One of the cool things about code is that you can build stuff out of thin air, basically for free. It's not like woodworking where you have to pay for the wood.
We are moving into a weird time where people are assuming that now we have to pay machines churn out code.
Somehow they packaged up our own ability to think and are selling it back to us. If they can get us to forget how to do it we'll be the perfect customers, dependent forever.
That’s what I don’t get about this whole AI push. It’s a global rug that everyone plans to pull, it’s sold at cost, it already presents major risks, everyone seems to be aware of all that and yet - the consumers and lawmakers are relatively quiet.
Its happening at a global moment of chaotic behavior. When prices are soaring, information and institutions are disintermediated, jobs are hard to come by and layoffs easy, insurance is tied to your job and your kids' well being to your location -- well, most people get real quiet about stuff.
Even then I think a LOT of people are saying something but the narrative mechanism of inevitability is really strong.
Say all the SOTA (remote) LLMs went away tonight, would programmers everywhere suddenly be giving up on programming? Maybe some, but unlikely that most would. It's not so much of a rug-pull if it's an inconvenience at most when it goes away. Maybe some would take some time to readjust (some days/weeks/months?), but then it'd be back to normal again.
I'm not forcing anyone to use tools they don't want. The Amish can continue making their handcrafted furniture, but I'll make sure to work on construction sites that have power tools.
In your mind this sounds like “damn kids, get off my lawn”, and to the kids it sounds like, “I will only ever plow my fields with donkeys, tractors are too complex for something like plowing”.
Just continue down the path of making it so only the oligarchs can afford to access them, power them, and buy capable hardware to run them locally.
Also helps the oligarchs to have their government assets issue impossible restrictions like “no foreigners can access” that isn’t technically outlawing models for everyone.
> Say all the SOTA (remote) LLMs went away tonight, would programmers everywhere suddenly be giving up on programming?
That's assuming that the code base in question is still well maintained and designed. If this were to happen I would bet that heavily vibe coded things - like Bun and Claude Code - would pretty much immediately hit a wall, where humans just can't comprehend the accumulated mess well enough to make productive changes.
> would pretty much immediately hit a wall, where humans just can't comprehend the accumulated mess well enough to make productive changes.
Plenty of people spent lots of time of their career essentially cleaning up code and refactoring after others. Yes it'd be a lot, but there are really giant spaghetti's out there that been 100% built by humans, then conquered by refactorers.
> Say all the SOTA (remote) LLMs went away tonight
Most likely that would cause company pulling the rug on their frontier model(s), to fall over themselves. Meanwhile the competition keeps going. Hence no top AI player does that unless forced to.
You can only build it for free if you don't value your time. That is good for personal projects and hobbies but which company can build stuff without costs? There was always an expense to generating code. Plus you have to pay for equipment, electricity, subscriptions to stuff for their staff etc.
I don't really think they are packaging up ability to think abd selling it back to you. There is nothing stopping you from not using AI and even companies, many firms just dont trust AI and don't use it.
But the idea that code could be built of thin air is not true in case of actual businesses.
I really dislike it but the effectiveness of code generation is just too good now. I really doubted it at first but I can churn out 10x the work. Is it the same quality as IC? no but I think that it doesn't matter anymore. How many of us have worked on some 10+m loc monstrosity for years never knowing more than 1% of the codebase? Probably most. Now I can generate that same monstrosity on my own. I've created a 1.1m+ loc and growing trading platform that automates options trades based on live ML edges from research papers. I never would have been able to do that on my own before LLM's. It has full telemetry and comprehensive error logging and testing as well. I'd have been strained to get MVP done before LLMs.
I don't think someone without coding experience can do it though. I've accepted that the new expected output for a single dev will eventually converge to 10x-100x what it is now.
I like the "maintainer stays in control" part, but isn't that also a problem in a way?
The AI provider gets paid, the platform gets paid (20% is a lot in my opinion!), and the maintainer gets more unpaid work: another PR to plan, review, revise, merge, and then maintain... that's a lot of work.
If people are willing to fund an issue, why should that money mainly cover LLM tokens rather than maintainer effort? Or at least, why doesn't the leftover money go to the maintainer instead of back to the donors?
Tokens are better than money. They let you train your skills to get top 1% on LeetPrompt so you have an entry in the lottery to interview for your next temp position.
While I definitely like the Patreon for Software Builders idea, that's got some moving pieces which take additional legal work. My hope is that could come in time as it would be really cool.
Regarding rewarding maintainer effort, I'm shooting for the value prop of "free AI", this only works if reconciliation is per-phase and liquidity is accessible across as many repos as possible. So if I had each reconcile drain the pool, there would be a lot of stalled work and human intervention required.
That said, there are probably some maintainers that don't want "free AI" and that's okay.
At first I was like "i want to use ai but dont have the money to burn for api tokens" cool.
But then I realized the backers are essentially saying "i have money and could support developers but i choose to give the money directly to a mega corp and skip the human".
I recommend you remove the policy of "Whatever the run didn't spend goes straight back to your backers' wallets." and make sure the human behind the wheel gets to eat. Somehow
While we're in the token-equivalent of ZIRP, tokens don't cost what they cost, so there's sort of arbitrage to be had. I have tokens I've been given than I'm not using, but that's not the same as me having been given cash in the first place.
> While we're in the token-equivalent of ZIRP, tokens don't cost what they cost, so there's sort of arbitrage to be had
Do you have a source for this? I believe “personal subscription” plans on OpenAI and Anthropic are likely ran at a net loss or close to it, but all indications elsewhere are that API pricing for these companies and likely Google as well are profitable per API call [1][2]. I would definitely believe that the Chinese players are operating at a loss though if that is what you mean.
Just finance the project for god sake, why introducing another "gift card" that they sell at cashiers? Let the maintainer spend the money, maybe he/she needs bread, not tokens
Just curious, why is there a login gate before seeing the list of projects that participate in the platform? Usually similar donation(?) websites list those publicly for better visibility and less friction.
Probably an oversight on my part. I was thinking that backers would find out about CleverCrow through the project maintainers so the public pages are repo specific.
As an aside, when you do login, CleverCrow shows projects that you've starred on GitHub to help find things you might want to support.
100% expected, in fact, desired (maintainers should be/are in charge). My goal is to create enough surface area that backers see their tokens going to many of the things they care about, not necessarily all.
Really interesting solution to the AI PR problem. Keeping the maintainers in the driver's seat for issue prioritization is definitely the right approach.
How are you handling the token allocation under the hood, is this managed via a GitHub App integration, and can backers target specific issues or just the repo as a whole?
I've got a couple good docs in the footer of www.clevercrow.io that will answer the token stuff. GitHub app integration for all repo interaction. Yep, backers can target specific issues or entire repos, their choice.
I always thought that the donate tokens thing would be done by sending some tokens from your personal sub to the maintainer's pool with some sort of proxy for tokens with rules, in a more direct way without doing it in cash, but yeah that's where the sweet fees live.
If this gets any traction, the "share tokens with a friend" could be good PR for the labs, instead of buy me a coffee, buy me a clanker.
Better yet: give them cold hard cash instead of what is arguably monopoly money for many OSS devs. Ironically this is something GitHub made "easy" with sponsorships several years ago.
You don't sponsor people or projects to complete specific issues or build specific features in the first place. Sponsorship is a reward and token of appreciation for doing good work.
Some don't mind doing the overall reward and appreciation thing. And some just have that particular issue that they want handled so the project works - better - for them. Both cases are valid.
I'm not sure how that's related to your initial question of "How do you ensure that funds ear-marked for a donor-specified issue goes toward that issue and not something else?"
The point is to get something funded. That's the goal. And negotiation doesn't scale. You can bet that nobody will negotiate contracts with 20 maintainers if they want particular features/fixes for 20 different projects that they're using. Otherwise it would be a thing today.
But instead we have these attempts and stopgaps, some of which have had some success here and there. This is something else in that pool making it easier to fund stuff, and if it gains traction than we'll know that it's serving its purpose. I think it has good potential.
I wasn't thinking about companies at all. Companies tend to only contribute to well-known projects, which leaves a very long tail.
I'm thinking about the far more random regular people using random projects that may not be that popular. Like I just did a rough count of the OSS apps on my phone: 60+. Of those there are 5 that I can immediately think of improvements that I'd like. One of them I've actually made improvements to already because it was a dead project that I really like and so had to revive anyway. And I used Claude to work on it as I'm not a mobile dev; been planning to republish but haven't gotten around to it. I won't bother to check my laptop but there's even more there.
Now imagine if I (and others interested in those projects) could contribute to small funding pools for various projects of interest, with the assurance that said funds go to that feature/fix which is of interest or gets refunded. I think that'd result in the general OSS support needle being pushed that much further over time.
You hit up the maintainer and negotiate a deal for that?
If all you’ve got is relative pocket change they probably aren’t going to agree but if you put real money behind it and it doesn’t go against their vision of the project then most people would be willing to accept actual contracting work to expand their project.
Pretty straight-forward solution: donors or maintainers don't use the service if it isn't good. But the maintainer is "accountable" (to the extent one maintaining a OSS project is) as they're the one actually providing the prompts and doing QA.
I think I've read from a few different sources that the Claude Code $100-$200/mo plans are subsidized so hard that it's basically $2k-$8k/mo in "would-be" equivalent API token usages.
This kind of makes sense in that space while the subsidies (if true) last?
Unrelated, "tokens" feels very like... back-then blockchain to me. All the craze.
Or the API is massively overpriced. What Anthropic/OpenAI are charging for tokens says almost nothing about their actual costs, just what people are willing to pay.
It’s just an obvious example of market segmentation by charging enterprise customers many times more than personal users while selling the same product
Yeah, rising token costs definitely played into my thinking too. I want builders to be able to use the best models and it seems like they are getting more expensive. But maybe local models will get there?
It does seem like a worse version of a FOSS donation platform that uses regular old money. I guess one advantage is that you can ensure your money goes directly to using AI to solve specific problems on the codebase, but what does that solve? Are people genuinely worried that if they donate to some FOSS platform that their dollars would go to something else? It seems to me like this removes agency from the FOSS maintainer and gives donators more control over their donation, even though it's explicitly designed not to.
Its efficiency also relies on being better than whatever other platform/harness the maintainer is already using. It's limited to whatever the harness the platform provides, and they're taking a 20% platform fee on top. So I have to, instead of taking $10 from a donor, i take $10 worth of tokens, which may or not be spent more efficiently than me just going in with my claude subscription and fixing it, and I get $8 of those to run in a platform I don't control? In what world as a FOSS maintainer would I sign up for this? It just seems strictly worse than just having a platform that can back resolution of issues with real money... which already exists.
Howdy Solar_Fields, these are great questions and I can give you my thoughts on how I feel it's different than cash donations (which are great if you can get them!). I look at this less like patronage and more an exchange of a resource to meet the needs of both parties. I want to support the projects I care about, some I'll give carte blanche, but some I have no connection to and really just want a bug fixed. Rather than fire up my own Claude Code and throw a PR at that maintainer, instead I'm saying, "hey, you know this codebase and can use this resource (tokens) better than me, please fix it with 'free' tokens." The platform fee is really just for AWS costs and is based on modeling, but I'm sure that's not the final form. Does that make sense?
Your reasoning is logical, but fails to pass the bar of "better than or even equal to just literally using some existing platform to attach a $X bounty to some issue I want resolved". There are several popular solutions that exist already to do that, your solution doesn't materially improve there, so what is the value add? It certainly gives the donor more confidence that the issue is being resolved in the way the donor wants it to, but if your problem is to make FOSS maintainer's life easier it doesn't move the needle in that direction, because it gives more power to already demanding FOSS users and less agency to the FOSS maintainer. And even if you solve that problem, does that value add cover a very-steep 20% platform fee?
I think it's a cool idea, don't get me wrong. But it has to be a very good solution to get adopted, like, it would have to significantly streamline the operations of getting bugs resolved by a FOSS maintainer, and I think it's going to be tough for you to try to beat "fire up my favorite agent in my terminal with an already optimized setup and give it this issue that has $X attached" rather than "I have perhaps inefficient token spend from a platform I have no control over and I have to take 20% less of a donation for that privilege"
In other words, I think you've built this solution for donors, and not FOSS maintainers, but really the bottleneck and problem and who you would be selling this solution to is not donors, but rather FOSS maintainers, and that's who you need to solve for if you want a platform like this to work. The donors have the easy job: they throw money at the problem to help it get solved faster. The FOSS maintainers have the hard job: They have to understand, accept the issue, propose a sensible solution, build it out, test it, etc. And your solution just makes it harder for those end users, because now they have two paths in their development workflow to getting issues resolved that they have to maintain, the non paid path and the paid path. So you're significantly increasing the overhead burden on these people and the material gain promised to those end users as the tradeoff is not convincing at all.
I hear you. I'm trying to balance the needs of backers and builders and it's admittedly a tough one. 20% is too high and I'll be adjusting that. The numbers I'm getting from usage today are really helpful in calibrating, so I'm sure I can get something more balanced there.
My hope is that maintainers can offload some busy work to helpful agents and create a more connected community. Likewise, that people otherwise unable to afford the high end frontier models would get to use them for free. Also not something everyone wants, but I know some do. I really do appreciate your feedback and your passion.
Can’t help thinking that if HN had a Black Mirror version (if this isn’t it), this would be one of the ideas in it.
If you like a project enough to donate to them, give them the money directly and let them decide how to spend it. This is just convoluted, weird and vaguely dystopian.
Well then, I'm hoping it's more of a San Junipero/Hang the DJ episode where good stuff happens by the end.
I think there's a trust continuum that culminates at giving money no strings attached, but it starts somewhere else. If my goal (as creator of CleverCrow) is to get more people to support maintainers, then I have to open the funnel wide and connect with where people start (transactionally, I'm assuming).
Ah yes, clearly the one thing I want from my favorite projects is for them to embrace AI coding and immediately deskill such that their value-add or passion for the craft evaporates in the next 3 months.
One of the cool things about code is that you can build stuff out of thin air, basically for free. It's not like woodworking where you have to pay for the wood.
We are moving into a weird time where people are assuming that now we have to pay machines churn out code.
Somehow they packaged up our own ability to think and are selling it back to us. If they can get us to forget how to do it we'll be the perfect customers, dependent forever.
That’s what I don’t get about this whole AI push. It’s a global rug that everyone plans to pull, it’s sold at cost, it already presents major risks, everyone seems to be aware of all that and yet - the consumers and lawmakers are relatively quiet.
Its happening at a global moment of chaotic behavior. When prices are soaring, information and institutions are disintermediated, jobs are hard to come by and layoffs easy, insurance is tied to your job and your kids' well being to your location -- well, most people get real quiet about stuff.
Even then I think a LOT of people are saying something but the narrative mechanism of inevitability is really strong.
Say all the SOTA (remote) LLMs went away tonight, would programmers everywhere suddenly be giving up on programming? Maybe some, but unlikely that most would. It's not so much of a rug-pull if it's an inconvenience at most when it goes away. Maybe some would take some time to readjust (some days/weeks/months?), but then it'd be back to normal again.
Short of outlawing all models, I don’t see a world in which we ever go back to only manual coding.
Who are "we"? Can't go back to what you never left.
I'm not forcing anyone to use tools they don't want. The Amish can continue making their handcrafted furniture, but I'll make sure to work on construction sites that have power tools.
In your mind this sounds like “damn kids, get off my lawn”, and to the kids it sounds like, “I will only ever plow my fields with donkeys, tractors are too complex for something like plowing”.
I'm totally fine with the kids' understanding, for whatever that's worth.
i am still manual coding
not gonna change
still gonna get paid $300k+ for it
No need to outright outlaw models.
Just continue down the path of making it so only the oligarchs can afford to access them, power them, and buy capable hardware to run them locally.
Also helps the oligarchs to have their government assets issue impossible restrictions like “no foreigners can access” that isn’t technically outlawing models for everyone.
> Say all the SOTA (remote) LLMs went away tonight, would programmers everywhere suddenly be giving up on programming?
That's assuming that the code base in question is still well maintained and designed. If this were to happen I would bet that heavily vibe coded things - like Bun and Claude Code - would pretty much immediately hit a wall, where humans just can't comprehend the accumulated mess well enough to make productive changes.
> would pretty much immediately hit a wall, where humans just can't comprehend the accumulated mess well enough to make productive changes.
Plenty of people spent lots of time of their career essentially cleaning up code and refactoring after others. Yes it'd be a lot, but there are really giant spaghetti's out there that been 100% built by humans, then conquered by refactorers.
Well, with AI there is mich more of them, and no more refactoring to reduce the complexity. But only to produce more spaghetti.
> Say all the SOTA (remote) LLMs went away tonight
Most likely that would cause company pulling the rug on their frontier model(s), to fall over themselves. Meanwhile the competition keeps going. Hence no top AI player does that unless forced to.
have you heard of the word Psychosis?
Personally, I think when theres this much money involved the psychosis would only intensify.
I wonder how much money is actually involved?
idiocracy in progress
You can only build it for free if you don't value your time. That is good for personal projects and hobbies but which company can build stuff without costs? There was always an expense to generating code. Plus you have to pay for equipment, electricity, subscriptions to stuff for their staff etc.
I don't really think they are packaging up ability to think abd selling it back to you. There is nothing stopping you from not using AI and even companies, many firms just dont trust AI and don't use it.
But the idea that code could be built of thin air is not true in case of actual businesses.
I really dislike it but the effectiveness of code generation is just too good now. I really doubted it at first but I can churn out 10x the work. Is it the same quality as IC? no but I think that it doesn't matter anymore. How many of us have worked on some 10+m loc monstrosity for years never knowing more than 1% of the codebase? Probably most. Now I can generate that same monstrosity on my own. I've created a 1.1m+ loc and growing trading platform that automates options trades based on live ML edges from research papers. I never would have been able to do that on my own before LLM's. It has full telemetry and comprehensive error logging and testing as well. I'd have been strained to get MVP done before LLMs.
I don't think someone without coding experience can do it though. I've accepted that the new expected output for a single dev will eventually converge to 10x-100x what it is now.
> Now I can generate that same monstrosity on my own.
For what? Why would you need to?
> I've created a 1.1m+ loc and growing trading platform that automates options trades based on live ML edges from research papers.
Sounds like a lot of "if/else" /j
> a 1.1m+ loc and growing trading platform
Why are you here writing comments kn HN? (Instead of counting money and investing it?)
I'm just curious, don't get me wrong.
your comment is pure derision, but its after 4pm markets are closed.
your supposed "honest" inquiry is if you are so smart why arnt you rich yet?
I like the "maintainer stays in control" part, but isn't that also a problem in a way?
The AI provider gets paid, the platform gets paid (20% is a lot in my opinion!), and the maintainer gets more unpaid work: another PR to plan, review, revise, merge, and then maintain... that's a lot of work.
If people are willing to fund an issue, why should that money mainly cover LLM tokens rather than maintainer effort? Or at least, why doesn't the leftover money go to the maintainer instead of back to the donors?
I don't think you understand if we were to just give people money then how are these platforms gonna take a cut of the action?
sounds like you do not support other people that have nothing to do with the code that you like
Tokens are better than money. They let you train your skills to get top 1% on LeetPrompt so you have an entry in the lottery to interview for your next temp position.
While I definitely like the Patreon for Software Builders idea, that's got some moving pieces which take additional legal work. My hope is that could come in time as it would be really cool.
Regarding rewarding maintainer effort, I'm shooting for the value prop of "free AI", this only works if reconciliation is per-phase and liquidity is accessible across as many repos as possible. So if I had each reconcile drain the pool, there would be a lot of stalled work and human intervention required.
That said, there are probably some maintainers that don't want "free AI" and that's okay.
"We can pay you in exposure... And b̶e̶e̶r̶s̶ tokens!"
At first I was like "i want to use ai but dont have the money to burn for api tokens" cool. But then I realized the backers are essentially saying "i have money and could support developers but i choose to give the money directly to a mega corp and skip the human". I recommend you remove the policy of "Whatever the run didn't spend goes straight back to your backers' wallets." and make sure the human behind the wheel gets to eat. Somehow
While we're in the token-equivalent of ZIRP, tokens don't cost what they cost, so there's sort of arbitrage to be had. I have tokens I've been given than I'm not using, but that's not the same as me having been given cash in the first place.
> While we're in the token-equivalent of ZIRP, tokens don't cost what they cost, so there's sort of arbitrage to be had
Do you have a source for this? I believe “personal subscription” plans on OpenAI and Anthropic are likely ran at a net loss or close to it, but all indications elsewhere are that API pricing for these companies and likely Google as well are profitable per API call [1][2]. I would definitely believe that the Chinese players are operating at a loss though if that is what you mean.
[1] https://newsletter.semianalysis.com/p/anthropic-growth-and-b...
[2] https://martinalderson.com/posts/are-openai-and-anthropic-re...
Just finance the project for god sake, why introducing another "gift card" that they sell at cashiers? Let the maintainer spend the money, maybe he/she needs bread, not tokens
Just curious, why is there a login gate before seeing the list of projects that participate in the platform? Usually similar donation(?) websites list those publicly for better visibility and less friction.
Probably an oversight on my part. I was thinking that backers would find out about CleverCrow through the project maintainers so the public pages are repo specific.
As an aside, when you do login, CleverCrow shows projects that you've starred on GitHub to help find things you might want to support.
Oh nice! Does it also require my bank's password?
What if the maintainer doesn’t want to implement a particular feature at the moment?
I suppose this is the most common scenario - I doubt features are not getting implemented because maintainers are lacking tokens.
100% expected, in fact, desired (maintainers should be/are in charge). My goal is to create enough surface area that backers see their tokens going to many of the things they care about, not necessarily all.
Really interesting solution to the AI PR problem. Keeping the maintainers in the driver's seat for issue prioritization is definitely the right approach.
How are you handling the token allocation under the hood, is this managed via a GitHub App integration, and can backers target specific issues or just the repo as a whole?
I've got a couple good docs in the footer of www.clevercrow.io that will answer the token stuff. GitHub app integration for all repo interaction. Yep, backers can target specific issues or entire repos, their choice.
Perfect, I'll check out the docs. Letting backers target specific issues is a great feature. Thanks for the reply, and best of luck with the platform!
I always thought that the donate tokens thing would be done by sending some tokens from your personal sub to the maintainer's pool with some sort of proxy for tokens with rules, in a more direct way without doing it in cash, but yeah that's where the sweet fees live.
If this gets any traction, the "share tokens with a friend" could be good PR for the labs, instead of buy me a coffee, buy me a clanker.
Better yet: give them cold hard cash instead of what is arguably monopoly money for many OSS devs. Ironically this is something GitHub made "easy" with sponsorships several years ago.
How do you ensure that funds ear-marked for a donor-specified issue goes toward that issue and not something else?
You don't sponsor people or projects to complete specific issues or build specific features in the first place. Sponsorship is a reward and token of appreciation for doing good work.
Some don't mind doing the overall reward and appreciation thing. And some just have that particular issue that they want handled so the project works - better - for them. Both cases are valid.
Yes, and donation/sponsorship is not the tool for the latter.
What matters is if it works out (gains traction) or not.
I'm not sure how that's related to your initial question of "How do you ensure that funds ear-marked for a donor-specified issue goes toward that issue and not something else?"
If you want that, negotiate a contract.
The point is to get something funded. That's the goal. And negotiation doesn't scale. You can bet that nobody will negotiate contracts with 20 maintainers if they want particular features/fixes for 20 different projects that they're using. Otherwise it would be a thing today.
But instead we have these attempts and stopgaps, some of which have had some success here and there. This is something else in that pool making it easier to fund stuff, and if it gains traction than we'll know that it's serving its purpose. I think it has good potential.
I don’t think companies are donating to 20 different maintainers in the hopes of getting 20 different specific features implemented, either.
I wasn't thinking about companies at all. Companies tend to only contribute to well-known projects, which leaves a very long tail.
I'm thinking about the far more random regular people using random projects that may not be that popular. Like I just did a rough count of the OSS apps on my phone: 60+. Of those there are 5 that I can immediately think of improvements that I'd like. One of them I've actually made improvements to already because it was a dead project that I really like and so had to revive anyway. And I used Claude to work on it as I'm not a mobile dev; been planning to republish but haven't gotten around to it. I won't bother to check my laptop but there's even more there.
Now imagine if I (and others interested in those projects) could contribute to small funding pools for various projects of interest, with the assurance that said funds go to that feature/fix which is of interest or gets refunded. I think that'd result in the general OSS support needle being pushed that much further over time.
Then you offer to pay the maintainer their consulting rate to do it if they are willing.
That's one way to go about it, but doesn't exactly work when one has targeted requirements in 20 different projects.
You hit up the maintainer and negotiate a deal for that?
If all you’ve got is relative pocket change they probably aren’t going to agree but if you put real money behind it and it doesn’t go against their vision of the project then most people would be willing to accept actual contracting work to expand their project.
Sounds like a lot of trouble to go through, vs just sending some funds to a wallet with the assurance it'll go where you want it to or return to you.
You actually hire a developer to work on that issue and not something else.
Pretty much what this ensures. Just that the "developer" is a LLM agent.
Yeah except that the agent cannot be held accountable if its fix is crap, or if it went off-track mid-fix.
Pretty straight-forward solution: donors or maintainers don't use the service if it isn't good. But the maintainer is "accountable" (to the extent one maintaining a OSS project is) as they're the one actually providing the prompts and doing QA.
I think I've read from a few different sources that the Claude Code $100-$200/mo plans are subsidized so hard that it's basically $2k-$8k/mo in "would-be" equivalent API token usages.
This kind of makes sense in that space while the subsidies (if true) last?
Unrelated, "tokens" feels very like... back-then blockchain to me. All the craze.
Or the API is massively overpriced. What Anthropic/OpenAI are charging for tokens says almost nothing about their actual costs, just what people are willing to pay.
It’s just an obvious example of market segmentation by charging enterprise customers many times more than personal users while selling the same product
Yeah, rising token costs definitely played into my thinking too. I want builders to be able to use the best models and it seems like they are getting more expensive. But maybe local models will get there?
Shameless plug, I submitted a similar thought in here the other day. https://news.ycombinator.com/item?id=48503555
I like your approach of pooling resources around specific issues. That seems a practical missing piece for aiding the maintainers.
Congratulations, you've fulfilled one of ThePrimeagen's predictions! (A donation platform for AI tokens)
NEW ACHIEVEMENT! ;)
God damn it, Donut.
Is this just basically a bountysource? or are there ways to give projects tokens without just sending them money?
It does seem like a worse version of a FOSS donation platform that uses regular old money. I guess one advantage is that you can ensure your money goes directly to using AI to solve specific problems on the codebase, but what does that solve? Are people genuinely worried that if they donate to some FOSS platform that their dollars would go to something else? It seems to me like this removes agency from the FOSS maintainer and gives donators more control over their donation, even though it's explicitly designed not to.
Its efficiency also relies on being better than whatever other platform/harness the maintainer is already using. It's limited to whatever the harness the platform provides, and they're taking a 20% platform fee on top. So I have to, instead of taking $10 from a donor, i take $10 worth of tokens, which may or not be spent more efficiently than me just going in with my claude subscription and fixing it, and I get $8 of those to run in a platform I don't control? In what world as a FOSS maintainer would I sign up for this? It just seems strictly worse than just having a platform that can back resolution of issues with real money... which already exists.
Howdy Solar_Fields, these are great questions and I can give you my thoughts on how I feel it's different than cash donations (which are great if you can get them!). I look at this less like patronage and more an exchange of a resource to meet the needs of both parties. I want to support the projects I care about, some I'll give carte blanche, but some I have no connection to and really just want a bug fixed. Rather than fire up my own Claude Code and throw a PR at that maintainer, instead I'm saying, "hey, you know this codebase and can use this resource (tokens) better than me, please fix it with 'free' tokens." The platform fee is really just for AWS costs and is based on modeling, but I'm sure that's not the final form. Does that make sense?
Your reasoning is logical, but fails to pass the bar of "better than or even equal to just literally using some existing platform to attach a $X bounty to some issue I want resolved". There are several popular solutions that exist already to do that, your solution doesn't materially improve there, so what is the value add? It certainly gives the donor more confidence that the issue is being resolved in the way the donor wants it to, but if your problem is to make FOSS maintainer's life easier it doesn't move the needle in that direction, because it gives more power to already demanding FOSS users and less agency to the FOSS maintainer. And even if you solve that problem, does that value add cover a very-steep 20% platform fee?
I think it's a cool idea, don't get me wrong. But it has to be a very good solution to get adopted, like, it would have to significantly streamline the operations of getting bugs resolved by a FOSS maintainer, and I think it's going to be tough for you to try to beat "fire up my favorite agent in my terminal with an already optimized setup and give it this issue that has $X attached" rather than "I have perhaps inefficient token spend from a platform I have no control over and I have to take 20% less of a donation for that privilege"
In other words, I think you've built this solution for donors, and not FOSS maintainers, but really the bottleneck and problem and who you would be selling this solution to is not donors, but rather FOSS maintainers, and that's who you need to solve for if you want a platform like this to work. The donors have the easy job: they throw money at the problem to help it get solved faster. The FOSS maintainers have the hard job: They have to understand, accept the issue, propose a sensible solution, build it out, test it, etc. And your solution just makes it harder for those end users, because now they have two paths in their development workflow to getting issues resolved that they have to maintain, the non paid path and the paid path. So you're significantly increasing the overhead burden on these people and the material gain promised to those end users as the tradeoff is not convincing at all.
I hear you. I'm trying to balance the needs of backers and builders and it's admittedly a tough one. 20% is too high and I'll be adjusting that. The numbers I'm getting from usage today are really helpful in calibrating, so I'm sure I can get something more balanced there. My hope is that maintainers can offload some busy work to helpful agents and create a more connected community. Likewise, that people otherwise unable to afford the high end frontier models would get to use them for free. Also not something everyone wants, but I know some do. I really do appreciate your feedback and your passion.
Can’t help thinking that if HN had a Black Mirror version (if this isn’t it), this would be one of the ideas in it.
If you like a project enough to donate to them, give them the money directly and let them decide how to spend it. This is just convoluted, weird and vaguely dystopian.
Well then, I'm hoping it's more of a San Junipero/Hang the DJ episode where good stuff happens by the end.
I think there's a trust continuum that culminates at giving money no strings attached, but it starts somewhere else. If my goal (as creator of CleverCrow) is to get more people to support maintainers, then I have to open the funnel wide and connect with where people start (transactionally, I'm assuming).
Where can I invest and trade in these ”tokens” you speak of?
I had this idea! Happy to see someone actually made it.
Give them money.
Tokens don’t come with CV clout which is what the vast majority of the slop PRs are really about.
interesting idea
Ah yes, clearly the one thing I want from my favorite projects is for them to embrace AI coding and immediately deskill such that their value-add or passion for the craft evaporates in the next 3 months.