pona-a
4 days ago
I do find using AI in a clean-room reverse-engineering project amounts to shooting yourself in the foot: you're sleep-walking into a legal minefield in an already legally gray field, and also risk over-relying on the AI's trial-and-error instead of developing a solid intuition of its functioning yourself.
At the same time, if we're being honest, Synaptics will have sent the C&D no matter what. They are an anti-consumer company with more lawyers than engineers, so of course they'll use all their resources to bully any competition, especially the most vulnerable.
adrian_b
3 days ago
The problem is not in reverse engineering, but in the existence of bogus software patents.
The threats of Synaptics are dangerous only because they claim patent infringement, for which it does not matter whether reverse engineering was clean or not.
echelon
3 days ago
I'm not sure patents make sense anymore in the post-AI era.
throwaway85825
3 days ago
Patents should require a prototype demonstration. Would clear up a lot of the patents for preexisting ideas.
tanseydavid
3 days ago
Software patents or all patents? Serious question.
DaSHacka
3 days ago
All patents. Rent seeking behavior around loose ideas of all things should be destroyed. The western world has an inherent desire to innovate, you do not need to provide overbearing protectionism to enable motivated individuals to innovate and invent technologies and products to progress society.
kube-system
3 days ago
Patents don't exist to encourage companies to innovate -- they exist to encourage companies to share information about their innovations.
I think you're exactly right that the western world would continue to innovate without patents. I think what you're missing is that the rent-seeking would continue as normal, simply by using trade secrets, obfuscation, services-models, etc.
dbgobrrr
3 days ago
But that's fine. I would prefer having the specs - that helps a lot in writing high quality drivers. That said, open source relied a lot on reverse-engineering in the past - not to mention that reverse-engineering is much easier these days. But not being able to e.g. merge a driver into the Linux Kernel because of a software patent is an actual, real blocker. Same thing with patent-encumbered codecs.
Patents was supposed to be a tool to enable competition, but does the reverse now - and it is holding as back as society. We should move away from it. Especially and specifically when it comes to vague, loose ideas just like the GP post says here.
StevenWaterman
3 days ago
But companies can already do that. So presumably patents are more effective at preventing competition?
kube-system
2 days ago
Of course, the existence of patents does not force everyone to use patents. But without them, those who would have used a patent would be forced to resort to secrets to protect their ideas.
rwz
3 days ago
Serious question: How does this worldview survive a collision with a stereotypical cancer-cure argument? Like, a pharmacological company needs to invest billions into R&D to invent a drug curing cancer. Without any kind of patent mechanism they will never be able to justify this kind of investment, cause the minute the release their drug, others can reverse engineer it and start selling generics of it for fraction of the cost because they never had to pay upfront R&D.
patrakov
3 days ago
And the minute they don't release their drug, others will use AI to make their own version, even better.
pona-a
3 days ago
Who's going to pay for collecting all the actual medical data, AI Jesus? Maybe the universities? Great, we agree R&D is best handled by public institutions funded by the taxpayer. But something tells me you wouldn't agree with me...
I swear most AI booster arguments can work about as well if we replace AI with the Holy Spirit.
(Apologies if this was sarcasm, I genuinely can't tell these days...)
shanoaice
3 days ago
Patents had good intentions, but they are long gone. Nowadays all patents are more or less rent-seeking tools. Software patents are just especially troublesome, I think FSF had some articles on it.
fulafel
3 days ago
They keep barrier to entry high.
PowerElectronix
3 days ago
They never did
marklar423
3 days ago
Why not?
ajb
4 days ago
Clean-room reverse engineering involves two teams: one with access to the thing being reverse engineered, who write documentation of it, and the one in the clean room who do the implementation based only on the documentation. There's no reason you could not use AI for the first part, as the documentation is then evidence that you are only looking at behaviour and not source code.
Using it inside the clean room is more of an issue.
ndiddy
3 days ago
Clean-room reverse engineering gets around the copyright on the software. What the Synaptics legal people are arguing is that the code is infringing on two of Synaptics's patents, not on its copyright. Even if you have a completely new implementation of the code, it's still patent infringement if the code implements patented inventions.
exceptione
3 days ago
As we are talking about software patents, these would probably have no meaning in Europe. So this might be a route to escape Synaptics.
voakbasda
4 days ago
If you use have the first team use one provider to write the spec and a completely different provider for the implementation team, I would expect you would be covered, but it probably will take the first players in this game to establish that precedent by having their cases tried in court.
pona-a
4 days ago
Why do providers even matter? They are nearly identical weights produced from mostly overlapping corpa. You can reduce it to two instances but that only shows how fishy this copyright-laundring is.
ajb
4 days ago
The reason it's not "fishy laundering " is fundamental to copyright: copyright protects expression, not ideas. That's why in cases where expressing the same idea would result in the same text, the courts have ruled that the expression is not copyrightable rather than that this gives a monopoly to the first person to copyright it. (NB I am not a lawyer, etc)
kube-system
3 days ago
Yeah, but the whole reason clean room RE uses two different teams of people is so you can demonstrate that the second team is not simply reproducing that the original expression -- it would have been impossible for them to do so because they had never seen it.
You lose that if you use an LLM which was trained on the original expression.
To keep the clean room clean, you'd need to use an LLM which you could prove had never seen the original.
pona-a
2 days ago
As others pointed out, it's hard to say how much copyrighted material the model already contains in the weights. The bigger problem is the legal system isn't a logical system, it's a social construct. The reason why it allows cleanrooming is its how it promotes understanding, creates labor, and is generally too expensive for your entire product to be wholesale copied the moment you publish it. There's an underlying social contract the legal contract only exists to mirror.
It's my personal opinion Synaptics should go bankrupt along with its cancerous mountain of patents, and the copyright system itself reformed to finally end its reign of terror of over both art and engineering, this requires coming up with a better social contract than "I got mine fuck you", or the entire world will turn into Temu crap.
kube-system
2 days ago
> The bigger problem is the legal system isn't a logical system, it's a social construct. The reason why it allows cleanrooming is its how it promotes understanding, creates labor, and is generally too expensive for your entire product to be wholesale copied the moment you publish it.
The law neither allows nor disallows clean room reverse engineering. Clean room reverse engineering enables a defendant to logically prove they have not copied the any protected expression, because the implementers have never seen any protected expression.
Forgeties79
3 days ago
> You lose that if you use an LLM which was trained on the original expression.
>To keep the clean room clean, you'd need to use an LLM which you could prove had never seen the original.
In this true though? It seems the jury is out on whether or not LLM outputs are considered “new” or not legally speaking.
kube-system
3 days ago
> It seems the jury is out on whether or not LLM outputs are considered “new” or not legally speaking.
Not exactly, and that's an unhelpful way to think about it. Copyright law, technically speaking, DGAF about whether you are using an LLM or not. Yes, there has been a lot of talk about whether LLM outputs qualify for copyright protection. And the answer isn't completely straightforward, but the ultimate answer doesn't have anything to do with LLMs -- the law requires a certain modicum of human creativity to qualify for copyright protection. So the real legal answer is that it depends less on whether you are using an LLM and more about how you as a human are using it. That line isn't well litigated yet -- but in general, if you want a copyright, you need to be ready to explain how human creativity was involved -- that's, by definition, the only thing copyright protects.
BUT, this doesn't matter at all in this situation, we're not questioning the copyright protection status of the LLM output, we're questioning whether it violates someone else's IP.
The point of clean room reverse engineering is a defense to prove you could not have violated IP rights of others (not just copyright, but also trade secrets, and patents). The entire point of the exercise is that the implementation team has never seen the original work, so it is impossible for them to have copied the expression. They only could have come to it coincidentally by implementing the idea in the same way. This protects you from a copyright lawsuit because copyright only protects expression, not ideas.
LLMs, are proven to possess the capability to reproduce expression: exhibited as overfitting and/or memorization. So, to form the same defense with an LLM, you'd have to prove that the LLM couldn't possibly be overfitting or repeating from memory. And the best way to do that is to do it the same way it is done with humans: you need to ensure they never saw (i.e. wasn't trained on) the subject of the reverse engineering.
If you don't do this, here's what happens:
You go to court, the plaintiff says "look, this code is basically the same as ours!". The defense says "no, we super duper promise we didn't copy it, we just coincidentally came up with very close to the same thing". They ask "well did your developers ever see our code before?" ...and... <drumroll> the outcome depends on that answer.
It's the exact same situation with an LLM. If it almost exactly produces something that is already copyrighted, the plain simple answer is that the model was overfitting or memorizing the training inputs. You need a believable defense to that.
Forgeties79
3 days ago
>unhelpful
I think that’s a bit of an undeserved framing personally
kube-system
3 days ago
I don't mean it in a personal way, I've seen many tech blogs trying to distill the topic down to something like that. It is inherently an oversimplification.
ajb
3 days ago
Yeah, that's why I only argued above that you could use an AI for the first team. I thought the comment you were replying to meant all clean-room is fishy.
bragr
4 days ago
Well since most providers have cross session memory or things like it, using another provider would ensure you aren't leaking any information from the first step to the second step aside from the spec produced in the first step.
freeAgent
4 days ago
This is, famously, how Compaq created IBM-compatible PCs.
ahknight
3 days ago
And let us never forget SCO vs Novell.
datakan
4 days ago
> you're sleep-walking into a legal minefield in an already legally gray field
I disagree. The WINE project settled all this reverse engineering legal nonsense a long time ago. They can file all the suits they want but they will just get dismissed.
jeremyjh
3 days ago
They will not "just get dismissed". They will get dismissed after the defendent spends thousands, if not tens of thousands on legal defense to establish these assertions of yours to the satisfaction of a judge.
fulafel
3 days ago
This was about patents (monopoly on published methods) so reverse engineering is a red herring.
PaulHoule
3 days ago
I thought synaptics had some sinister hold on the computer industry which sabotaged Windows 8.
Like when you looked at all the confusing and unwanted "2-in-1" and hybrid computer designs the thing you looked at was that they twisted themselves into pretzels to include... a trackpad! For most applications other than "working on an airplane" a true tablet with a bluetooth keyboard and mouse is so preferable but for various reasons the industry never showed it to you.
Forgeties79
3 days ago
> and also risk over-relying on the AI's trial-and-error instead of developing a solid intuition of its functioning yourself.
Not something everyone values unfortunately.
squidmaster
3 days ago
[dead]