Idea: Bounties for new product additions

[captn3m0]

We currently have ~3.5k USD in our Open Collective account. Our expenses are fairly low, and I don't see that changing. We currently have 178 open requests, out of which 30 are tagged as very-helpful. I'm guessing another round of triaging can get this to 50-60 at the very least.

I'd like to propose a 64USD bounty on the "Very Helpful" requests, to be paid out on a merged PR via our Open Collective.

Pros:

1. The hardest work is research and adding new products. We get to delegate this with real incentives.
2. We find a way to use our funds.
3. More products = the project becomes more helpful.

Cons:

1. The second hardest work is reviewing new products. If we get a flood of new product PRs, we can get overwhelmed.
2. We might face issues around payouts. We are limited by what our Fiscal Host (Open Source Collective) can pay out to, and that might have further limitations as well as delays out of our hands. Eligibility should be made clear to participants.
3. We might get a lot of AI-slop. We're dealing with it anyway, might as well try to put it to good use?

Rules around AI-usage

I've been following LLM usage guidelines in various projects, and the one I really liked was the Jellyfin one. Broadly, these 3 things:

1. LLM output is expressly prohibited for any direct communication
2. pure 'vibe coding' will be rejected
3. you are responsible for what you commit

I'd propose adopting something similar here as well. I don't want to review PRs that have not been looked at and reviewed by the creator. If it works well for the bounty context, we can apply this project wise.

If a reviewer asks a question about a specific fact, the author should be able to explain it themselves.

Why $64?

At 50 closed requests, we get to our current budget.

We might run out of money

That's a good thing. We convert money we aren't using into useful pages. If we run out and the experiment is successful, I'll figure out how to get more of it, and continue the experiment.

Rules of engagement

Most bounties I've seen create "please assign this to me" spam, which I hate. We can have something like:

1. The first "mature" PR is considered. This means your PR cannot be a stub or a placeholder, or an AI-generated one.
2. You can have one open PR at a time.
3. It is on you to check if someone else has filed a PR before you pick up work.
4. Open to maintainers as well.

This results in: file early, but not too early?

Sounds good? Implementation-wise, I can propose:

1. A /bounty page on the website, based on a BOUNTY.md file which sets the rules.
2. Some clarification around eligibility due to money issues.
3. Maybe a site-wide banner. I'd filed Site-Wide and per page Banners just-the-docs/just-the-docs#1784 today, but we can maybe use a simple flash banner meanwhile.

[captn3m0]

@endoflife-date/everyone Thoughts?

[marcwrobel]

I am not a big fan of this. I have the feeling that it betrays the open-source nature of endoflife.date. And I fear it could be a hassle to manage.

Having said that, if you like to try it:

• 64$ seems way to much for a adding single product. A lower price (10$ ?) would seems better and would allow to make bounties a long-term way of adding very-helpful products,
• Some products flagged as "very helpful", such as Spring Security or Apache Shiro, are very easy to add and don't require much work. Other products may be more difficult. We may want to only reward difficult contributions.
• I have the feeling that most products I review requires a lot of additional work (sometime a few hours), so it seems a bit unfair for reviewers, as it does not reward their work.

[captn3m0]

All fair points. It isn't too hard for us to raise funds for this so I am not too concerned about funds.

Reviewer burnout and Very Helpful being too easy is good feedback - thanks. Let me think about this.

[BiNZGi]

I like the idea but I am unsure about the approach. I would also start small and adjust later depending on how it goes.

[Chriz76]

I am new but I hope it is still ok to state my opinion. I like the idea.

Also took a look how urllib handles it. They pay differently on how hard it is to solve the problem that might address the point of @marcwrobel .

I also had the feeling that reviewing is a large part of the work. Would it be possible to pay the reviewers part of the money? They are all known and trustworthy or could it be somehow misused and lead to low quality products?

Like the others said: Start small and with a lower reward. Most people will do it as kind of incentive not to earn money. Maybe a honor list would work as well (or is this already existing?)

[captn3m0]

All good opinions in this thread. I'll fill a PR for the page soon with the ideas from here.

[terrybader]

You already have a realized con... there are 117 PRs waiting with the tag very-helpful (which I would in interested how that gets applied, have a PR waiting myself) thats 30-50ish, already a good number in the pipeline. So what is this solving? - reducing the bankroll only appears to be the motive.

If you do do this, no matter how little, then you will get flooded and then need to $$ for the PRs, monitoring, etc.

And what's with all the "Rules around AI-usage"? The only thing that should matter is the standards for the submission.

[usta]

My 2 cent is how you spend that 3.5k is just buy a large pizza every weekend and eat it for us. ( but we want proof as photos each week )