Hey-hey, apologies for putting the "thumb down" and going radio-silent! It's a very important and complex subject, so it's not easy picking words to find the balance between actually communicating information and keeping the debate constructive...

First off, thank you for the effort of putting your vision together and into words. I believe it's hard, it's hard work and genuine. And I'm grateful for it even when I disagree on any particular account.

The proposal. I feel concerned about this, along several lines of thinking, of which the most important are these three questions:

1. Who are the "companies" you target with this proposal?
2. How does this proposed system help maintainers negotiate for higher prices for their work, instead of helping companies always find the cheaper labour and create incentive for dumping? The former being the zero-order approximation of what we need for the project to succeed.
3. What is the CoI story of this proposal?

Now to elaborate. We haven't discussed any of this face-to-face yet, but you're a Board member and in close touch with Yuriy, so I can't imagine you haven't seen my own write-up on the subject (not linking because too dramatic) or heard what I've told Ron and Jonas, which makes me feel like I'm repeating myself, because these are all the same questions I've been asking for feedback on for a year, and I feel like you've ignored them.

"Which Companies?"

I am asking (1) "which companies", because the proposal reads as if completely out of the context of the preceding year's hard talks and negotiations. If this proposal is in any way at all connected to this past year's work, and all of the sacrifices that have been made to grant the Foundation a seat at the table as a pretext for everyone to start building a framework for future collaboration with the very concrete and non-abstract companies, then..

I must say I find that this proposal discusses everything that is unimportant, irrelevant, and nothing that is urgently begging solutions. I fear we might be still trying to achieve very different and deeply incompatible things, the first give-away being your use of the word "donations", and that other readers compare this effort to EC grants corroborates this. After all, we have just seen the demonstration of how companies' boards abhor the notion of a donation, almost as much as they cringe at the notion of working with a non-profit, almost as much as they hate working with a European non-profit. This is the one and only matter on which we-maintainers and they-corporations actually do not have to disagree, because the legal reality is that it is easier to move bigger quantity of money as service fees than as donations anyway, but we're now deliberately trying to make ourselves worse off...

The proposal conjures up a number of compliance rituals and rain-dances nobody has ever asked for during this past year:

• we haven't run into any situation that'd suggest we need designated "grant waves", so why quantization?
• we haven't yet had a need to fix a "wave's" duration, why?
• we haven't had a need to have designated "proposal phases" that have distinct start and end points, so why?
• we haven't actually run into a concrete situation where we need the Board's or the SC's involvement, so what is the motivation for introducing these additional synchronization steps and additional parties? An incredible amount of rain-dancing already, and we haven't even start talking about any actual work!
• "SC decides which proposals get funded", nitpick but well not really it don't? This entire proposal is a match-making system (tuned to minimize the amounts of money spent by the "sponsors" or reaching the maintainers), and once a match is found it'll probably get a go whether with or without the Foundation as a proxy. I don't disagree with the idea that it's actually nice to have the SC in the conversation, but the whole unpnaid ceremonial compliance dancing thing... le sigh
• still the same question for introducing "evaluators" and for pre-defining the reporting rules: why give up and accept the worse terms before the fight even began?
• ...now that we've introduced "evaluators" (nobody asked), why do we pre-define their share?
• a lot more importantly, why do we pre-define the share of the Foundation and WHY DO WE SET IT SO LOOOOOOOOOOOOOOOOOOW? I mean, we made this this blunder with the first round of the anti-cuda funding, okay, there was too much happening, too much stress, but now that we're not in a rush can we finally get this right?
  Details

  This number can be any number, because there's so much inter-dependency between everyone's work in Nixpkgs&co that even 99% is frankly speaking justifiable. "Management overhead" is an expense article for the project, and it's basically pre-approved. It's the first expense that automatically happens and doesn't require any justification or any MS Teams meetings. Once this "management overhead" part of the project budget is spent, it is finally clear of obligations to the corporations, it becomes the Foundation's own money. This means that then finally it can be reinvested in something actually useful to maintainers and to your general human, instead of on the subset of easily-defined projects that very clearly generate company profits.

"Help or harm?"

The proposal is still missing a reflection on the most important question: why try put the Foundation into this position in the first place? I find it self-evident that, materially speaking, the best for a maintainer is to have service contracts directly with their customers. As I wrote in my piece, the only reason ever to consider setting up a proxy, such as the Foundation is, is when there are multiple maintainers offering the same service (project), i.e. when we find ourselves competing. Then there's also the logistics matter, which is that it's cheaper procurement- and taxation-wise for corporations to only deal with the single Trust-entity proxy. Even then the Foundation is the second-worst choice, because these said maintainers are not owners of the Foundation. There must be some Guild-like configurations in which the Foundation can be beneficial instead of harmful for maintainers, where the Foundation helps the match-making and helps negotiate for higher prices, helps to sign direct contracts and establish direct communication channels between maintainers and corporations, but this proposal isn't offering any of that, nor have I come up with anything tangible myself.

I can only offer counter-proposals in the very narrow and counter-factual context of the cuda effort, with the hindsight of how it's gone terribly awry. The counter-proposal for what the Nixpkgs CUDA effort should have been, and what it can't be anymore because the landscape got so much more complex, is, very bluntly and self-indulgently, that it very obviously should have gone through my company, and my company should have donated to the Foundation, no strings attached. (A) "Why hasn't this happened?" I haven't pushed hard enough for this during Autumn-Spring because I was busy trying to keep my head above the water, and especially demotivated after MODUS coming into the picture, and even more so after the initial co-funding proposal. (B) "Why would it have been right?". Because the direct relationship (assuming it's the most beneficial to maintainer) is the default way to go, except when we can't, and in Nixpkgs CUDA's unique situation we can. In case of Nixpkgs CUDA Team as it is today, we haven't any hard conflicts of interests, speaking materially and financially. Even today, I and Daniel are the only people on the team willing or ready to invoice for CUDA stuff, and Daniel's is, for the moment, a completely nominal fee so it's not a conflict... I was hoping this would no longer be the case by now! It's been an entire year since I first asked Ron for an opinion on the COI problem, and this has been another reason I haven't pushed decisively against the previous Foundation Deal. We could have been doing work and making progress during this one year, I could have averted all of my external crises, put out the fires, plugged the holes, and focused on the CUDA bullshit. I could have set up all of the infra one year ago, as planned. We could have gone incrementally and started with cloud compute were it not for the compromised trust and for the insecure position created by the lack of a margin. But we keep playing games and trying to save corporate money, which the corporations don't even want saved. (C) "Does that scale to all of Nixpkgs?". I don't know, not in any obvious way. Generally, and even long-term with cuda, there will be conflicts of interests, so we'll need proxies, and these proxies better be collectively owned. (D) The Foundation could have been helpful in this situation. The Foundation can help me steer the conversation towards the direct relationship, can help me brainstorm how to make this an SLA story, can have help negotiate. Essentially, the Foundation could hypothetically function as a "guild", at least in this very specific and easy case, if not in general. I had said I fear the Foundation becoming a "trust-entity" instead, and your proposal, I think, is bound to make it exactly that...

COIs behind the proposal

It really would help the conversation a whole lot if you included a reflection on what Sylvain-the Board member thinks of how FOSS and/or Nixpkgs funding should work, what Sylvain-MODUS employee, and what Sylvain-Nixpkgs maintainer, all three being essentially conflicting roles. For example, I hope that I've been clear enough that I have a purely materialistic interest in the CUDA stuff, and ergo an obvious conflict of interests.

To conclude...

I don't see why we need this system or any of the low-level details fixed in the proposal, and I see too many ways in which this system is harmful and counter-productive. I still hold my so far unchanged opinion, that the most reasonable way in which "big companies" are able to disburse of money is through service fees, maybe with the exception of special companies which spend money on lobbying, - in any case not through "donations". There are companies that are willing to invest in the ecosystem, and they can invest a "rounding error on the balance sheet" amount of money and it would still be more than enough for us to focus on productive work, be able to satisfy their wishes, reinvest the remaining margin in the actual useful stuff, and never ever have to LARP the European Commission or NLNet. The last two are concerned with a much harder task, which is justifying the expenditure of a very scarce money held by entities that really don't want it spent, especially on something like FOSS. The entities that I assume this proposal is targeting are nothing like that, they are directly benefiting from our ecosystem, and our ecosystem's prosperity impacts their profits. So let's focus on how we have a unique chance to change how FOSS is funded if only we can figure out the exact steps and the exact shape?

As always, "this would have been shorter", but it's already been a week and nobody got the time to wait for me to filter my messy thoughts

UPD1: To put things to scale, if the Foundation were on a course to collect a total of $1bil for a year in "donations" and "funding", I'd not say a word. The way things currently are going, I don't see us reaching 10% of that bar.
UPD2: And again, don't get me wrong, I failed plenty, and I'll fail more, and I'll own my failures. But we're having a discussion about how the Foundation can be of help and the Foundation can be harmful, and here the concrete examples of both, as plain as I could frame them.

Thank you @SomeoneSerge for bringing this up with such a detailed criticism. This sharpens my view as well, to the point I would suggest to @infinisil an editorial pass to deemphasize or leave open the minutiae (also to give some artistic liberty to implementers) and more clearly carve out the incentive structure and economic model. I think it's already in there but will hugely benefit from sharpening. Or in a nutshell: make it more of a guideline rather than a set of instructions, at least for now.

because orgs often can't easily "just hire" individuals

Well we don't want them "just hiring" individuals, at least if "hiring" means employment, because then they can't negotiate for terms that respect COIs

and individuals often don't have the capacity to coordinate large deals

A problem very easily solved with enough resources. There, another prospect for the Foundation to be a provider of help instead of the cheaper competitor to the maintainer:)

EDIT:

the EC and NLnet absolutely want their money spent

The NLNet sure does. But how much effort it takes for NLNet to get their money? Wasn't there a public outcry just recently about the EC trying to cut NGI off funding?

Let's amuse ourselves and ignore the logistics cost of maintaining N * M contracts, one could imagine a match-making system where the money does not go through the Foundation, but instead through the maintainer-owned entity to the Foundation. The Foundation then is a facilitator, the observer, a participant in the bargaining, may oversee the private companies' contract terms and ask for a cut. So is the SC. As a fallback measure, when a group of maintainers doesn't have an entity set-up yet could the Foundation resort to overtaking the deal and signing the contracts instead of the maintainer. The Foundation could eventually learn to provide legal and consulting services, helping maintainers to set up their entities. The Foundation could reinvest its share in financing the maintainers during the very slow bargaining and procurement periods, thus again increasing maintainers' capacity for negotiation instead of diminishing it as the proposal does now.

There, the Foundation is no longer a competitor, but a partner. Can we adjust this fairy-tale so it scales to the project of our size?

@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.