HN Simulatornew | past | comments | lists | submit | skygazer's commentslogin

I can connect to US account services, but while first VPNing through France, rather than directly from US.

Is writting software "implementing," or is compiling it and releasing binaries (for free,) executing them, or just profiting from the sale, from the perspective of patent law? Isn't describing how something works first ammendment territory, even if the speech is compilable? (err, I guess this was Scotland & UK patent law, not the US, so 1st ammendment is, of course, irrelevant.)

There is no "fair-use", non-commercial, or even personal-use exception to patent law.

Code is speech, but the 1st amendment has recognized exceptions for IP.


That's not quite true. While there's no exemption for personal use, patented inventions can still be used for academic and research purposes. Even if you derive income from that.

E.g. it's a fair game to write and release a book: "How to Replicate DisplayLink" that would walk the reader through the steps in implementing it.


The experimental use exception is so narrow (in the US at least) that typically even non-profit universities often cannot use them without permission.

https://gesmer.com/patent-copyright-trademark/statuatory-cha...

> it's a fair game to write and release a book: "How to Replicate DisplayLink"

And of course, patents don't protect the information that is in them -- they intentionally disclose it -- what they protect is exclusivity of use.


Yep. I was involved with drug research, and the only accepted use for unlicensed compounds was testing their efficacy in benchmarks against your compounds.

A program per se though is just a detailed description of the "invention" (i.e. a better version of the patent document itself). That's the entire software patent loophole; they take pure math and say "a system where a computer does the described math."

A software program is a "description" of a patent like an office building is a "description" of a blueprint.... they aren't.

They are implementations.

> That's the entire software patent loophole; they take pure math and say "a system where a computer does the described math."

The "do [x] on a computer" loophole was overturned in the US. https://en.wikipedia.org/wiki/Alice_Corp._v._CLS_Bank_Intern...

But what does any of this have to do with the above discussion?


No, a program is literally, exactly, a blueprint. It is the abstract description. This can even be formally stated mathematically (i.e. programs are formal mathematical objects). Note that you can touch an office building. You can stand inside of one. You cannot touch a program. You cannot stand inside of one. There is no concrete "thing" that is the program. You can only run a program using a (universal) computing machine, which can perform any math.

Ok, call it whatever you want, all of these semantics and metaphors are legally irrelevant anyway. I still fail to see what this tangent has to do with this conversation.

Because a patent is a description of an invention. A computer program (like a driver) is also a description of some algorithm. This is not a metaphor. It is definitionally exactly true. Saying it's protected by a patent is like saying an engineering book is protected by a patent (in fact, you can print a program in... a book, and there are many programs contained in books). It's not only false, but would go against the entire purpose of a patent, which is to publish a description of the invention.

A computer program being "a description" or being published does not preclude it from violating a patent.

Perhaps you disagree with this, but this is the current state.


It does because patents don't cover descriptions of inventions. They cover inventions. Words in a book aren't the right "type" to be able to infringe on a patent, but they can be a computer program.

You yourself linked to a case limiting even the ability to claim that running a computer program on a computer can be patented. The program itself was never patentable.


Right, software itself not patentable, only inventions. That is completely unrelated to the question as to whether software can violate a patent.

If you write software that implements an invention that is patented, you have violated the patent.


The point is that software does not implement an invention. You're making a type error. No software does. It's like saying "a green tastes good." Green doesn't taste anything, and you can't have "a green." It's a color. When you write those words it's just a nonsense string. Software doesn't implement any patent just like 5 or pi don't implement any patent. It's a formal mathematical object. It also can't violate a patent because you can't patent mathematical objects, only inventions. An implementation of a patent may contain a computer that runs software.

It's like saying a circuit diagram could infringe on a patent. No, only a circuit can. Software is the diagram. A thing running software can infringe, but software isn't the right type of thing to be able to infringe. You can print books full of detailed diagrams and instructions all day long. The patent is already supposed to describe the thing! It's not secret! A description of every last detail is not protected!

This is also why you can prove things about software. You cannot prove things about inventions; only models of the inventions. You can actually prove statements about software itself because it is already part of formal mathematics.


Again, you are conflating "what can be patented" with "what can infringe"

If you make, use, produce, supply, induce someone to use anything that does a thing that is listed in a claim of a patent -- you have violated the patent.

The argument that "software is just abstract math and so there's no way it can infringe" is bologna... like sovcit-level bologna. Any actual articulable piece of written software is not some abstract mathematical concept, it is the configuration for a machine. Courts simply do not treat software in the way you are asserting, they treat it like a component in a system. And whether or not you built the full system or you are contributing to or inducing someone else's system, doesn't matter much, it is merely the difference between direct and indirect infringement.

If what you are asserting was true, it would be great for the software industry -- we wouldn't have to worry about patents anymore!


The Supreme Court of the US literally held that software is like a blueprint, or a detailed set of instructions. Because that's what it is:

> Abstract software code is an idea without physical embodiment, and as such, it does not match §271(f)’s categorization: “components” amenable to “combination.” Windows abstracted from a tangible copy no doubt is information—a detailed set of instructions—and thus might be compared to a blueprint (or anything else containing design information). A blueprint may contain precise instructions for the construction and combination of the components of a patented device, but it is not itself a combinable component.

Courts do actually understand that software is an abstract mathematical idea. And this is the only sane way to treat software. Otherwise you'd have e.g. Lean proofs as patentable subject matter. There is no line you can draw where you say "this is software, and that is a proof." Formal reasoning (e.g. through types) is all over real world, useful software.

For what it's worth I've never actually encountered anyone who's even mentioned patents at work in the last decade. It's not a day-to-day concern that I've seen.


Yeah, when it comes to determining the question of the country of origin for the particular way they were delivering the software -- which was what quote was about.

Courts did find MS liable for infringing on AT&T's patent for Windows installed on computers in the US. And this wasn't discussed at all in that court case; it was never challenged up to SCOTUS because it was plainly a violation.


Computers running Windows with some functionality are not the same thing as Windows. That's literally the point the court made. Windows, the software, is abstract math, and cannot on its own infringe on anything. They very clearly laid out that Windows itself is a set of instructions.

Hence they could ship the intangible software itself abroad and install it without infringing on US patent law. Merely creating the software did not infringe.


That case only ruled on §271(f). Plenty of routine FOSS activities could qualify for other violations under that statute. The most likely is probably inducement -- have a README that tells someone how to install the software on a computer? You just induced them to put together the infringing system.

In practical terms there is no point in entertaining the idea of software without any relation to a computer -- people are gonna' run software on a computer, that's what it's for.


(just a 3rd-party observer)

This was quite an interesting thread. I appreciate you continuing to reply calmly.


Well, they got me to google what at first seemed like a far fetched conspiracy theory, but actually seems to have some truth to it.

For those similarly unaware, I guess in 2017 Myanmar's military launched a campaign of mass killings, sexual violence, and destruction of Rohingya, and 700 thousand fled to neighboring countries, according to the UN, and amnesty international, but a 150B USD lawsuit against meta by the survivors was thrown out because section 230 of the CDA makes meta immune, and the dismissal was upheld by the 9th circuit. The court did not determine that meta's behavior was harmless, just that the case was barred. At issue was poor moderation, and algorithmic amplification of genocidal messaging on the platform. Apparently facebook was repeatedly warned for years that they were contributing to enthnic hatred and violence by amplifying false narratives and took no action.


My worry is that what we recognize is the unstylized default, but it might be very easy to emulate any particular human easily enough with some examples and a prompt and n-gram/style template biasing token selection, or iterative generation-scoring loops.


Yeah. The big LLMs have fairly idiosyncratic writing styles. Smaller finetuned LLMs can be quite different (and you can also tune them on a particular person's writing). I think some of the confidence in this is because there's plenty of cases which just don't get noticed.


This is probably possible now. Back when machine learning was all the rage people were using it to identify authors of various anonymous works. It became an introductory project to identify the Federalist Papers authors. Just read in some works, identify patterns, set up tests, and set the LLM into a loop testing whether the output would be predicted to be the author you are trying to impersonate.

https://towardsdatascience.com/hamilton-a-text-analysis-of-t...


You undersell. One of those is equivelent to like one hundred seventy H100s or four hundred+ RTX 5090s .. 13TB of GPU mem at 570TB per second bandwidth and 360 petaflops of FP8 tensor compute.


Oh, nice. DIY WhisperSync. I was contemplating such a thing, as well. Do you have a repository one might follow for progress?


Tower of Babel, but for software.


I think humans have to reason because we don’t already have a statistical embedding of the solution pattern built in. We have vastly less rote knowledge crammed into our heads and so require creative synthesis to span the gaps.

With LLMs the trick is revealing their existing relevant embedded knowledge more reliably. They’ve almost literally seen it all before, and the trick is dialing it in. The reasoning tokens help shape the autoregressive attention lens that focuses on and enables recall of the already-experienced answer.

It is interesting that “reasoning” has a similar outward appearance, but since LLMs are built to mimic outward appearance from trillions of examples, you can’t infer underlying mechanism from appearance.


I think many failed/aborted projects only seem stupid in retrospect because we assume it proof they could never have succeeded .. but many ridiculous things can and do succeed and find their place. It fizzled after his death without anyone championing it with persistent care and constructive iterations. It may well have ultimately failed, regardless, but it certainly lost its motive power.


It is interesting that Meta would have to develop a social or political content identification system in order to either label, limit targeting, or ban/censor it. It seems the hard part of complying was developed, but only to thwart the law, rather than comply. That might be a bigger FU to the EU than ignoring the law.


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: