How we got here

Everything on this site is what's left after a lot of other ideas were tried and turned down. This page keeps those ideas, and the reasons we said no.

We think the no's are the most useful part. If you're wondering "why didn't they just…", the answer is probably here — and if you think one of these deserves another look, the reasoning is laid out so you can argue with it.

The licence
Who holds it
Money and price
Votes and decisions
When things go wrong
Enforcement
Words

The licence

Plain MIT

The idea: Release Kanecta under MIT and let it go. It was the original plan.
Why not: Permissively licensed software tends to end up steered by whoever has the most money, and can be out-competed by a well-funded rival giving away a better free version. MIT also says nothing about who keeps the software maintained. "We just wanted it out there; now we want it out there in a good way."
Instead: Open ownership, for big user-facing software only. Small libraries should still be MIT.

Pay to read the source

The idea: An early licence draft with a paid tier just to read the code, a fee schedule, and a special founder role.
Why not: Nobody adopts something they can't evaluate first. A fee schedule the licensor can change on its own is an instant no at any corporate legal review. And every restriction depended on one institution behaving well forever.
Instead: The full source for everyone, a free trial, and one price set by the members.

The hundred-year problem

The idea: Obligations owed to a named organisation, such as "pay the society".
Why not: A fee owed to an organisation that no longer exists is impossible to pay. Any obligation that points at one institution inherits that institution's mortality.
Instead: Every obligation only applies while there's a working way to meet it (see the gradient).

AGPL

The idea: Ship under the strongest existing copyleft licence. It needs nobody to behave well.
Why not: It protects the runtime but can't reach the types, components and functions people build in Kanecta, and that's where the value ends up. Bolting extra terms onto AGPL doesn't work either, because its rules only let you add permissions, not conditions.
Instead: A purpose-written copyleft that names types, components and functions. One AGPL idea survived: removing a legal notice (for us, the membership reminder) is itself a breach.

Offering AGPL as a second option

The idea: Offer unmodified AGPL alongside our licence, to answer "why no OSI licence?".
Why not: The exposure wasn't worth winning a debating point.
Instead: One licence, honestly labelled.

A "pay it forward" licence

The idea: An existing licence: MIT plus a sentence saying you should give something back.
Why not: It's the worst of both. It asks for nothing binding, yet still gets flagged by compliance tools as non-standard. A company can't budget for "do something kind", and a lawyer reads it as open-ended risk. Encouragement without obligation is just a minimum of zero.
Instead: A small, defined minimum. And it isn't money: it's five minutes of attention a year.

A disclosure-only licence

The idea: The only obligation would be to publish a yearly statement of what you're built on, what you gave back, and why you think that's fair. No minimum.
Why not: A minimum of zero can't fund big, broad software. The idea was good in itself, though, so it's been kept as a separate project with its own timeline.
Instead: Membership as the licence's one condition.

"Join the society or publish a disclosure"

The idea: Let people choose between membership and publishing a statement.
Why not: It brought back the hundred-year problem, and it read as paying to avoid transparency.
Instead: Membership for everyone. You choose which mutual, but not whether.

Leaving components outside the copyleft

The idea: Treat things people build in Kanecta like plugins, exempt from share-alike.
Why not: Kanecta progressively builds itself out of its own data, so the code shrinks and the components and types grow. Exempting them would protect the part that matters least.
Instead: Copyleft covers types, components, functions and templates. Only the data in your fields stays private.

Share-alike owed only to users

The idea: Like AGPL, owe the source only to the people using your service.
Why not: That's why almost nobody ever sees anyone's AGPL source. An obligation to the public is more visible, and simpler to write: nobody has to define "user" or "request".
Instead: Share-alike runs to the public.

Open core

The idea: A free version with the good bits held back for paying customers.
Why not: It wears away under revenue pressure, and the small user should get exactly what the big one gets.
Instead: Everyone gets everything. No open core, ever, written into what a mutual must be.

A paid marketplace for components

The idea: Let people sell components to each other.
Why not: It splits the commons and muddies who's licensed to what.
Instead: Everyone pays a little and gets everything. Services, such as hosting, are fine to charge for.

Applying for OSI approval

The idea: Apply, eventually.
Why not: The licence breaks "freedom zero" on purpose, so it can't be open source by definition. Applying would only manufacture a "licence fails review" story. Being vague about it is what sank other ethical licences.
Instead: "Source available, not open source", said plainly and without apology.

Who holds it

Three boards

The idea: A workers' board, a users' board and a custodian board with a veto, plus a protected founder role.
Why not: Too much machinery, and every board is something to capture.
Instead: Two permanent ends, a trust and the members, with everything between them replaceable.

An incorporated society that owns everything

The idea: Give Kanecta to a members' society.
Why not: A society can always vote to wind itself up and share out its assets, so its members could vote to sell. The asset has to sit above the members, out of their reach.
Instead: A trust holds the asset.

A society as the "democracy" layer

The idea: Trust as the vault, a society for the democracy, mutuals for the work.
Why not: The society would have been the soft underbelly: the only layer with both real power and the ability to change itself. Capture it and you could reach the vault through the price vote.
Instead: No central membership body at all. The members are whoever belongs to a qualifying mutual, and the register is spread across the mutuals.

A company as the permanent owner

The idea: Keep the copyright in an ordinary company until the direction is clear.
Why not: A company can be sold and can rewrite its own rules. That's fine for something temporary, and fatal for something permanent.
Instead: A company only as a short bridge, if at all. The trust is the permanent home.

One worker co-op at the centre

The idea: A single worker co-operative owning and running Kanecta, with shares split between workers and users.
Why not: If one group can always outvote the other, it isn't really shared. Shares that can't be sold or profited from don't make sense to buy into. And one co-op is one thing to capture.
Instead: Many mutuals, anyone can start one, and none of them holds anything precious.

A founder power to amend the deed

The idea: Keep a power to fix the deed later, in case the first draft was cheap and wrong.
Why not: Three reasons. Security: if one person can change it, every threat to Kanecta becomes a threat to that person. Capture: a special power is a prize. Credibility: "community-owned, except the founder can rewrite it" is exactly the gap between paper and practice we object to in others.
Instead: A tiny core nobody can amend, and rules that can be extended around it but never used to weaken it.

A protected founder role or job

The idea: Entrench the founder in the constitution, removable only by a supermajority.
Why not: It makes the founder a target and everyone else suspicious. Being an actual equal is better.
Instead: Security comes from there being many mutuals to work for, all non-profit and worker-run. It's structural, not contractual.

A royalty to the founder

The idea: A small percentage of the trust's income to the founder.
Why not: There's nothing to take a percentage of, and a capped royalty looks like selling in instalments.
Instead: Wages for real work, from a mutual, like anyone else.

A trust that "does things"

The idea: Let the trust step in and sort out problems, for example a failing mutual.
Why not: Discretion is the thing worth capturing.
Instead: A deliberately dull trust, ideally one any lawyer could run. Every would-be decision becomes a stored default that applies automatically when an objective condition is met.

The trust approving mutuals

The idea: The trust decides which organisations qualify.
Why not: A trust that can say no is a trust that can be captured into saying no to the wrong people.
Instead: Notification, not permission. Publish rules that meet the criteria and you're registered.

A charitable trust

The idea: Register as a charity.
Why not: Three reasons, all pointing the same way. It probably wouldn't qualify, because Kanecta benefits its members rather than the public at large. It doesn't need to, because the trust's no-benefit rule already locks the asset. And we don't want to: we're happy to pay tax, and a charity halo would undercut the plain honesty of "ordinary people paying their bit".
Instead: A private purpose trust, subject to legal advice.

Building the whole structure first

The idea: Don't release anything until the complete, perfect trust exists.
Why not: Building heavy governance before anyone uses the software is backwards, and it held everything up for a long time. A rushed deed would need tightening later, and tightening reads as betrayal where loosening reads as generosity.
Instead: Build in layers. A tiny permanent core first; the rest can be added later, around it.

Money and price

Institutions pay; individuals chip in a token amount

The idea: The usual model: the real money comes from a few big payers.
Why not: Funding is control, whatever the constitution says. If most of the money comes from a few organisations, they own you.
Instead: A little from everyone. Individual members are the main act, not a side stream.

Organisations contributing engineers instead of money

The idea: Big users "pay" by lending developers.
Why not: Engineer-hours bring the contributor's priorities with them. That's how big projects end up steered by their largest sponsor.
Instead: Money is preferred. It pays a mutual's own people, so control stays with the project.

Organisations voting through paid seats

The idea: An organisation that pays more gets more say.
Why not: Then money buys influence, which is the thing we're trying to stop.
Instead: Organisations can belong, but only individual people vote. The money scales; the say doesn't.

Each mutual sets its own price

The idea: Let the market decide.
Why not: Mutuals are non-profit and their main cost is people. Competing on price would force them to squeeze wages and poison the culture.
Instead: One price everywhere, so mutuals compete on how good they are to belong to.

The mutuals agree a price between them

The idea: The mutuals get together and set the common price.
Why not: That's a cartel.
Instead: The price is a term of the licence. The members vote on it directly, and the mutuals have no say. (This still needs a competition lawyer's sign-off.)

A cheap way out

The idea: A commercial licence, for organisations that don't want membership or share-alike, at about twice the member price.
Why not: Too cheap. Buying out should be a deliberate, slightly uncomfortable decision, not a routine line item. It's the price of not taking part.
Instead: A commercial licence priced well above membership. The exact multiple is still open.

Prices written into the licence

The idea: Put the fee, or a schedule of fees, in the licence text.
Why not: A number in binding text is a fee schedule. It invites the "royalty" reading, and lawyers see open-ended liability.
Instead: A stable licence that points to a single price, which the members set and the trust records.

The trust collecting the money

The idea: Everyone pays the trust, and the trust funds the work.
Why not: The trust should never touch members' money. That's a pile of cash worth capturing.
Instead: You pay your mutual, the way you pay a registered plumber rather than the regulator. The trust gets small registration fees for its own costs.

Votes and decisions

A committee or assembly sets the price

The idea: Members gather, debate, and decide.
Why not: The vault is only as safe as the process that feeds it. Capture the forum and you poison whatever the trust records.
Instead: A direct poll of every member, counted by a fixed rule. A forum can host the debate but can never stand between the members and the result.

One big central election

The idea: The trust runs a global vote.
Why not: A billion-person central election isn't dull. It needs a central roll, servers and staff, which is exactly the kind of capability the trust isn't supposed to have.
Instead: Each mutual counts its own members and publishes its tally, and the trust adds them up. Publishing every tally makes cheating visible.

Someone calls a vote

The idea: Votes happen when someone asks for one.
Why not: Whoever controls the trigger controls the agenda, and can spam it.
Instead: Recurring decisions happen on a fixed schedule. New questions need a petition with enough signatures. Either way, the trigger is a fact, not a choice.

A separate price election

The idea: A dedicated vote on the price every year.
Why not: Turnout. Special elections reach the keen few.
Instead: One line in the annual declaration everyone already does: "What do you think the price should be?" The rule that comes with it: one line, maybe two someday, never a ballot paper.

Voting on the code, or a maintainer class

The idea: Use the same voting machinery for technical decisions, or define an official group of maintainers.
Why not: A price has a middle and a yearly rhythm; code has neither. An official maintainer class is one more thing to capture.
Instead: The mutuals are the maintainers. The main version is a commons they choose to pool into, so it's the main version because it's good, not because it's official.

When things go wrong

A deadline after which it becomes MIT

The idea: After a set number of years, everything becomes MIT.
Why not: It ties the escape hatch to the calendar instead of to anything actually going wrong.
Instead: See the next two.

The "poison pill"

The idea: A permissive licence that springs into effect if the trust fails or is sold.
Why not: The attack is built in. A hostile company could kill the trust to force everything permissive, then out-compete it.
Instead: The gradient (next).

What replaced them: the gradient

The idea: Every obligation applies only while there's a working way to meet it. No mutuals, nobody to pay. No registry, nothing to publish to. When someone rebuilds the machinery, the obligations come back.
Why it works: There's no single trigger to pull. You'd have to kill every mutual and every registry everywhere, and stop anyone reviving them. That's genuine death, the one case where going permissive is right.

Naming who gets Kanecta if the trust ends

The idea: The usual wind-up clause: the asset goes to the government, or to a charity.
Why not: A government would then have both the reason and the means to capture it. A charity has no particular reason to keep the source open, and might sensibly sell it for cash: a well-meaning back door. An asset lock is only as strong as its dissolution clause.
Instead: Nobody gets it. On dissolution the conditions fall away and Kanecta becomes free to everyone, so killing the trust to grab it achieves nothing. Leftover cash and equipment go to an open-source or digital-commons organisation, named by category rather than by name, because a named charity may not exist in fifty years.

"It can never be sold for profit"

The idea: The original wording of the core promise.
Why not: "For profit" is a loophole. It doesn't catch handing it over for a dollar, at cost, licensing it exclusively (selling control without selling the asset), or mortgaging it.
Instead: Two separate locked clauses. The asset can never leave the trust by any route, and nobody gets private benefit from it.

Keeping attackers out

The idea: Use identity checks and membership rules to stop a hostile takeover.
Why not: A large enough organisation can flood any pool with real, named, genuinely contributing people. It just hires them.
Instead: Make the prize worthless. A captured trust can't sell, can't profit, can't close the source and can't stop a fork. The security isn't the lock on the door; it's that the room is empty.

Enforcement

Licence keys, activation, phone-home

The idea: Software that checks you're a member and stops working if you're not.
Why not: It would turn Kanecta into the kind of software it's meant to replace, and it betrays the people who are supposed to own it. Make the right thing easy and the wrong thing clearly wrong, then stop.
Instead: Honour and law. A friendly reminder that never blocks, and a licence term that makes removing the reminder a breach.

An identity check on the free trial

The idea: Verify who's trialling, to stop people re-trialling forever.
Why not: It means building surveillance, it punishes honest people, and it wouldn't stop the determined. Friction on a trial is almost fatal: we abandon software that makes us jump through hoops just to try it, too.
Instead: The full app, unlimited, for three months, then a reminder that stays. Serial re-trialling is simply a breach of the licence.

Email required to download

The idea: Collect an email before the download.
Why not: Mirrors are allowed, so a gate only catches the most trusting people. The download page is the moment of most doubt.
Instead: No gate. An empty email field quietly shows what we mean.

Means-tested hardship

The idea: Prove low income, or have a panel of peers decide.
Why not: A decision process needs a living institution, and creates a record of who is poor. It burdens honest people with paperwork while liars sail through.
Instead: You declare it, and that's final. It's visible only inside your own mutual, among people you know. A separate rule stops companies dressing up as paupers. Trust for persons; look-through for schemers.

A written exemption for people at risk

The idea: A clause excusing dissidents, or people with no access, from membership.
Why not: To claim it, they'd have to identify themselves. That's worse than nothing.
Instead: No special clause. If no mutual is reasonably available to you, you decide that yourself, and nobody asks.

Only joining a mutual in your own country

The idea: Members join a mutual where they live.
Why not: It isn't needed, and it would turn "none available" into an everyday escape.
Instead: One global pool. Join any mutual, anywhere.

A different badge for people who don't pay

The idea: Show paying and hardship members differently.
Why not: Then the badge is a means test.
Instead: The same green dot for everyone: paying, hardship or in kind.

"Powered by Kanecta"

The idea: Sites running Kanecta must show a "Powered by" banner.
Why not: Compulsory "powered by" banners have a long history of being resented.
Instead: A badge about the operator, "run by members of a co-operative", which a commercial licence removes.

Words

"Community ownership"

Why not: "Community-owned" invites an "is it really?" argument, and clashes with how community energy and land trusts use the term.
Instead: "Open ownership". It makes the safe, true claim that ownership is open to anyone. It also sits naturally beside open source without claiming to be it.

"Contributors"

Why not: In open source it means people who write code, and it scares off everyone else.
Instead: Owners. Owners, not users.

"Ownership at an affordable price"

Why not: It sounds like you're buying ownership, which is precisely what we're against.
Instead: Ownership comes from being a member, one person one vote. The fee is dues, not a stake. Someone on hardship owns exactly as much as someone paying.

Leading with "chip in"

Why not: It makes money the point.
Instead: Ownership first. "Owned by the people who use it, which means you join, chip in a little, and get a say."

Explaining everything on the home page

Why not: People skip the home page and go straight for the download button.
Instead: The same short message in three places, each a little more specific: the home page, the download page, and the app's first run. The long version is here, for people who chose to read it.

"I"

Why not: It centres a lone founder, which is the story we're building away from.
Instead: "We".

Still undecided: the questions we haven't settled are listed at oosl.org/open-questions.