Proposal: Grant program and purpose-bound funding

[infinisil]

In recent months, primarily the NixOS Foundation board members, me and @Ra33it0, have been drafting this proposal to fill some gaps in the NixOS Foundation's funding model. Note that this is not yet fully board or SC approved, which we'll need to get before this is implemented.

Brief summary

Motivation:

• Enable companies to finance Nix ecosystem improvements in a more streamlined and transparent way
• Enable contributors to get paid for working on Nix ecosystem improvements in a more streamlined way
• Provide a way to connect funding partners to contributing individuals, e.g. for when:
  • Individuals don't want to be employed/contracted directly by a company, and don't want to be limited by receiving funding from only one company
  • Companies don't want to employ/contract an individual directly, and want to provide funding in a more flexible way to multiple individuals
  • Companies want a trusted third-party to hold the funds until work is complete
• Strengthen the ecosystem by providing funding resources to contributors using general NixOS Foundation donation profits not tied to any companies

There are two inter-related parts to this proposal:

• Grant program: A way for contributors to propose ideas and get paid by the NixOS Foundation for delivering milestones. The funds can come from both general-purpose donations and purpose-bound funding.
• Purpose-bound funding: A way for companies to provide funding for specific grants proposals or team initiatives.

Current status

• Revised after feedback from board and SC members
• Revised after feedback from official teams
  • Pinging these teams for feedback (somewhat arbitrary subset from the homepage): @NixOS/nixpkgs-core, @NixOS/security @NixOS/infra @NixOS/marketing-team @NixOS/moderation @NixOS/cuda-maintainers @nixos/nix-team @NixOS/documentation-team @NixOS/nix-formatting @NixOS/ngi @NixOS/nixos-hardware @NixOS/hydra-maintainers
• Shared with the wider community
• Shared with companies
• Approved by board and SC

If we need to add more context behind individual parts, we'll add this on demand.

[fricklerhandwerk]

I think this is a very good idea overall, and the rough structure seems sound. It's a miniature version of what the European Commission does. If the costs and benefits of doing such a thing are well understood, this can be a good mechanism for coordinating action that would otherwise not happen.

The only thing I'd caution against is assuming that the capacity to operate such a scheme will magically appear out of the blue. This needs considerable up-front investment from everyone involved, and quite a bit of time for building experience and routines, including those to trust and delegate and exercise the control mechanisms outlined here. What may help bootstrapping it is managing expectations with all parties: Initially we can expect that sponsors won't get a good bang for their buck, implementors and reviewers won't get good rates, organisers will do serious work on their free time, balls will get dropped, and there won't be too many exciting success stories for the general audience of users or maintainers. But that's how things go and will be fine if it's approached as a learning process aimed at getting it right eventually.

[fricklerhandwerk]

This section underspecifies how and when evaluators are appointed. Not saying you need to nail it down now, but since evaluation (or call it "review") is a well-known and somewhat well-understood bottleneck on delivery, it probably needs to be planned ahead of time. I'd even say it needs to be priced right into the original proposal. Which is why it sticks out to me that you fix the review overhead by fiat; I argue this will be highly domain- or even case-specific, because reviewer qualification and time isn't fungible.

[infinisil]

Resolved by 4b5b23a

[fricklerhandwerk]

That commit does not address my comment. What you get by a low flat review fee is likely no reviewers or useless reviews. What I meant by "price it into the proposal" is that a proposal should allocate the cost of reviews, which will be domain-specific and highly dependent on who's involved in the first place. Imagine: I propose to add a non-trivial feature to Nix, and maintainers get 5% for review. This is how you make nothingburgers.

Conversely, if any interesting work can only be done with maintainer involvement anyway, why not just scrap the public competition and give the money to maintainers directly to have them figure out on their own?

[cafkafk]

I already gave feedback in the call yesterday, so I've approved it. I'd still like to see more teams feedback of course

[YorikSar]

I think it would be better not to use "decided" with the board. Maybe "calculated" or "tabulated"?

[lisanna-dettwyler]

Nothing especially stands out to me as problematic, but I think @fricklerhandwerk made excellent points.

[infinisil]

@fricklerhandwerk @SomeoneSerge Thanks for the feedback, we'll consider it for the next revision. To leave more room for other people I'd like to request you continue your discussion among yourself in an out-of-band channel for now.

[tomberek]

Looks good, just would like to balance out the criteria with positive reasons as well.

[tomberek]

Most of this section goes over reasons to reject; harms and their various categories. Recommand having at least a section of reasons to accept; positive reasons about evaluating benefits of a proposal in terms of accomplishing the Foundation's goals or being in line with the Values.

Otherwise. Looks good overall!

[cafkafk]

I’d think about it in terms of which criteria are actually useful for applicants and for the selection process.

From an applicant’s perspective, it helps to understand what would make an application ineligible or clearly out of scope. That avoids unnecessary effort, makes the process feel more approachable, and gives people a straightforward way to assess whether it makes sense to apply.

I would be more cautious about adding criteria for what a "good" application looks like. Once you start listing positive criteria, anything not on that list can easily be read as less relevant or less desirable, even if that was never the intention. This is a dynamic seen in hiring, for example, where people often self-select out when they do not see themselves reflected in the stated criteria.

Not having positive criteria gives the selection process more room to recognize strong applications on their own merits, including ones that may be good in ways we did not anticipate. The criteria here are useful as a way of defining the boundaries of the process, not as an exhaustive description of what we might value. I think a fuller discussion of what "good" can look like would be a separate document entirely and would be misplaced here.

[infinisil]

Me and @Ra33it0 have now made a number of revisions based on the feedback so far. I encourage everybody to re-read the proposal in full, but a notable point is that we've strengthened the team initiatives concept in comparison to grant proposals: Purpose-bound funding for grant proposals now has an additional 20% maintenance and 5% evaluation charge (in addition to 10% for management), while purpose-bound funding for team initiatives only has the 10% for management. Furthermore, funding for team initiatives is encouraged to be recurring instead of one-time.

In general I'd also like to point out that this proposal is not perfect, and will need changes over time. While we try to be as forward-looking as possible, at some point it's better to try something out and iterate on it with real-world data, instead of never getting it out of the discussion phase.

[nixos-discourse]

This pull request has been mentioned on NixOS Discourse. There might be relevant details there:

https://discourse.nixos.org/t/nixos-foundation-draft-proposal-for-grant-program-and-purpose-bound-funding/79732/1

[SomeoneSerge]

I only briefly skimmed the document. I think I have the same principal question as before: what are some of the real-world efforts we're trying to accommodate here, so it's possible to evaluate the proposal against them?

[SomeoneSerge]

Do such individuals exist? Why should the Foundation cater to such individuals, if there were any, instead of helping individuals to get direct contracts?

[SomeoneSerge]

Why should the Foundation help companies avoid direct contracts or create hidden employment relationships?

[SomeoneSerge]

Why should the Foundation help Companies in their anti-competition efforts?

[SomeoneSerge]

Suggested change

	These programs should incorporate messaging to encourage contributing individuals to seek employment contracts or contracting agreements with funding partners.
	These programs should incorporate messaging to encourage contributing individuals to seek contracting agreements with funding partners.

...but I can't claim I understood the intent of the paragraph in the first place

[SomeoneSerge]

Already asked in the previous round, but why "waves"? Why not a continuous call?

[SomeoneSerge]

Please let's omit all numbers that aren't the order of $10^6..10^9$ from the document (applies to the diagrams too)

[SomeoneSerge]

Suggested change

	- Procure funding for experimental projects, fundamental research, and the "invisible work" that is otherwise hard to market.
	- Provide contributors and their projects with interim funding during sponsor search or bargaining periods.

...here is one (and, fwiw, only) use-case for a "grant system" that I currently see. All of the work that is too removed from "profit-making" for the market to show interest in, maybe projects whose time hasn't come yet. Running a mainline kernel on nokia 3310 or whatever weird stuff Matthew does...

EDIT: added the 2nd, a variation of the 1st

[SomeoneSerge]

I'm getting the impression that "purpose-bound funding" is expected to be the main source of cash for the grant system, and I really hope that's not the case? Anything that is "purpose-bound" is probably clear-cut enough to be a commercial contract between for-profit entities, but then what the Foundation should do is collect a sort of a "tax" (no strings attached) from such commercial fees in exchange for facilitating such deals and helping companies and contributors connect...

[fricklerhandwerk]

I agree that the focus should be for the foundation to provide a service to the community, if nothing else then merely to keep its non-profit status. It is easy to see such a scheme as a match making mechanism, but how is that not an agency?

[SomeoneSerge]

Should be possible to have private submissions until they're approved

[SomeoneSerge]

Ideally the team submits something to the grant program that the company then submits funding for using the purpose-bound funding mechanism

Something I'd very specifically recommend discouraging! On the very rare occasion of a "double coincidence of wishes", the Foundation should be helping the team create an upward pressure for the price instead of fixing it low...

[fricklerhandwerk]

The following are goals, not motivations. The document doesn't state why these are goals to begin with.

[fricklerhandwerk]

This document is missing measurable outcomes or success criteria, making it impossible to judge whether the stated goals were reached by the implementation.