[michaelt]

  An honest FOSS license + dual licensed for commercial
  would say "it's fine if you make money, and we don't need
  a cut... unless you've integrated our software AND you
  don't want your final product to be similarly
  open-source."

I think the "problem" the commons clause is aimed at is where AWS starts selling managed deployments of your product, and they make $$$$ from your work without giving you a cut. And maybe your business plan was to do that, but AWS has an existing billing relationship with everyone making them a much lower friction choice.

[yebyen]

So, the model that AGPL was designed to combat, right?

Under the AGPL, Amazon would be free to do that, as long as they released the code for their managed deployment systems.

(This doesn't get you any money, but it saves you from the sense that your FOSS-work has been exploited by a commercial entity that doesn't give anything back, so long as it's actually enforceable.)

[Bartweiss]

I think this is a lot of people's complaint with the Commons Clause.

It's defended by citing a real problem, but the AGPL also addresses that problem. Meanwhile, its impact on non-Amazon players like "some random user who wants to write and share a handy tool" is much less pleasant, because it effectively turns that user into a free profit source for the license holder.

I can imagine a healthy role for some intermediate license that goes beyond AGPL to say "actually Amazon, you have to pay us for this". But it would need to not distort what FOSS means for everybody the way this does.

[slededit]

AGPL says nothing about monetization. It only says Amazon would have to open any modifications they made.

Unless Amazon becomes the new primary developers it does not solve the problem of paying the people actually doing the work.

[yebyen]

Right, what it does potentially solve (comparing GPL to AGPL) is this:

Amazon takes your work and solves the "service layer" story, but having never distributed any binaries, they are not obligated to share their modifications in any way.

Under the AGPL, if you solve the core issue, and they borrow your solution adding a proper service layer to it, that would need to be released as source code, in a way that users could repeat the deployment on their own.

I say "potentially solves" because nobody is solving 80% of the problem better than anyone else can and releasing their 80% solution as AGPL, saving "Amazon jobs" for the Amazon people. And if they were, it would be easily circumvented; Amazon would simply never take the bait. Best case, they would figure out what makes your solution so much better and then implement those ideas for themselves in a clean-room.

[dragonwriter]

> Unless Amazon becomes the new primary developers

Which, if they really are the dominant beneficiary, they probably will; whoever is making the most money from the software has the most financial interest in maintaining it, and that's the reason so many big corps that benefit from FOSS invest heavily back into those projects, either via sponsorship of projects or paying their own developers to work on the codebase or both.

But that doesn't serve the interests of the investors in the company doing the initial work, who invested in the hopes of capturing the downstream revenue later development of the work would produce. Which is understandable, but they shouldn't attempt to trade on the popularity of open source licenses with their non-open alternatives.

And that, not the “people doing the work”, is what the Commons Clause is about.

[slededit]

Doesn't really happen with Openssh. Or MySQL or Postgres for that matter. With Open Source there are a lot more takers than givers.

[dragonwriter]

> Doesn't really happen with Openssh.

Who is the big player dominating revenue from OpenSSH? Projects with extremely diffuse benefit like that (or, perhaps more clearly, OpenSSL) do often have a problem here, but that's not what Commons Clause claims to be directed at.

> Or MySQL or Postgres for that matter.

Postgres definitely gets lots of support from significant commercializers of Postgres like EnterpriseDB, Postgres Pro, Citus, and Heroku.

MySQL, maybe not, but a GPL arrangement, especially one requiring copyright assignment, discourages third-party commercial contributions, especially when the owner is actively dual-licensing. Who wants to pay for development of IP someone else can sell more freely than you can?

(Permissively-licensed SQLite, OTOH, has huge contributions for downstream commercial users.)

> With Open Source there are a lot more takers than givers.

By design. There are also more givers than with propietary-licensed software, also by design.