Three Answers to Who Pays the Maker: Fair Source, Post-Open and the Commons Stack
Part one took Fair Source apart: a non-compete licence with a two-year fuse, defined by the vendors who use it, wearing a word that belongs to farmers and workers. It’s an easy target. The harder part is that Fair Source is an answer to a real question, and dismissing the answer doesn’t make the question go away. Who pays the people who make the software?
The Commons Has No Ledger for This put that question at the centre of the Digital Commons series: free software licences protected the artefact but never built a way to pay for making it. Three proposals now take that gap seriously enough to rebuild something around it. They head off in three different directions: fence the market, levy the large users, or fund the work from the act of joining. Comparing them says a lot about what each one is really trying to protect.
The Man Who Wrote the Definition
The second proposal comes from someone who should know. Bruce Perens wrote the Open Source Definition, adapted from the Debian Free Software Guidelines he drafted in 1997, and co-founded the Open Source Initiative with Eric Raymond the following year. When the person who wrote the definition says it has stopped working, that’s worth listening to.
His diagnosis, on the Post-Open project site, is blunt: “the economic structure of Open Source is one of resource extraction”. Businesses profit, the developers carrying the load don’t, and the strain shows. Perens points to the 2024 XZ Utils backdoor, where a lone, overstretched maintainer was worn down by a patient attacker until malicious code got in, as a warning of where that strain leads.
His answer is a licence that “will never call itself Open Source, because it has different rules”. It works roughly like this:
- Free for most users. Individuals and smaller organisations get the whole Post-Open Collection at no cost, under a Zero-Cost Licence with rights similar to open source.
- A levy on large users. Any entity with annual revenue above USD 5 million, or that runs the software as a service for others, embeds it in a paid product, or wants to keep its modifications private, signs a paid contract: 0.5% of revenue between USD 5 and 10 million, 1% above that.
- One collection, one bill. Users license the entire collection, not individual programs, and report once a year what they use. Infringe or breach the contract and you lose rights to the whole collection.
- Money to developers, not companies. A not-for-profit collects the fees, a CPA firm handles the confidential usage reports, and after costs, the money is split per program by usage, then among each program’s contributors.
- Governance by makers. Governance is “exclusively by individual software creators”. Users get a voice; corporations don’t get control.
Perens’s own model for this is the music industry’s performance rights organisations, the collecting societies that license songs to broadcasters and venues and pass royalties back to songwriters. Australians know the local version as APRA AMCOS.
Credit Where It’s Due
The big difference between Post-Open and Fair Source is who ends up with the money.
Fair Source protects a company by fencing off a market. The non-compete applies to everyone, a solo developer as much as a hyperscaler, and the beneficiary is the company holding the copyright. The developers who contributed to that codebase from outside get nothing from the fence, and whoever wants to compete with the vendor is locked out for two years.
Post-Open leaves the market open, in the sense that matters to small users: the free tier carries rights close to open source. Then it charges the organisations big enough to pay, in proportion to their size, and sends the money to the people who did the work. That’s the right direction. It’s also close to what Leanne Ussher described in the ledger post: a contribution priced like a tax, where individual benefit outweighs individual cost and the proceeds fund the commons, rather than a market price that collapses to zero for anything digital.
Post-Open is also honest in the way Fair Source is honest. It says outright that it isn’t open source, and it doesn’t borrow a word with someone else’s reputation attached.
So it’s the more serious attempt, and the one closer to what this site argues for. For that reason it deserves the same scrutiny I gave Reticulum's licence. Good intentions on their own don’t make a governance model.
The Collecting Society Problem
The collecting society analogy is a good one, but it cuts both ways. A collecting society works because it’s the only one: one licence covers every song, so venues pay one fee and stop worrying. That same structure is why APRA sits under continuing scrutiny from the ACCC. Its blanket licensing has been authorised since 1999, most recently in September 2026 for another five years, a decision now before the Competition Tribunal, and only on conditions: it must explain its licences and distribution rules, report its revenues, distributions and costs, and run a dispute resolution scheme. The ACCC’s determination calls APRA a “virtual monopoly” with “access to a great deal of market information that is not available to APRA Members and APRA Licensees”. Nobody is suggesting collecting societies are villains. The conditions are there because a single collector, sitting between everyone who pays and everyone who gets paid, holds a lot of power that nobody else can see into.
Post-Open would build that same kind of collector, and give it a few extra powers on top.
The whole-collection licence means a dispute over one program puts a company’s rights to every program at risk. That’s a very strong enforcement lever, and the problem is who gets to pull it: whoever runs the collection. The paid contract asks software-as-a-service businesses to “submit the identity of customers” they run Post-Open software for. That’s a list of who depends on what, held centrally, for every SaaS provider in the scheme. And the usage reports, the numbers that decide who gets paid how much, are confidential to a CPA firm by design. That’s sensible for protecting commercial information. It also means developers have to trust a distribution they can’t audit.
I’m not suggesting anything sinister here. It’s the same structural point I’ve made about Reticulum’s founder holding every lever, and about TradeLens in The Decentralisation Paradox: a system that routes everything through one trusted centre is only as good as that centre. Post-Open would be a better centre than most, but it would still be a centre.
A Ledger That Only Counts Commits
The part that matters most to me is how Post-Open decides who did the work.
The answer is git: “To understand a developer’s contribution to the Post Open Collection, we will instrument git repositories.” Payment starts with developers and documentation writers. “Payment for other roles will be developed later.”
That’s the same filter Martin Owens described in the ledger post, just turned into an accounting system. The person who triaged 400 bug reports, translated the interface into Vietnamese, tested a release on hardware the maintainer doesn’t own, or answered the same forum question for the fortieth time doesn’t appear in a commit graph. Counting commits also rewards the work that’s easy to count. Anyone who has seen a team measured by lines of code knows how quickly people start working to the measure.
There’s a better model already available, and it’s the one we’re building GrowGood on. In Valueflows, work is a first-class economic event with its own effort, provider, receiver and timestamp, whether it produced code, a translation, a test report or an answer to a confused new user. Sensorica’s contribution accounting goes a step further and publishes its value equation before the work happens, so every contributor knows in advance how an afternoon of documentation weighs against a week of circuit design. Next to that, a confidential usage report, git instrumentation and a distribution formula that hasn’t been written yet look pretty thin. Post-Open has put a lot of care into collecting the money. The ledger, the part that decides whether the payments are fair, is the least developed piece of the design.
Identity, Passkeys and Who Gets Left Out
Post-Open’s answer to the XZ attack is to “positively identify all participating developers” and issue them hardware passkeys, with every change signed and tracked from check-in to user. As supply-chain security, it’s hard to argue with. A known, accountable contributor with a hardware key is much harder to impersonate than a GitHub account with a reused password.
The cost lands on a group the commons has always protected: pseudonymous contributors. Plenty of solid free software has been written by people known only by a handle, some for privacy, some because their government or employer would make their life difficult if their name appeared. Identity checks also need someone to do the checking, which brings us back to the single centre. This could be a membrane in Martin’s sense, a boundary that lets trust move between strangers, or it could end up as a gate that only some people can get through. The project hasn’t said yet how it will handle this.
Still a Draft
Fairness also means being clear about where Post-Open actually stands. By its own account, the licences are drafts, a law firm has agreed to review them pro bono, many documents and processes aren’t written yet, and the project “will require external financial support for its first several years of operation”. In 2024 Perens told The Register some terms “would have to be written better to be upheld by a court”. The 2024 draft also barred using the code to train AI models, the same kind of purpose restriction I argued against in Reticulum’s case. Here, at least, it comes in a licence that never claimed to be open source. I haven’t found evidence that any developer has yet been paid through the scheme.
I don’t mean that as a criticism. Building institutions takes years, and starting from a published draft is how it should be done. But it means Post-Open is a proposal, not a working alternative, and nobody should relicense a project onto it yet. The project’s own site says as much.
The Third Answer: Fund It by Joining
The third proposal starts from the shared pool of money rather than from a licence.
Jeff Emmett co-founded the Commons Stack in 2019 with Griff Green and Michael Zargham. Michel Bauwens credits him with the idea of “peer for peer”: moving beyond peer-to-peer networks, where equals talk to each other, to infrastructure communities build to provide for one another. It’s the shift I wrote about in The Monasteries of the Höllental, and the subject of Emmett’s own session at the Valley of the Commons in August.
The Commons Stack answers “who pays the maker?” with two mechanisms that work as a pair.
- The augmented bonding curve funds the pool. Anyone who wants in buys into the community at a price set by a published curve, a share of every purchase goes into a common pool, and leaving early costs something. In Emmett’s words, “Funding for shared work then comes from the act of joining rather than from a grant round”.
- Conviction voting spends it. Members put their weight behind proposals, support builds up the longer it’s held and decays if withdrawn, and a proposal is funded once its accumulated support crosses a threshold that rises with the size of the request. Big asks need broad, lasting backing. As Emmett puts it, “Nobody has to be online at the right hour, and a whale who shows up on the last day has bought nothing”.
What I like most is that Emmett insists the models behind a funding system be published and runnable by anyone, because “a simulation only its designers can run is an authority claim wearing a lab coat”. It’s the same thing I’m asking of Post-Open’s confidential usage reports. The rules that decide who gets paid should be visible before anyone contributes, and in a conviction-voting system they are: the curve, the thresholds and every vote sit in the open.
The approach is also close to the Valueflows model we’re building GrowGood on, and Emmett says so himself. In his Valley of the Commons session, he walked the room through REA accounting, “resources, events, agents”, as tracking “a more holistic picture than just transaction back and forth”. He described peer for peer as a channel that becomes “a recorded event with named agents”, and asked for “the contributory account of the collective” in place of user counts and KPIs. Both treat an economy as flows between agents rather than balances in an account. Emmett’s own tooling models funding as flows with conservation enforced, and his book Exploring MycoFi, written with Jessica Zartler and Sadie Morris, takes fungal networks as its design pattern. The ledger we’re designing for GrowGood aims to go one step further and record a mycorrhizal network as an agent in its own right, with its phosphorus transfers logged as economic events. Where Emmett uses the mycelium as a metaphor, we want to actually record it in the books.
Where the two differ matters for the question this post is asking. The Commons Stack is very good at allocation: deciding, in public and continuously, where the pool goes next. Its record of contribution, though, is the token balance, which shows who paid in, not who did the work. Valueflows records the work: the hours, the translations, the diagnosed bugs, whoever did them. Each one fills a gap in the other, and I haven’t seen anyone put them together yet.
The Real-World Test
These mechanisms have actually been tried. The Token Engineering Commons (TEC) used an augmented bonding curve and conviction voting to fund research and education in token engineering for several years. In July 2025 its community began proposing a sunset, and a formal shutdown proposal followed in October. The claims window, extended a week past its 20 June 2026 deadline, let holders redeem their share of what remained. The final results, published in July: 70 of 724 eligible wallets claimed, taking 65% of the treasury, about USD 146,500. Roughly 90% of the claimable allocation sat with the top 100 wallets. The unclaimed remainder went to Giveth.
To me that’s a lesson about the token rather than the voting. Conviction voting kept the spending rules public for as long as the TEC ran. But the bonding curve turned membership into a tradeable position, and tradeable positions concentrate. A whale who shows up on the last day can’t buy a vote, but one who accumulated early holds most of the pool by the time it’s shared out. I’ve written about this before with blockchain governance: decentralised mechanics, centralised holdings.
To Emmett’s credit, when the TEC’s forum went offline with the shutdown, he rebuilt it as a read-only archive from Wayback Machine and Common Crawl captures, which is where I checked these figures. “If the host disappears, the data does not,” as his site says. That’s the commons practising what it preaches.
My own view is that the voting mechanism works without the token, and Emmett seems to agree. When someone in the Höllental audience suggested blockchains were “the tool that was missing”, his answer was “We don’t need necessarily blockchains”: lighter tools such as CRDTs can give similar guarantees “with about 1% of the infrastructure cost”, without “validators, tokens, shilling, all that kind of stuff”. Conviction voting needs weighted members, not a market. A co-operative’s member shares, one per member or weighted by recorded contribution, would do. That’s the version I’d want to test: Valueflows to record the work, conviction voting to allocate the pool, and no token anywhere in between.
The Three Answers, Side by Side
| Fair Source | Post-Open | Commons Stack | AGPLv3 | |
|---|---|---|---|---|
| Who it protects | The vendor | Individual developers | The common pool | The commons |
| Mechanism | Non-compete for two years per release | Revenue levy on users above USD 5 million | Joining tribute plus conviction voting | Copyleft, including network use |
| Can a co-op offer a competing service? | Not with current code | Yes, though running it as a service triggers the paid contract, whatever the co-op’s size | Not a licence question; depends on the software’s own licence | Yes, if it shares its changes |
| Who sets the rules | Sentry and the adopting companies | A developer-governed not-for-profit | Token-weighted members, in public | Published licence, no central body |
| Status | 14 companies listed | Draft, unfunded | Flagship deployment (TEC) wound down 2025–26 | In production for 19 years |
| Who gets paid | The vendor | Code and documentation contributors, by git measure | Whoever proposals fund | Nobody, by default |
The last row of that table is where my own preference runs into trouble. The AGPLv3 protects the commons better than any of the alternatives, and it pays nobody. That was Martin’s point and Leanne’s point in the ledger post, and it hasn’t gone away.
What a Ledger Would Need

Fair Source answers “who pays the maker?” with “the company does, from a protected market”. That’s just proprietary software with readable source. Post-Open answers “large users do, through a collective”, which is the right answer to the right question. The Commons Stack answers “everyone who joins does, and the members decide in public where it goes”, which gets the allocation right and leaves the record of work to a token balance. What all three are still missing is the part of the commons that isn’t just a payment pipeline.
From what I’ve followed so far, it needs at least five things:
- A published formula. How money moves from usage to programs to people should be public before anyone contributes, the way Sensorica publishes its value equation and the way conviction voting runs in the open. A confidential report can stay confidential. The rules that turn it into payments can’t.
- Contribution beyond commits. Testing, translation, documentation, triage and support recorded as work from the first day, not “developed later”. Valueflows already has the vocabulary.
- More than one centre. Collecting societies are mostly national, and answer to national regulators. A federated model, with regional or project-level stewards under shared rules, would remove the single point that holds every lever. Emmett makes the same case for money itself: “Monocultures are efficient right up to the point where the single crop fails.”
- A ledger of work, not of holdings. What gets someone a say in the pool should be what they did, recorded as Valueflows economic events, not what they bought. That’s the join between the accounting we’re building into GrowGood and Emmett’s allocation, and nobody has built it yet. It also needs the kind of reputation checks Sensorica pairs with its ledger, because, as Emmett warned in the same session, “As soon as we’re tracking metrics, those metrics can and will be gamed.”
- Contributor-elected governance with a way out. Post-Open’s governance by creators is a good start, and conviction voting is the best continuous allocation method I’ve seen. Commons governance also needs an answer to “what happens when this body gets it wrong?”, and in a commons, that answer always comes back to the right to fork and leave.
Fair Source puts a fence around a market and calls it fair. Post-Open is the first serious attempt in decades to put a price on the commons without fencing it in, and it has worked out how to collect the money. The Commons Stack has the best method I know for deciding, in public, where that money goes. What’s still missing is a ledger of who actually did the work, and joining that up with the other two is what peer for peer is really about.
Part two of two. Previously: Enclosure on a Timer: Who Fair Source Is Actually Fair To. Related reading: The Commons Has No Ledger for This on Valueflows, contribution accounting and funding pro potentia, The Network Is Not the Owner on who owns a network versus how it’s wired, and The Monasteries of the Höllental on peer-for-peer production.
Sources
- Post-Open, About Post-Open, and the draft Zero-Cost Licence (0.07, August 2024). The “resource extraction” diagnosis, the free and paid tiers, the USD 5 million threshold and 0.5% and 1% rates, whole-collection licensing, git instrumentation, developer identification and passkeys, SaaS customer disclosure, and the need for external funding.
- The Register, interview with Bruce Perens on the Post-Open draft, 30 April 2024. The Zero Cost Licence draft, the AI-training restriction, the performance-rights analogy, and Perens on the draft’s legal weaknesses.
- Slashdot, “Slashdot’s Interview with Bruce Perens”, December 2024. Licence vetting and project status.
- Open Source Initiative, The Open Source Definition, and its history as an adaptation of the Debian Free Software Guidelines.
- ACCC, Australasian Performing Right Association Limited, authorisation AA1000661, final determination, 11 September 2026. Five-year authorisation with conditions; “virtual monopoly”.
- Emmett, J. jeffemmett.com. The augmented bonding curve, conviction voting, published models (“an authority claim wearing a lab coat”), MycoFi and the monoculture argument.
- Bauwens, M. (2026). “Understanding the importance of the P4P moment in the long evolution of the commons”, Fourth Generation Civilization, 16 August 2026. Credits Emmett with the peer-for-peer concept.
- Token Engineering Commons, “TEC Claims Process: Final Results”, 11 July 2026, via Jeff Emmett’s read-only archive of the TEC forum.
- Valueflows, valueflo.ws, and Value Equations. Work as a first-class economic event, and published benefit-redistribution formulas.
- Freund, A. (2024). “backdoor in upstream xz/liblzma leading to ssh server compromise”, oss-security mailing list, 29 March 2024.
- Post-Open
- Commons Stack
- Fair Source
- Jeff Emmett
- Conviction Voting
- Open Source Funding
- Valueflows
- Commons Governance
Comments
Be the first to comment! Reply to this post from your Mastodon/Fediverse or Bluesky account, or mention this post's URL in your reply. Your comment will appear here automatically via webmention.
Follow this blog on Mastodon at @gaggl.com@web.brid.gy or on Bluesky at @gaggl.com