbrella ("Fair Source") with a formal definition requiring delayed open-source publication; adopters include Sentry, Codecov, GitButler, Keygen, and PowerSync. **Status:** growing but niche. OSI-side critics reject the category: the freedoms "do not apply to everyone; they discriminate based on legally fuzzy noncompete rules" (Thierry Carrez). **How RETURN differs:** neither user limits nor time delay nor noncompete clauses. And where Fair Source's restriction is *qualitative* (what kind of use), RETURN's is *quantitative* (how much revenue) — which makes it computable but not automatically fairer. **What RETURN inherits:** the "legally fuzzy" charge lands on RETURN's AttributableRevenue definition exactly as it lands on FSL's "competing use." RETURN's § 3.1 fallback rule (0.1 × total revenue ceiling per dependency) narrows but does not eliminate the fuzziness. ### 2.6 Commons Clause (2018) and Elastic License 2.0 / Confluent Community License The Commons Clause — a rider forbidding "selling" the software, applied by Redis Labs to some modules in 2018 — produced immediate, broad backlash ("Tragedy of the Commons Clause," RedMonk) and was replaced by Redis's own license within six months. ELv2 and the Confluent Community License survive as cleaner anti-managed-service prohibitions. **How RETURN differs:** all three are *prohibitions* with no payment path. RETURN prohibits nothing; every use is permitted, priced, and dischargeable in cash or work. ### 2.7 Peer Production License / Copyfarleft (Kleiner 2010; Bauwens) Commercial use forbidden *except* for cooperatives, non-profits, and commons-based organizations — an identity-based exclusion designed to keep value inside the cooperative sphere. **Status:** marginal since publication. The critiques are structural: excluding users of non-rival goods harms the commons itself; successful commons projects do not use identity exclusion; one academic treatment calls the PPL a rent-seeking instrument. **How RETURN differs:** RETURN excludes no one by identity. A venture-backed corporation and a workers' cooperative face the same formula; only revenue differentiates them. ### 2.8 Bruce Perens' Post-Open (2024, draft) — the closest relative This deserves its own weight, because it is the only prior approach in RETURN's mechanism family, and it comes from a co-author of the Open Source Definition itself. Post-Open, drafted from early 2024, proposes: free use below **USD 5 million** annual end-user revenue (an aggregate, veil-piercing definition); above it, a **stepped fee of 0.5 % (USD 5–10 M) or 1 % (above 10 M)** of revenue, paid once for the *entire* Post-Open collection; a developer-owned non-profit collects and distributes, with a CPA under NDA sequestering revenue data; and — critically — **in-kind substitution exists**: entities that publish sufficient contributions to the collection owe nothing and may even be paid. **Status:** explicitly a non-operational draft. No software released, no confirmed operational funding, and the license header itself warns it is not the product of an attorney. Documented criticisms: 1 % of revenue is enormous for thin-margin businesses and large companies may simply blacklist such software; per-line distribution metrics invite make-work gaming; a single license-mandated collecting organization is a centralization risk (critics invoke the ISOC/.org affair); an FSF representative compared it to a collecting society with "inherent problems… in distributing the funds fairly and transparently." **How RETURN differs from Post-Open — the five real deltas:** 1. **Continuous progressive curve** (asymptotic max 1.5 %) instead of stepped tiers — no cliff at any revenue level. 2. **Karma memory** — a time-decaying reputation multiplier ([0.5, 1.5], three-year half-life) rewarding sustained contribution and pricing sustained extraction. Post-Open has no memory; **no predecessor anywhere has this**. It is RETURN's strongest novelty and, precisely because it has no precedent, its least-tested hypothesis. 3. **Formula fixed by charter, changeable only by fork.** Post-Open's distribution runs on instrumented metrics plus organizational discretion; RETURN's Executor is charter-locked with zero discretion, and evolution happens through parallel successors. 4. **Per-work obligations** computed individually, rather than one fee for a pooled collection with centralized redistribution. 5. **A much lower floor** (EUR 100k attributable revenue vs. USD 5 M total revenue) — RETURN engages mid-sized commercial users that Post-Open exempts entirely. Whether that is a feature or an adoption barrier is an open empirical question. **What RETURN inherits from Post-Open, unchanged:** every major criticism. Enforceability against unwilling corporations. The collecting-society centralization worry (RETURN's answer — redundant Register mirrors, multi-sig treasury, charter lock — is a design response, not proof). The blacklisting risk. And the honest acknowledgment that revenue-based obligations may simply cause large procurement departments to route around the licensed work. ### 2.9 Voluntary funding, quantified (Family B) The empirical case for *mandatory* proportionality rests on the gap between value taken and money returned: - Harvard Business School Working Paper 24-038 (2024) estimates the demand-side value of widely used open-source software at **USD 8.8 trillion** — firms would need to spend roughly 3.5× more on software if OSS did not exist. - Against that: GitHub Sponsors has moved **over USD 33 million cumulatively** since 2019. The Open Source Pledge (launched October 2024, ≥ USD 2,000 per developer per year, backed by Sentry) has moved **~USD 4.5 million cumulatively**. Tidelift's own maintainer survey reports six in ten maintainers unpaid, with direct corporate payments at 5 % of maintainer income. Germany's Sovereign Tech Fund has invested **~EUR 23.5 million across 60 projects** — meaningful, state-funded, and structurally incapable of scaling to the size of the gap. Voluntary mechanisms work, measurably, at roughly **five to six orders of magnitude below the value consumed.** That gap is RETURN's founding premise. It is also fair to note the counter-reading: perhaps the gap persists because *any* payment friction — voluntary or mandatory — exceeds what the market will bear for non-scarce goods. RETURN is a bet that the friction, not the obligation, was the problem. --- ## 3. Comparative mechanism table | Approach | Commercial trigger | Threshold | Rate / price | Time delay | Cash vs. in-kind | Memory | Governance | |---|---|---|---|---|---|---|---| | **RETURN v0.2** | revenue-proportional | 100k EUR attributable | progressive, asympt. 1.5 % | none | cash OR in-kind 1:1 | **karma, 3-y half-life** | foundation, formula fixed, fork-only evolution | | License Zero / Prosperity | binary commercial | none | fixed license tiers | 30-day trial | cash only | none | commercial agent | | PolyForm | category (NC / SME) | SME definition | negotiated | none | cash only | none | licensor | | BSL / BUSL | production / competition | Additional Use Grant | commercial license | 4 y → OSS | cash only | none | licensor | | SSPL | "as a service" | none | none (disclosure) | none | stack disclosure | none | licensor | | Fair Source (2016) | user count | per-org user limit | license fee | optional | cash only | none | licensor | | FSL (Sentry) | "competing use" | none | commercial license | 2 y → OSS | cash only | none | licensor | | Commons Clause | "selling" | none | prohibition | none | — | none | licensor | | ELv2 / Confluent | managed service | none | prohibition | none | — | none | licensor | | Peer Production License | identity (non-coop) | none | prohibition for capital | none | — | none | licensor | | **Post-Open (draft)** | revenue | **USD 5 M total** | **0.5 / 1 % stepped** | none | **cash OR published code** | none | developer non-profit + CPA | | Open Source Pledge | voluntary | — | ≥ USD 2k/dev/yr | — | cash only | none | voluntary | | Tidelift | subscription | — | ~USD 100–150/dev/yr | — | cash only | none | company catalog | | Sovereign Tech Fund | state grant | — | project funding | — | state cash | none | state agency | --- ## 4. Positioning: deliberately not OSI open source The Open Source Definition, point 6, prohibits discrimination against fields of endeavor. RETURN attaches an obligation to commercial use. Therefore **RETURN is not an open-source license under the OSI definition, and does not claim to be.** Every predecessor that blurred this line — SSPL most visibly — spent its public debut litigating definitions instead of mechanisms. RETURN's category is stated plainly: a **market mechanism for non-scarce goods** — a fourth category alongside proprietary, open source, and the delayed/restricted "fair source" family. Its nearest declared neighbors are Post-Open and the Fair Source movement, both of which have likewise stopped claiming the "open source" label. The OSI maintains open source as a binary gate, not a spectrum; RETURN does not ask the gate to move. --- ## 5. What RETURN inherits unsolved — stated plainly Intellectual honesty requires a consolidated list of predecessor criticisms that apply to RETURN **without modification**: 1. **Enforceability.** No revenue-based obligation has ever been litigated to judgment. RETURN's formula-derived damages (§ 8.2 of the protocol) simplify the quantum but not the jurisdictional reality of suing an unwilling corporation. 2. **Definition fuzziness.** "AttributableRevenue" is the same species of legally contestable term as "competing use" (FSL), "commercial purpose" (PolyForm), or "end-user revenue" (Post-Open). The proportional fallback ceiling narrows the dispute space; it does not close it. 3. **In-kind valuation.** Pricing contributed code, documentation, and data at "market value" via a fixed price table is untested at any scale, and the make-work gaming critique aimed at Post-Open's metrics applies to RETURN's attestations. 4. **Centralization / trust.** A single canonical Executor is a collecting society, with the fairness-of-distribution problems the FSF already named for Post-Open. Charter lock, multi-sig, and Register mirrors are mitigations designed on paper, not proven in adversarial practice. 5. **The rational-avoidance scenario.** The most likely large-firm response may be neither payment nor contribution but *avoidance*: procurement policies that exclude RETURN-licensed dependencies entirely, as critics predicted for Post-Open's 1 %. 6. **The eternity trap.** A charter-locked formula that turns out to be miscalibrated cannot be fixed, only forked — and forks fragment exactly the network effects the karma system needs to matter. Two RETURN-specific hypotheses have **no precedent at all** and must be labeled as design hypotheses rather than proven features: the karma memory (no prior mechanism has priced history) and the continuous progressive curve (no prior license has avoided threshold cliffs). Their parameters — 100k floor, 1.5 % asymptote, 3-year half-life, [0.5, 1.5] bounds — are calibration guesses, not empirically derived values. --- ## 6. Signals worth watching Three external developments would materially change RETURN's position: - **Post-Open becomes operational.** Its success or failure is the single most informative natural experiment for RETURN's category. - **State funding scales.** The proposed EU Sovereign Tech Fund (feasibility study 2025; a suggested EUR 350 M from the 2028–2035 EU budget) and the Cyber Resilience Act's maintenance pressure could partially solve the funding gap through public money — making mandatory-return licenses less necessary, or complementary. - **A fourth category formalizes.** If the fair-source / source-available space consolidates a recognized definition beyond OSI's binary, RETURN should dock there rather than stand alone. --- ## 7. Summary in three sentences Every individual building block of RETURN — revenue thresholds, percentage obligations, in-kind credit, foundation governance — exists somewhere in prior art, most completely in Bruce Perens' still-unbuilt Post-Open draft, and every predecessor's unsolved problem (enforceability, definitional fuzziness, centralization, avoidance) transfers to RETURN intact. What no predecessor combines is RETURN's specific mechanism set: a continuous progressive curve instead of cliffs, immediate permanent freedom instead of delays or limits, 1:1 in-kind substitution, a karma memory that prices behavioral history, and a charter-locked formula whose only evolution path is a fork. RETURN is therefore best described not as a new idea but as a new *combination* in a category with one other serious draft and no operating precedent — which is exactly why it must launch as an open draft seeking falsification, not as a solution claiming validation. --- **Sources:** This document condenses a structured research report (July 2026) drawing on official license texts, OSI board statements, vendor announcements (MongoDB, Elastic, HashiCorp, Sentry, Redis), LWN.net, The Register, InfoQ, TechCrunch, RedMonk, postopen.org, fair.io, the Harvard Business School Working Paper 24-038, Sovereign Tech Agency press releases, and the Tidelift maintainer surveys. The full report with citations is preserved in the project archive.