That's also a reason the big labs stopped. Publishing is most valuable to people who have no other way to get the attention of smart strangers. Once you can hire nearly anyone and everyone already returns your calls, the main remaining effect of publishing is to tell your competitors which things worked.
This is what happens to every field as it turns from a science into an industry. Chemists published freely until dyes started being worth money, and then the interesting work moved into company labs and stopped coming out.
Which is a very ironic and selfish situation when your business model dependent mostly on model training based on available published data, academic and non-academic.
That has nothing to do with anything. If you publish a copyleft paper, that doesn't compel someone who makes a product based on your paper to publish more papers.
The GNU RPL (research public license), a viral knowledge license. By reading this paper you are legally obligated to openly publish all vaguely related future research that you perform.
Your RPL wouldn't be enforceable. Copyright doesn't deal with abstract ideas passing through people's minds. Even the GPL is kind of in a gray area because the virality feature and its definition of "derivative work" have never been tested in court, to my knowledge. Maybe under contract law, no idea. If nothing else, I'd love to hear a verdict.
Well it was a joke and is obviously quite silly but I believe it would be enforceable to the extent that the licensor could terminate the agreement and sue for damages. If I can agree to pay you not to talk about something (ie an NDA) or not to work in a field (ie a non-compete clause) then why can't I pay you to be required to publish all future work you do in a given area? ("All future work" might well be overly broad; you'd likely have to restrict it to only a few years but I'm no lawyer.)
I don't believe the GPL is in a grey area. When you license a work all sorts of strings are commonly attached. Rather IIUC no one has gone out of the way to test the GPL largely because it is clearly within bounds, plus any violation has the potential to be a PR disaster since violations are quite literally examples of looting the commons.
Meaning what? Claw back the ideas from people's minds? You can terminate the agreement in the sense that you revoke access to the paper, but presumably the person you find in breach has already used the research for something that you find them in breach for. You're kind of closing the gate after the horse has bolted.
>sue for damages
I honestly have no idea what damages you could claim from not publishing research. I think you would need to set a value ahead of time on the agreement.
>If I can agree to pay you not to talk about something (ie an NDA) or not to work in a field (ie a non-compete clause) then why can't I pay you to be required to publish all future work you do in a given area?
Not sure why you added the word "pay" to your clauses, but anyway. The reason is that the existing contracts have well-defined boundaries. An NDA stops you from divulging a very specific set of information. A non-compete clause stops you from working in a very specific field. Your proposal has an undefined reach. What counts as research? What counts as "related"? It would seem that if I agree to such a contract, my entire life, both private and professional, after reading the paper is covered by the contract, and anything I do might come under scrutiny. There's never a point when I can go off-duty. "What's that? You read my paper on compression and were working on a side-project that uses compression? Gonna have to see some publication on it."
>Rather IIUC no one has gone out of the way to test the GPL largely because it is clearly within bounds
No, it's because the status quo is convenient and no one wants to be the first guinea pig. It's definitely not obvious that the terms are legally valid, but it's ambiguous enough that people don't want to test it.
If something is so obviously wrong then perhaps take a minute to consider that your interpretation isn't what the other party intended?
If I pay you not to do something and then you breach the contract I can terminate the agreement and seek damages. Ditto if I pay you to repeatedly do something and then at some point you fail to do it. So if I pay you a recurring fee to publish all your research on a given topic and then you fail to make good on that I can seek damages, right? Now what if I paid you a lump sum up front? Now what if I licensed a patent to you in place of that lump sum? What if instead of a patent it was the right to make use of a piece of software?
> I honestly have no idea what damages you could claim from not publishing research.
Aside from whatever was stipulated for breach of contract I expect it would be extremely situational. I agree that you'd probably want to stipulate a penalty ahead of time. The original idea was a two sentence joke after all ...
> Not sure why you added the word "pay" to your clauses
Because contracts involve consideration and payment is easy to understand and reason about. FOSS software licenses obviously substitute "right to use the code" for "payment".
> A non-compete clause stops you from working in a very specific field. Your proposal has an undefined reach. What counts as research? What counts as "related"? It would seem that if I agree to such a contract, my entire life, both private and professional, after reading the paper is covered by the contract, and anything I do might come under scrutiny.
This is incredibly contrived. You could ask the equivalent about a non-compete. The reach is whatever is defined in the contract that both parties agreed to. If the contract stipulates something overly broad then possibly a judge would invalidate it. This is business 101.
> It's definitely not obvious that the terms are legally valid, but it's ambiguous enough that people don't want to test it.
An assertion entirely without evidence from my perspective. I'm going to assume that all the lawyers who have sure left me with the impression that it would be a bad idea to violate it know what they're talking about.
>So if I pay you a recurring fee to publish all your research on a given topic and then you fail to make good on that I can seek damages, right? Now what if I paid you a lump sum up front? Now what if I licensed a patent to you in place of that lump sum? What if instead of a patent it was the right to make use of a piece of software?
Uh-huh... This doesn't answer my question of what terminating the agreement of access to the paper does, besides what I've already said. You've licensed to me access to a paper under certain conditions. I've breached the conditions, therefore you terminate the agreement, therefore you revoke access. Am I missing anything?
>FOSS software licenses obviously substitute "right to use the code" for "payment".
Hence my question. The hypothetical license/contract under discussion is about access to research results, not about a monetary transaction.
>This is incredibly contrived.
Well, the idea of viral abstract ideas is stupid, so it forces me to give contrived examples.
>I'm going to assume that all the lawyers who have sure left me with the impression that it would be a bad idea to violate it know what they're talking about.
What point do you think you're making? Something can be ambiguously (but not certainly) risky and a bad idea to do. I have two coins, one with two tails and the other a fair one, and I offer you to gamble everything you own on one of these coins of your choosing, or walk away. I assume you wouldn't pick the unfair one. Therefore if you would rather walk away than gamble everything you own on the normal coin, the toss actually has a 100% chance of you losing?
> I've breached the conditions, therefore you terminate the agreement, therefore you revoke access. Am I missing anything?
You're missing the part where I seek punitive and actual damages under the terms of the contract. No different than violating an NDA - I paid you a lump sum up front, after a while you breached the contract, the agreement is null and void, what's the consequence?
> Well, the idea of viral abstract ideas is stupid, so it forces me to give contrived examples.
On the contrary, presumably it was because you lacked the ability to roundly refute anything I had put forward. Otherwise I assume you would have done so.
> Therefore if you would rather walk away than gamble everything you own on the normal coin, the toss actually has a 100% chance of you losing?
But in this analogy it is you baselessly making that claim. There's every expectation that it's a fair coin, many experts have carefully inspected it and authored opinions on it, and some have put forward theories that it slightly deviates in one direction or another. Then you show up and confidently assert without any evidence that there's some wild deviation from fair, hand waving that you would have proof if only someone wanted to bother testing it.
Out of curiosity, what is it that has you so bothered about the idea of viral licenses? What do you find so objectionable about attaching arbitrary terms to contracts?
You keep handwaving away any requests for something concrete to discuss. If it’s so easy to construct a viral license for access to science, just give an example of it.
> Out of curiosity, what is it that has you so bothered about the idea of viral licenses? What do you find so objectionable about attaching arbitrary terms to contracts?
The only example of them is open source that is routinely violated because it’s so weak in practice. LLMs reproduce it frequently with no attribution and nobody is successfully suing them over this.
>I seek punitive and actual damages under the terms of the contract.
Right, so you're agreeing with my original interpretation of "termination".
>You can terminate the agreement in the sense that you revoke access to the paper
The termination and the lawsuit for damages are separate events. Why did you accuse me of being purposefully dense if you're agreeing with me?
>But in this analogy it is you baselessly making that claim.
We're both making baseless claims. I said A, and then you said ¬A, and neither one has backed anything up. I don't even necessarily believe you talked to any lawyers, I just granted it for the sake of argument because the statement you made was so weak that I didn't need to cast doubt on it.
Also, in case you missed it, the fair coin in the analogy represents the case where the GPL might not be fully valid, not the other way around. If the coin is unfair then the GPL is fully and obviously valid, and if you get sued and you did infringe it then you're certain to lose.
>Out of curiosity, what is it that has you so bothered about the idea of viral licenses? What do you find so objectionable about attaching arbitrary terms to contracts?
Since I'm not the topic of discussion, I won't answer these questions beyond saying that I didn't opine on viral licenses.
This reads like a company requiring you to hand over your first-born kid in the TOS. I would happily be the test case and violate your license if you want to sue me!
To paraphrase your last paragraph, an idea that hasn’t been tested is either perfect, or just bad enough that no one wants to test it, as, as you say, testing it is would be at least somewhat inconvenient.
but you patent something requires proving that you have done something novel. You're effectively asserting that the thing you invented is new and different enough to be not a derivative of a previous patent.
sure to build a new system, you might need other patents. but a pateent it's self should be unique enough to defend. Which means its copyrightable as its own thing
An idea can be both novel and derivative at the same time.
You create idea B.
It absolutely builds upon idea A, and is in some sense derivative.
But your idea B introduces a lot of new novel concepts on top of idea A that makes it novel.
An idea can absolutely be both novel and derivative at the same time. Most of the best ideas are.
For a hypothetical example, an innovative Linux filesystem requires having the Linux OS, or similar, as a base to operate. In some sense, it is derivative of that prior work. But the new ideas extend the old ideas enough to be novel.
From: rms@prep.ai.mit.edu (Richard M. Stallman)
Date: 25 June 1986 at 18:45:28 GMT+2
To: a2deh@ai
I figured out what a copyleft is.
It's the sort of thing that I put on GNU software and manuals.
The left-wing version of the copyright that rightists use
to exploit the masses.
But copyrights are not patents, and the opposite of "patent" is "latent", so we need a Latent Office that protects inventions from the people who claim to have invented them, and issues warnings to patent trolls:
LATENT (L)
All rights unreserved.
This invention may already exist, but hasn’t become obvious yet.
Latentleft — publish the invention, conceal the monopoly.
Latent Pending — the idea is out there somewhere.
Latent Troll — remains dormant until an industry becomes profitable.
The patent system is a pretty useful compromise here. The applicants are forced to disclose all knowledge required to practice the craft. They get a 20 year monopoly in return.
I realize everyone always wants everything for free, but if you want someone to put the energy into writing up their invention you need to give them an incentive.
I don't understand why from a legal perspective, but it seems like over the last decade or more the tech industry stopped caring about patents. Startups used to talk about their parents and large tech companies would encourage you to file. While people say software patents make no sense, when I've seen cases involving EE, ME, and software patents, the software patents are usually the most likely to hold up in court and get damages.
Everyone says this in the abstract but to concrete examples they shug and say, of course that approach doesn't work, they did X, Y, Z wrong, they should have given it more effort, it could have worked if done properly / this can obviously never work, everyone knew already, it's nothing new etc.
It would need to be beautifully done science - indeed the level of rigour required to confidently show no effect is probably quite a bit larger than the level of rigour required to confidently show a large effect.
And that's probably one of the main reasons it's not done more often.
Isn't that exactly the point? If in my conception that concrete example eliminates that single branch, then I can try the others, and only the others. Or confirm the negative. Which is also valid work.
In general, your point stands for armchair researcher.
No, good researchers also dismiss such things, because most research out there is crap. If you reward negative results, you'll have a flood of them, and the reason that they are negative will mostly look like it's incompetence (and mostly will be, but even the small not-incompetent will be hard to distinguish).
It's asymmetric. If you make something work, beat a benchmark, invent a drug that works etc, the exact way you arrived there has some leeway. In the end, the thing worked out so it's worth knowing about.
If your project failed, there can be a million reasons for that, and it's not necessarily that the initial hypothesis or initial idea has been refuted. E.g. in ML, your model didn't learn the task. Okay, there could be a million knobs (hyperparameters) that you set up wrong, or you implemented it wrong, or you should have just added learning rate warmup, or this or that, a million things possible. People fail all the time at projects that others then manage to do later on. Science is not like simply asking the universe some clear question and getting a clear answer. It's a very messy process and even professionals are not super great at it, or at least they simply cannot afford to put so much effort in each single project to make it absolutely airtight such that the failure to make it work can be a legit refutation of the main idea.
> I wish more people would publish the things that didn't work.
This would be a pointless endeavour. One of the most basic mantras of science is "absence of evidence is not evidence of absence". So just because something didn't worked out for you that doesn't mean it doesn't work out for others, or even yourself in the future.
You are talking about a different thing, i.e. you have slipped in a level of abstraction that was not there before:
In 'searching a path from A to B in a maze' language:
The original statement was: (1) The branch to the left from A is a dead end.
Your interpretation: (2) There is no path from A to B.
(1) is still very useful (reducing the wasted effort) for those trying to find a path from A to B. The OP's point is that in the current environment only positive results are rewarded (I found the path from A to B!), not the negative ones like (1).
> The original statement was: (1) The branch to the left from A is a dead end.
This is where you get things wrong at a very basic and fundamental level.
Just because you failed to explore branch A, that does not mean it is a dead end. It just means you came up empty.
That is why science is based on observations and theories: it is based on building up on ideas and what works and can be proven. Otherwise you will left with useless papers such as "Bicycles are a dead end because I tried to ride one and I fell".
You are assuming incompetence on the scientist saying that the branch from the left of A is a dead end.
While nobody is perfect, there are numerous perfectly valid scientific negative results. You know, there exist things like impossibility proofs in mathematics and computer science. There are equivalents in other sciences (e.g. if X was true, that would lead to Y that is easily observable and clearly not observed). Sometimes that implication has assumptions that might change once the technology/society changes, other times it holds true regardless.
Unicycles are a dead end as a practical transportation, because the bicycles have them beat in every way (except portability).
A scientific result would be much more along the lines of 'Bicycles without gears have limited applicability, especially in hilly terrain.'
There are different types of research. Negative AI results are not like Michelson-Morley experiment in physics that can positively establish the lack of something. It's more of a craft, the "make it work" style of thing. Give the same task or goal to two different teams, you will see massively different outcomes. Simply failing to make it work is not conclusive.
To make such a negative result acceptable in AI, you'd have to have some clear reason why you think that your particular setup should produce the result you want, that exact configuration and architecture, dataset etc. There are countless projects in AI that fail. And it's not clear at all that it refutes any abstract hypothesis. It's a get-your-hands-dirty field. It can make or break a project whether someone has that tacit knowledge, that black magic experience to know how to properly do the project.
People can generate extremely many ideas. You'd need to convince me that your idea (among a million others that people are trying each day) is so significant that its failure is in itself interesting. If you were to review for AI conferences, you'd see the flood of papers that claim to achieve 0.5% or 1% improvement on some benchmark. Now imagine that they didn't even have that to show for it. It got worse by 2% after trying their random idea. Who cares then? Even the +1% with a random idea is quite annoying to accept. But if their random idea really made something work much better, I will at least have some reason to want to see what may be going on there, there can be some signal. With negative results, it's very uninteresting.
I did read OP's post much more generally than only about AI research. There are many research areas where (properly done!) negative results would be very valuable, but are not properly encouraged.
I do agree that in a new area with too many degrees of freedom (and yes, AI research is one of those), negative results (especially poorly done) are of limited usefulness.
> Chemists published freely until dyes started being worth money
Notably this is exactly what patents are intended to combat. And while US IP law is clearly very broken it does at least largely accomplish this stated goal. Much (but certainly not all) industrial chemistry has made it into the academic literature.
Not that the same logic necessarily applies to AI research (ie algorithms aka math and their implementations). And I'm actually happy about that because the cost of doing the research is so much lower. There's a long list of reasons that the average person living in a residential area can't do industrial chemistry as a hobby.
To your dye example, yttrium indium manganese blue was the first commercially viable inorganic blue pigment discovered in ~200 years, is the only known environmentally safe one, and was openly published in the literature. It's also under an exclusive license. (TBF though unless the chemical is unusually difficult to synthesize not publishing would be rather pointless in this day and age given the utterly absurd capabilities of modern analytical techniques.)
It depends on the field, and to some extent on the particular patent officer assigned, but one of the failings of the US patent system is many patents don't actually provide enough details to reproduce the result. And even if they do they don't (usually) include ancillary information like the discovery process, properties of the invention etc.
>> To your dye example, yttrium indium manganese blue was the first commercially viable inorganic blue pigment discovered in ~200 years, is the only known environmentally safe one, and was openly published in the literature. It's also under an exclusive license.
Gee, I wonder what was wrong with the previous blue pigments and why it was so important to have this one under an exclusive license.
Cobalt blue is a blue pigment made by sintering cobalt(II) oxide with aluminium(III) oxide (alumina) at 1200 °C. Chemically, cobalt blue pigment is cobalt(II) oxide-aluminium oxide, or cobalt(II) aluminate, CoAl2O4. Cobalt blue is lighter and less intense than the (iron-cyanide based) pigment Prussian blue.
Cobalt poisoning sounds scary, but I see nothing on the linked Wikipedia page that would present a risk of accidental consumption of medically relevant amounts of cobalt, whether in one sitting or through prolonged exposure.
Though the mention of Bolesławiec makes me worried a little; my wife loves their designs and we have a bunch of plates and bowls and such from them, and we serve food on/in them regularly...
> my wife loves their designs and we have a bunch of plates and bowls and such from them, and we serve food on/in them regularly...
I don't think you need to worry about that. The pigment is already reasonably benign when in solid form; once embedded in or under a glaze I'm not aware of anything that would suggest it carries any health risks whatsoever.
That said I'm unclear how safe direct exposure to the pigment itself is (such as when suspended in a liquid for painting). There's not a lot of data available that I could find, and of course due to having a unique crystalline structure it won't necessarily have the same properties as the component products, however cobalt(II) oxide itself is extremely hazardous which is at least cause to be cautious.
Other than manufacturing safety primarily the new pigment is just incredibly vivid.
>> That said I'm unclear how safe direct exposure to the pigment itself is (such as when suspended in a liquid for painting). There's not a lot of data available that I could find, and of course due to having a unique crystalline structure it won't necessarily have the same properties as the component products, however cobalt(II) oxide itself is extremely hazardous which is at least cause to be cautious.
The consensus is that artists' colours are hazardous waste that needs safe disposal. You can find instructions how to do that in many arts supplies' sellers sites and the like, for example one I've used:
The comment about not licking your brushes is a reference to the common practice of licking them in the wargaming miniatures painting community (you know, Warhammer and all that). You can find videos of influencers explicitly telling you that the best way to shape the tip of your brush or remove excess water is to lick it. For example:
If you ask for a rationale it's usually that wargaming and generally hobbyist paints are non-toxic, but the truth is that hobby paint manufacturers never list their paints' ingredients so there is no way to know. And of course there's different degrees of toxicity, not everything needs to be cobalt(II) oxide-level toxic to hurt you if you consume "large enough" quantities of it; the question is what that "large enough" means and without knowing the substance, there's no way to know.
The startup landscape has also changed noticeable compared to 5 or 10 years ago. A team of smart/credible people could get funding for an idea and build a product + publish, knowing there was a six month lead time for anyone to copy them and ship. These days, the barrier to ship code is zero. People can copy your business over a weekend, so there is much more urgency to establish product market fit and build a “moat”. Publishing timelines are now at odds with the pace of go-to-market and VC funding timelines.
I think in theory this is true but in practice I don't see loads more good apps or products. There's a paradox here I think. I just don't see loads of quality competitors popping up I actually think it makes building something harder because the barrier to entry just gets higher somehow.
I think the revolution is more of a gradual thing than an overnight one. New features and more complex apps in days or weeks instead of months. I think the world will be a much different place in 10 years than it is now, but it will be a gradual change rather than the overnight switch flip some AI maximalists would have you believe.
At my company, AI has had a huge positive impact in helping us manage technical debt that we just didn't have time to deal with before. This simple thing will have a compounding efficiency and profitability effect over the next several years.
The other thing that we have found is that AI coding abilities can crank out features faster than we can provide human support to our customers using said features. We have feature PRs that have been open for months without being merged because our company does not have the human bandwidth to provide support for them. In case you can't tell, we prioritize human support and individual connection with our customers. We actually value them.
As with all human endeavours, as the tooling gets better the expectation for a launch/product/feature increases dramatically too. I probably do 3-5x the features I used to before AI, yet there's still more to do. I'm not running out of work but making incredible progress on features that save the place I work at weeks and weeks of time every month which frees people up to do more of the other things. It still takes time to build and check and test things though.
it goes to show that, despite what we want to think, developers just aren't that important. The code is maybe 10% of what it takes to make a business, even a software tech business, a success. So whether it's human generated or LLM generated, the code is just one slice of the pie that has to be complete and effective to make a product successful in the market.
Over the weekend, I analyzed public YouTube tutorials of 4 SaaS products in the same domain. Then I asked the coding agent to define an API where they converge — prior art. Today I'm working on creating dashboards that won't touch anyone's copyright or IP.
A couple of days ago, something interesting happened. The agent works in an iteration loop where each iteration is an endpoint. I let it run overnight. In the morning it was still cranking away even though it had finished all the API endpoints. It had found a changelog from one of the companies listing every single feature and bug, and decided on its own to implement every item as an iteration.
We are 2 to 3 months away from coding agents replicating solving the edge cases of most SaaS applications.
There are other reasons. One is for career development of your researchers. Another is to be good citizens, in good standing, in the community of scientific researchers. Another is to flex on people and buy class and respectability - to comport oneself as the “old money” does.
On the contrary, I've found that stuff in the mining industry is actually well published, to the point of replicability (one of our projects is essentially copying the patented design of a well known mining technology company, albeit with some modifications, as the patent will expire next month).
Perhaps because the real moat in the industry isn't the technology per se, but the concessions.
Such a biased and utilitarian view. The goal of publishing is to communicate your findings to the scientific community for the sake of advancing our knowledge as a society.
Of course you might want to keep _some_ of this knowledge as trade secret, but then don’t claim you are doing research.
You are advancing your shareholder’s interests, not that of the broader society.
This academic publish and industry doesn't is an incorrect view. Academics also hold off from publishing groundbreaking science in order to protect their research edge. You shouldn't see this as industry bad as driven by profits and academia is free from career incentives and money.
This is what happens to every field as it turns from a science into an industry. Chemists published freely until dyes started being worth money, and then the interesting work moved into company labs and stopped coming out.