Breaking the rules is in Big Tech's blood – now it's time to break the habit Microsoft's journey through intellectual property has been a multi-year saga that makes Game of Thrones look like a haiku. From the young Bill Gates and Paul Allen plundering software companies' garbage bins for source code listings, Gates writing a snotty letter to computer clubs a few years later complaining about software …
Or any other of the big publishers. I hate their guts, the whole modus operandi is aimed at screwing over scientists. But here is a place they might have a use.
Lots of them have now Open Access models, and those open texts have licenses. As do moat open source projects, but they have no money. I'm pretty sure scraping them and training LLMs from them is verboten as per T&Cs.
Streaming showed that if you give people the opportunity to view/listen to stuff in a convenient way for not a lot of money then suddenly Bittorrent etc become a lot less attractive.
As the market for streaming video becomes more fragmented and you need a zillion subscriptions to view everything[0], maybe people will rediscover piracy...
[0] Here in Switzerland it's often not even possible to pay to watch stuff
In part because the contract negotiations between artists and record labels opened up loopholes. Artists wanted so many dollars per LP sold and so many cents per air play. Industry lawers, being the sneaky b..tards they are, agreed. Knowing that any new technologies (downloads, streaming) would have to be renegotiated (i.e. more work for lawyers).
Musicians should have argued for one payment per copy sold, no matter what the format (cassette, CD, thumb drive) and the eqivalent for "broadcast", whether on the local FM station or the 'Net.
If it were just that simple, I'm afraid. Music industry revenue basically comes from three streams
The music industry (at least in the US), uses two forms of rights to the product, rights to the master physical copy and rights to the work itself.
Record companies keep rights to the master which gives them the rights to copy and control who can distribute copies made form that master. The record companies can offer a royalty to the artist which goes to pay the advance given by the record comapny that the artist used to make the master. Only the very highest selling artists will ever actually see these royalties, as most won't ever cover the advance. Honestly, most artists don't really value these rights as they never really do much for them.
Publishing rights are generally held by publishers and writers, often in a 50/50 arrangement. These are the rights needed to broadcast the music, use it for an event/commercial/movie/etc, allow it to be played in a venue, or allow it to be used on an album. The artist only ever sees money from this stream is if they wrote the music. Publishing rights are traded rather frequently, too, so the rights holders change on these. Artists tend to get more out of these rights, even long term, but it takes a few hits and years of writing to turn this into a steady income. You never know when a movie, series, or commercial will come along and decide a song you wrote thirty years ago is perfect.
The third stream is basically income from venues as well as merchandising. This is where most bands make their money. The reason for getting a record with a label is because it means marketing and distribution of albums that get people into seats.
The internet disrupted the industry by heavily blurring the line between what counts against master rights and publishing rights. Spotify is an easy example. Should they be paying for rights against the master? against the Publishing? or Both? Streaming would seem to be akin to broadcasting, but Spotify allows you to download them as well, so that would count against the master. Then there is the fact that streaming is on demand and used in the same way a copy of the master is used, does this make streaming more like owning a copy of the master and less like broadcasting? Should the publishers or the labels be paid more, then?
Everyone wants their pound of flesh and the artists are often the runts being pushed to the back. Also, when you do get a big name artist in the fray, they are typically after something that benefits themselves more than artists in general.
Music’s fine. There are tons of bands and artists writing, performing, and recording all over the country the world. On any night in pretty much any town you can find talented people playing and singing their hearts out for you, hawking CDs and T-Shirts out of a cardboard box, and touting their social media links for the pure joy of it. I should know, I’m regularly one of them[1].
The music industry however is another matter - that’s dead[2] And to be honest I don’t see any reason why anyone would miss it. If (the Taylor Swifts of the world aside) the recorded music business implodes leaving the space it used to occupy to hyper local scenes of passionate amateurs and semi-pros making music out of compulsion rather than to put cars in the garage and yachts on the water for shareholders in corporations that’ll be a huge step forward for humanity…
[1] In my case I’ll leave judgement on the talented bit to the audiences
[12] Although to a greater or lesser extent the corpse is still twitching to give an illusion of life…
We should note that the successive waves of piracy and streaming have indeed pretty much killed music.
I keep seeing people saying this, but it's simply untrue.
It's easier than ever for musicians to make their content available, either online under their own control, or through their choice of streaming platform. For the consumer, they now have ready access to millions of tunes whereas most people previously had pitiful collections in the 60-100 album range. Consumers pay (though the large services) for the content they listen to, instead of paying ten bucks for an album with one or two good songs and a load of filler tracks. They don't have to find the space and orgnisation for a large physical collection. They can explore new music without the risk of buying an album in hope and being disappointed. The low costs of exploration, and opportunity for the services to proffer suggestions is fantastic - I've found loads of new artists as a result of using streaming services who I would never have heard of, or never risked buying an album "cold".
People seem to be harking back to the "gold old days" of big record companies, ignoring the fact that a few large companies decided who (musically) lived and who died, ignoring their exploitative execs, producers and A&R w@nkers, that the charts were actively rigged, and most people of couldn't afford to buy more than a handful of crap, poorly made vinyl records. Somebody will be itching to say that the streaming services make most of the money, and that's probably true, but how's that different to the good old days when the record labels made most of the money?
"Somebody will be itching to say that the streaming services make most of the money, and that's probably true, but how's that different to the good old days when the record labels made most of the money?"
It's largely true AFAIK, Spotify pays a pretty miserly rate to the artists based on what they make apparently. The issue though is going to come when the record companies do what the film and TV companies have done and decide that instead of letting a few content aggregators (Netflix and Amazon Prime) make all the money each month, they'll start their own streaming services and we're back to the bad old days of the record companies being in control. Even worse is that hedge funds and the like have realised how much money is in artist's back catalogue rights and they will take every opportunity they can to commercially exploit that.
>> We should note that the successive waves of piracy and streaming have indeed pretty much killed music.
> I keep seeing people saying this, but it's simply untrue.
It depends on the context.
It has become far easier to produce and distribute music. However, this has resulted in a great increase in the volume of music being produced - a quick glance indicates that there's roughly 120,000 new songs released to streaming platforms every day.
And sadly, that means it's much harder to get people to actually listen to your music. Which in turn means that it's much harder to monetise your music, and thereby makes it more difficult to fund production of new music.
There's always going to be exceptions to this - some people will be in the right place at the right time, and will ride the zeitgeist to fame and fortune. But there'll be tens of thousands of others left behind. And with people now starting to use AI to produce music, the signal/noise ratio is going to get even worse!
Only music in the "charts" has gone South. And that is probably not due to piracy. In fact a pirate warbling in a croaky Somerset accent would probably sound better than most of the current chart music.
I have never before in my life had so much access to fantastic music that I do like. It's just not in the pop charts anymore, that's all.
Music was killed by the greed of the studios. The only people making money out of it are Apple, Amazon and Spotify with Napster being causal in this process of greed schadenfreude.
The same is under way with Streaming with likely winners being Amazon Prime Video, Apple, YouTube, Netflix and an inevitable consolidation around Disney+/Max/Peacock/Discovery/Paramount + studios last stand.
Schadenfreude Part Deux pending in about 5 years, with a renewed war on Pirate Bay declared a lost cause.
i'm increasing uncomfortable with streaming. New content is increasingly awful at least to my taste and price keeps going up. I know it's a running joke,"i've watched the Netflix". But I'm actually starting to think that out of the stuff I want to watch I am getting to the point of having watched it all. We tend to watch older stuff either because it's better or we just like it more. There is very little that is new that we watch. The older stuff is also becoming increasingly hard to find.
Lots of very good older (not necessarily old) films and tv shows are vanishing. Not streaming and not available on DVD.
For music and film that I really like I've started rebuying DVDs and CDs. Mostly second hand as they are often cheap as chips right now.
I've had the conversation with my kids and despite having a family subscription to Sportify they have also started building their own CD collections. Again often second hand very cheaply.
Streaming can be convienient, but it's not to be trusted.
The other issue I have with streaming services like Spotify is they change the mixes of songs from time to time - mostly for the worse. And once they are changed that's it, you can't hear the original/real version anymore.
Another good reason to have CD's or at a minimum buy a downloadable version of tracks you really like.
Indeed, any license that explicitly requires attribution is already being expressly violated because no attribution is being granted. This will end up in courts but the fight is nonsense because Open Source and most Creative Commons licenses are unambiguous in their demand for attribution.
I fear that a nonsense battle is actually what the corporations want and foresee two awful but non-exclusive outcomes:
1. They want to argue that attribution-at-scale is just not practical – they would, in essence, have to cite every public web site ever published – and use the impracticality to somehow neuter the very concept of "attribution"
2. They intend to capitulate and try to strike a compromise: they wreck pre-AI copyright and, in turn, precedent that AI-generated "content" is uncopyrightable (being built from the ashes of pre-AI copyrighted works) is carved in stone...
But 2 is a trap, not their end-game! Their end-game, then, will be to punt the idea that, given the right model and the right, designed prompt, AI algorithms can output *any* target content – this has already been demonstrated in the lab.
They'll use that to muddy the waters and cast doubt on whether anything was ever human work and push for precedent that basically ends copyright for works post 2023.
The resulting free-for-all would hand a huge advantage to whoever has the "biggest AI". In a world devoid of attribution, there ceases to be any reward for independent artists, authors, musicians or writers or other creatives to participate. Independents will surely persist out of vim and vigour, in their niches, earning a pittance in kudos and currency but AI content can and will be churned out at scale, heretofore unseen, and eclipse their already meagre visibility.
Borking copyright will further exacerbate the imbalance of power that gives big content houses free license to dictate what the vast majority see, hear or read, whether it comes with ads – even at premium tiers – which devices it can be played on, which lands it can be viewed in, etc.
I don't particularly like the idea of Gates, Bezos, Zuck and the rest choosing the landscape of arts and culture.
#1 is the standard operating model of the social media companies. Simply argue (but not in so many words, obviously) that you've built something so large that you can't possibly police it, except with AI, which in the case of the GenAI companies will create a lovely closed loop where they have to be allowed to do what they want.
The lesson from the social media companies is that we need to stamp hard on that argument early, before they grow too large to stop.
We need a legal framework for GenAI (and for the stuff that everyone was previously calling 'AI') that says if it would be illegal for one of your employees to do it then it's also illegal for your algorithms to do it. Copyright infringement, discrimination, recommending smut to kids, promoting terrorist content... whatever. If your algorithm does it then you're responsible for that in the same way that you would be if an employee did it.
As additional weight behind the law, make advertisers responsible for anything they fund, so if an ad appears on illegal content, the advertiser is also responsible for having funded it. That would clean up the behaviour of big tech unbelievably quickly.
To be fair, he did add: "There's a separate category where a website or publisher or news organization had explicitly said, 'do not scrape or crawl me for any other reason than indexing me,' so that other people can find that content." But he went on to add that this was a "gray area. And I think that's going to work its way through the courts."
I don't think this is a very grey area, although I guess protections could be extended by adding permissions or denials to robots.txt. But then MS and the other data rapists will probably just ignore it. Copyright and IPR has long existed, and has protected MS's own IPR. Publishing copyright material does not automatically grant anyone any rights to create derivative works, unless that permission has been granted. Fair Use doctrine does not yet (AFAIK) permit derivative works, especially if that's for commercial gain.
But MS seems to acknowledge some limitations on scraping and crawling, and these should probably be more thoroughly tested in courts. I guess there may also be complications with privacy policies and T&Cs, if publishers or websites have agreed to share data in exchange for advertising revenue.
Licensing for open source software is both necessary and prudent. Licensing for every file that gets posted to the internet is not the right approach here. This is true for a few reasons. First, AI companies have already tried to ignore that and are fighting to do so in court. Microsoft and its AI providers are major culprits with their Github Copilot product, but that is not the only bot to have scanned lots of licensed open source code and then ignored the licenses. They have already acted as if licenses don't count and they're trying to somehow make that legal. I don't think it will work.
The more important reason is that licenses should not be necessary to deny permission to scrape. If Microsoft forgets to apply their boilerplate license notice to some code and I get a copy, I do not have permission to do whatever I want with it. The same thing should apply to other types of work, and surprisingly enough, national copyright legislation and international copyright treaties confirm that it does. If we establish a precedent that, if you didn't stick a license on your content, any treatment of it is fine, we're making things more annoying for no good reason, and if we're assuming that copyright law will still apply where explicit license terms haven't been applied, then there's basically no benefit to doing it in the first place. A license that requires them to do something they don't want to do should be no more powerful than a license that denies them permission unless they negotiate with you, and that's what we should have by default with normal copyright, no explicit licenses required.
that if I pirate a movie and post it on youtube, then I'll get take down notices (and maybe criminal notices ) spaffed at me like no tommorrow
If I publish said script to said movie, equally as much , I'll get take down notices, if I write a script using said previous script as a basis (a derived work.. think changing Harry Potter to Simon Claythrower), again I'll get takedowns and copyright violations thrown at me
All by the big media companies who own the rights to said media.
However its perfectly fine for m$ to scrape my website, then publish derived works from that......
This pretty much hits the nail right on the head. AI related copyright issues won't be a legal issue until AI has progressed to the point where it affects copyright holders with deep enough pockets.
I suspect that what may be the deciding factor is when generative AI gets good enough to be used to make commercially successful new video entertainment releases based on a supplied script. If anybody can just feed existing movie and TV libraries into an AI model, then give it a movie script, and then get an acceptable new movie out, existing copyright holders won't stand for it. If it is ruled as being legal, then the laws will be changed to prohibit it.
Software copyright is based on the same laws that protect books and movies. Software licenses are based on granting users permissions to use that software under those laws.
What may be needed is some sort of GPL equivalent which says that anyone using a (let's call it GPL V4) work for training must publish all related training material under the same or compatible license.
Most MIT licenses (which don't require attribution) probably won't require a change, as they implicitly give all permissions required for using as training data.
Apache will have to think about their license, as they fall in between MIT and GPL in terms of what they require (I.e. the patent clause).
If you read anything and learn from it (or even better: make some kind of profit from it) then you should have all your "earnings" (stealings) taken away from you, by force.
There is no justification for anybody to "steal" knowledge from somebody else (who had published it "for entertainment purposes") and then derive some kind of ill-gotten income from that stolen insight, it's just common sense people!
OpenAI researchers are absolutely free to read the whole public Internet, if they want, and learn it, if they can. As is anybody else.
But they are not doing that.
They are downloading it, processing it, using it as input to a computer program, taking the output of that computer program, and making money from it.
Nothing in there looks remotely like reading and learning.
What the program is doing could look like reading and learning. But the program is not the researchers, and nobody is suing the computer program.
Even if this were possible,
A) the software would still be microsoft code, full of bugs and bottlenecks and spyware.
B) giving away the software for free would not stop companies from paying $millions to microsoft for the same or inferior versions. If there is no money involved, how do you skim any of it off for yourself and who will bribe you to sign the contract?
What happens if I train an AI on this software to modify it to remove the registration/copy protection?Surely that's fair in Microsoft's view of the online world?
Hmmm...How about this: train the AI on Windows, and have the AI produce a version that does not phone home? Or has a solid, stable, properly working UI? Or actually fixes some number of the myriad bugs that Micros~1 can't/won't fix?
TReko, you may be on to something there!
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
