> That's not legal use of the material according to most copyleft licenses.
That probably doesn't matter given the current rulings that training an AI model on otherwise legally acquired material is "fair use", because the copyleft license inherently only has power because of copyright.
I'm sure at some point we'll see litigation over a case where someone attempts to make "not using the material to train AI" a term of the sales contract for something, but my guess would be that if that went anywhere it would be on the back of contract law, not copyright law.
Indeed, the GPL's definitions of "modify" and "propagate" restrict the license's scope to actions that would otherwise infringe on copyright if not permitted. And fair use and similar doctrines generally act as carve-outs to copyright infringement.
> I can see why people were tempted to cut corners, especially given past tolerance…
Which is one of the reasons that the pre-trump executive orders that granted leniency and amnesty at times were all terrible terrible things to do. We really have a problem in this country where we've decided that the laws suck, but we don't want to do the hard part of changing the law, so we just decide to ignore it. Until at some point someone comes along and decides to enforce the law and now a whole bunch of people who were acting on the de facto state of the law now have to deal with the consequences of the de jure state of the law.
Immigration is a place we've done this a lot, but things like the status of marijuana across the country is also predicated on this sort of arbitrary non-enforcement of the law. Obviously the states are not obligated to enforce federal law, but the feds absolutely can. The feds could raid every marijuana dispensary in the country and take them all down with barely a hiccup, at least from a legality standpoint. Yet they don't because we have decided to arbitrarily not enforce the law, even if we haven't changed the law.
I had really hoped after Trump's first term, we would have seen a real awakening to the amount of things that are allowed only because we don't actually enforce the laws that are on the books, and a real push to both fix the laws and roll back the abuses of executive power like this. But we didn't seem to learn that lesson, and sadly it doesn't look like that lesson is going to be learned this time either.
> The feds could raid every marijuana dispensary in the country
This is not a realistic scenario, but does seem to present the argument, that Federal law and the power to enforce it is inherently and legally superior to individual State laws?
> fix the laws and roll back the abuses of executive power
Perhaps we could start with the federal law that made marijuana a Schedule One controlled substance: so that ordinary residents of responsible and responsive states would not have to rely on "arbitrary non-enforcement of the law" in order to get their smoke on.
Think about it! Every one of those citizen-puffers are in violation of federal law! Send in the troops!
The argument seems to be that federal laws overrule states', and that underlying idea justifies sending military forces into cities, because, you know, federal laws are being violated, everywhere, like, all the time, man.
> This is not a realistic scenario, but does seem to present the argument, that Federal law and the power to enforce it is inherently and legally superior to individual State laws?
That's the current state of US law yes. That's why California needs explicit permission from the federal government to have more restrictive air quality laws than the federal standards. It's also quite literally baked into the constitution:
>This Constitution, and the Laws of the United States which shall be made in
>Pursuance thereof; and all Treaties made, or which shall be made, under the
>Authority of the United States, shall be the supreme Law of the Land; and the
>Judges in every State shall be bound thereby, any thing in the Constitution or
>Laws of any State to the Contrary notwithstanding.
Subsequent rulings and case law have largely established that the states and their law enforcement agencies are not inherently bound to enforce federal law, but that is a very grey area with lots of edge cases, and can change substantially depending on the "incorporation" status of the underlying constitutional basis for the law. This is why the drinking age is controlled by highway funding rather than direct legislative imposition, but it's also why the feds can and will send the military in to ensure your schools are integrated. The interplay between the supremacy clause and the 9th and 10th amendments is a very complex part of the legal system but this has been the state of the country for a very long time.
> Perhaps we could start with the federal law that made marijuana a Schedule One controlled substance
You'll get no argument from me on this front. Especially since there is a federally legal synthetic form of THC that is actively prescribed by doctors for cancer patients. It's called Dronabinol and it's a Schedule III substance. Yes you're reading that right. Psychoactive THC can be obtained via prescription in the US from any pharmacy and that synthetic version is a lower control level than xanax. And all of this in the face of the fact that the plant source of the THC is a Schedule I substance which in theory is supposed to mean there is no known or accepted medical use for the substance. Which seems like a lie.
> The argument seems to be that federal laws overrule states', and that underlying idea justifies sending military forces into cities, because, you know, federal laws are being violated, everywhere, like, all the time, man.
It's not "the argument", it's the actual law of the land. The fact that the feds can't send in the military is that we explicitly disallow the use of military forces for domestic law enforcement. But that restriction doesn't apply to non-military federal LEOs and even with the military, it's a grey area (see aforementioned use of the National Guard to enforce racial integration). And it's one of the reasons why we should have been concerned about the ever expanding federal powers in general and executive powers LOOONG before Trump had ever considered running the first time. It's part of what makes "legislating" via the courts so dangerous, and what makes the impulse some people seem to have to respond to the Trump administration by way of things like court packing and other attempts to simultaneously empower the federal government without also empowering Trump a dangerous impulse.
I did wind up replacing the USB C ports on a 4 year old computer recently because it was dodgy as hell. When i got it under the microscope it the longer bus power pin contacts (and one or two of the others) had been badly worn/squished/stretched in a way that I guess was causing them to bridge to other pins. I assume some USB-C cable had some gunk in the connector which was hard enough to damage the contacts on the center blade, and the user didn't notice (because how often do you look into the end of your USB-C cable?). It probably presented as a cable that wasn't seating right or didn't go all the way in and whatever was inside probably fell out when it was removed and they tried again.
And for what it's worth, damage to the center blade does seem to be a common failure mode for USB-C and mini-usb connectors. Less frequent for something like HDMI but it does seem to happen from time to time. Lightning never felt like it locked in as securely as USB connectors do, but at the same time, every time I saw a damaged lightning connector it was always on the male (and therefore usually cheaper accessory) side.
Stop telling people what will or won't irritate others right off the bat. There's no better way to wrongly speak for someone else. :)
More seriously, it's probably a good bet that most people haven't heard of McMaster Carr in general, and if they have, probably not in the context of having a smart or well designed website. So it's not a terrible article headline. And of all the click bait style headlines that could be used, it's probably the least offensive. At least they give you the name in the headline.
McMaster Carr is probably the most famous industrial supplier in North America. It's a $1.3 billion company that's been in business for over a century. Anyone who makes physical stuff probably goes on their website multiple times per day. I currently have 17 different McMaster tabs open.
It's fine if your personal experience has never lead you to encounter this site, and it's fine to remark on your unique perspective as someone observing it not as a user but as an admirer. But the presumption that others share your ignorance is at best impolite, and the suggestion that some may be irritated should be taken as useful criticism, because plenty are.
Even if “anyone who makes physical stuff” has heard of them, manufacturing jobs represent 12.8 million jobs[1] compared to 158 million non-farm jobs[2]. That’s 8% of the workforce. Let’s double that to account for people that used to work in manufacturing or are in maker hobbies and the like. Even at 16% I would comfortably stand by the assertion that most people have not heard about McMaster Carr’s website being a very high functioning web site.
Well by that logic, there are 4.4 million software developers, much less than half the employment of manufacturing, in the US. Should we introduce stack overflow as a website you've never heard of? I'm sure plenty of people haven't heard of it, but I wouldn't assume any particular person hasn't.
> Should we introduce stack overflow as a website you've never heard of?
Yes? Obviously who you're writing for matters. I wouldn't write an article specifically targeted at and submitted to software developers and use that for a title, but I would absolutely have no problem at all introducing Stack Overflow as a site you haven't heard of (or maybe I'd qualify it as "probably haven't heard of) if writing something for a general audience. It's a safe bet that most people you talk to wouldn't have heard of Stack Overflow, so there's nothing wrong with a headline for a broad audience that assumes that.
Would you object to a headline for an article written for a general audience that calls "lwn.net" a site you haven't heard of? Or maybe an article that calls Admiral Grace Hopper a computer pioneer you haven't heard of? Sure everyone on this site almost certainly has heard of them. But in the general public?
Perhaps, but the Norway tax mentioned in the article kicks in at $174k net worth. That's a paid off house and a nearly drained 401k for even the poorest of Americans. Yes there is an exemption for part of the house, but even if it were 100% exempt, I think you're going to have a rough time getting support for taxing 1% of a retirement account worth less than the code section it's named for.
Replying to myself since it's too late to edit, but according to these numbers[1], it looks like this tax would hit about 52% of American households, so my "even the poorest of Americans" is a bit overwrought. And if we take the US median home price (~410k as of this year[2]) and exclude 75% of that (~307k), then this tax would hit ~30% of American households (~$481k net worth). Even at that, it's still quite a hurdle to clear to convince the top 1/3 of households support a 1% tax on their accumulated wealth.
This is in part because the term "Asperger Syndrome"[1] (which has always been related to / a subtype of autism) is no longer used and is now just under the Autism label.
Together with PDD/NOS (which is an entire category on its own, because you also could be diagnosed with this if you "just" have brain damage, leading to autistic like symptoms), Childhood disintegrative disorder, and Rett syndrome.
But to simplify diagnosis they swept it all up and since DSM V was released all those things are now under the ASD (Autism Spectrum Disorder) umbrella
I imagine that prosecutors don’t pierce the corporate veil for public companies often because “DA charges 87 year old grandmother with stake in Evil Corp in her pension fund with manslaughter, local union members brace for additional charges against members” doesn’t make for good headlines or good justice either.
The officers and board of the company aren’t protected by the corporate veil concerning their actions. They retain some degree of protection from actions of others within the corporation provided they did not have (or did not have a reason to suspect) knowledge of those activities. But to my knowledge that’s not special to officers, it applies to any employee which is why the rank and file Enron employees didn’t get prosecuted.
Learning to program a computer makes it clear that a computer is incompetent. How do we find out if something is in a list? Does this one match? No. Does this one match? No. Does this one match? No. And on and on. One of the biggest hurdles I ever had to get over starting out was understanding how much of what made a computer feel “smart” was just a lot of really dumb things done really fast (and hopefully in the right order and with the right logic).
But that means if you don’t get that logic right, nothing else is right, and that’s where these AI tools lose their magic. When they get it right they’re impressive and when they get it wrong, it lays bare that your computer is very very stupid, just very fast
"Competent" and "smart" are two different things. A computer that correctly executes a program every time is competent. It's reliably doing what it is intended to do.
These genAI tools don't do that, though, so they aren't competent in the way that a computer is.
From that perspective, these AI tools are perfectly competent. They are programmed to take user input, run it through various LLM models and use that output to act on behalf of the user. And they are reliably doing that thing exactly as they are instructed.
The problem isn’t “competence” then, it’s a failure to create a program that does what the user wants it to do reliably. In this way, these AI programs are no more “incompetent” than any other poorly designed program. Any system that applies randomness to its outputs is inherently going to get things wrong because randomness implies a lack of accuracy.
AI is absolutely getting shoved hard and fast in places it’s just not ready for yet, due to the overhyping of its boosters. But personally I think these sort of articles and complaints that treat them as somehow unique from all the other poorly implemented and rushed software that has plagued the industry since the first paycheck was written is accepting the booster’s hype. It’s poorly implemented software running on the same dumb computers we’ve been using forever, of course it’s going to be getting things wrong.
You have to beat inflation. It's not long or short term profitability, it's taking an asset and making it return more than inflation. Treasury bonds or stocks or mobile home parks or monkey nfts, it's all the same thing.
Again, that's with the mindset that the reason to own shares is because they're an appreciating asset.
If you study the history of what investing in a company via shares has historically meant, you'll find that it's only really in the last half century that the "appreciation game" came to the forefront. Dividends yields (or equivalent) were the evaluation criteria for these investments prior to our current era.
You speak about inflation and stock as if the cause/effect relationship is crystal clear and unidirectional, I wouldn't be so confident about that.
Wouldn’t you want the returns represented by your dividends to beat inflation too? Assuming you’re not investing in something for non-monetary gains, if the dividends aren’t beating inflation at least most of the time, why would you stick your money into stocks when you could just by I bonds from the government?
Why wouldn't they? If the company is healthy the dividends paid will reflect macroeconomic inflationary processes.
I don't think "you don't want to beat inflation?!?!?!" is a good faith interpretation of what I'm talking about higher up in the thread - which is rampant speculation for the publicly traded companies, and behavior tantamount to rug pulls in the case of private equity.
If they contribute nothing back, what are all the `google.com` email addresses in the git history doing? If they contribute nothing back, why are they listed as a customer of `fflabs.eu` which is apparently a private consulting company for ffmpeg run by some of the ffmpeg lead maintainers?
What do we think the lesson corporations are going to take from this is?
1) "You should file patches with your bug reports"
2) "Even when you submit patches and you hire the developers of OSS projects as consultants, you will still get dragged through the mud if you don't contribute a patch with every bug report you make, so you might as well not contribute anything back at all"
The text and context of the complaint can be used to steelman it, adopting the principle of charity.
From that perspective, the most likely problem is not that bugs are being reported, nor even that patches are not being included with bug reports. The problem is that a shift from human-initiated bug reports to large-scale LLM generation of bug reports by large corporate entities generates a lot more work and changes the value proposition of bug reports for maintainers.
Even if you use LLMs to generate bug reports, you should have a human vet and repro them as real and significant and ensure they are written up for humans accurately and concisely, including all information that would be pertinent to a human. A human can make fairly educated decisions on how to combine and prioritize bug reports, including some degree of triage based on the overall volume of submissions relative to their value. A human can be "trained" to conform to whatever the internal policies or requirements are for reports.
Go ahead and pay someone to do it. If you don't want to pay, then why are you dumping that work on others?
Even after this, managing the new backlog entries and indeed dealing with a significantly larger archive of old bug reports going forward is a significant drag on human labor - bug reports themselves entail labor. Again, the old value proposition was that this was outweighed by the value of the highest-value human-made reports and intangibles of human involvement.
Bug reports are, either implicitly or explicitly, requests to do work. Patches may be part of a solution, but are not necessary. A large corporate entity which is operationally dependent on an open source project and uses automation to file unusually large volumes of bug reports is not filing them to be ignored. It isn't unreasonable to ask them to pay for that work which they are, one way or another, asking to have done.
> Even if you use LLMs to generate bug reports, you should have a human vet and repro them as real and significant and ensure they are written up for humans accurately and concisely, including all information that would be pertinent to a human.
Look at the report that's the center of this controversy. It's detailed, has a clear explanation of the issue at hand, has references and links to the relevant code locations where the submitter believes the issue is and has a minimal reproduction of the issue to both validate the issue and the fix. We can assume the issue is indeed valid and significant as ffmpeg patched it before the 90 day disclosure window. There is certainly nothing about it that screams low effort machine generated report without human review, and at least one commenter in this discussion claims to have inside knowledge that all these reports are written by verified and written by humans before submission to the projects.
I won't pretend that it's a perfect bug report, but I will say if every bug report I got for the rest of my career was of this caliber, I'd be a quite happy with that.
> It isn't unreasonable to ask them to pay for that work which they are, one way or another, asking to have done.
Google quite literally hires some of the ffmpeg maintainers as consultants as attested to by those same maintainer's own website (fflabs.eu). They are very plainly putting cold hard cash directly into the funds of the maintainers for the express purpose of them maintaining and developing ffmpeg. And that's on top of the code their own employees submit with some regularity. As near as I can tell, Google does everything people complaining about this are saying they don't do, and it's still not enough. One has to wonder then what would be enough?
That probably doesn't matter given the current rulings that training an AI model on otherwise legally acquired material is "fair use", because the copyleft license inherently only has power because of copyright.
I'm sure at some point we'll see litigation over a case where someone attempts to make "not using the material to train AI" a term of the sales contract for something, but my guess would be that if that went anywhere it would be on the back of contract law, not copyright law.
reply