There's a related problem with copyright. It's not about the fact that entity has the rights to make money from something the entity created. It's that they "lock up" the idea behind an exclusive license that has to be negotiated separately, with no way to arrive at a fair deal.
For example, no-one else can try to produce a better Star Wars movie or video game that what the copyright holders permit. Harry Potter famously has a ton of fanfiction, but no one can sell it. Tons of people want to simply make youtube videos of them playing a Nintendo game and...
A "fair deal" would be that some % of the revenue for selling a derivative work goes back to the original holder, recursively, where the % is defined by law, and the original holder does not get the right to block the distribution of the derivative work, their consideration is financial. @gwern might have a clearer idea of how this might work mechanically.
There is a similar problem where to make a new and innovative service work, you need some way to make use of a large quantity of copyrighted content. The Google books case is similar to the AI training lawsuits now, and Google seems to have eventually won in court.
What's the actual legal argument against copyright holders from being able to block certain uses they dislike?
There isn't one. Copyright owners can block uses unless certain fair use conditions are met.
I am claiming this is wrong, others should be able to build on the ideas others have created so long as they share revenue with the first creator, recursively. Ironically fair use, which llms may turn out to fall under, mean the fair user doesn't need to compensate the original owners a dime.
The Google books case, the 2nd circuit agreed that copying the full contents of most books ever published was fair use, and providing search engine services where the full text contents of all those books are searched, and small snippets provided to online users, was fair use, and compensation is not required.
I am not a copyright lawyer, but the LLM cases seems to have the same elements as Google books case.
To actually change the real world in this regard, at least in the US, there will need to be arguments to cross many hurdles, such as to convince a majority of congressmen, and inevitably many judges when it get's challenged. And probably even beyond since the USG has ratified most of WIPO, which it can't unilaterally change.
Inevitably there will have to be convincing legal arguments or else this won't get far enough for it to practically matter.
If you don't have any right now, maybe try focusing your efforts on coming up with some?
There are lower hanging fruit with greater ROI.
Copyright law doesn't protect scientific facts so progress is able to be made, and it probably doesn't protect authors and artists, so data inefficient AI can be trained.
I was simply noting it causes a dead weight loss in creative output, which has become much worse by extending copyright to effectively eternity. Almost everyone alive when steamboat willy hit theaters is not.
I was simply noting it causes a dead weight loss in creative output, which has become much worse by extending copyright to effectively eternity.
How do you know the 'dead weight loss in creative output' outweighs the positive effects in the first place?
It doesn't seem obvious at all to me if there are no such arguments put forward.
Copyright, both in the US and internationally, is JUSTIFIED as protecting creators and ensuring they get some of the value they bring to the world. It's ACTUALLY standard politics and balancing what big lobbyist corporations want and what lawmakers think won't get them recalled by the public.
If it were possible to have a sane system, it would NOT be one-size-fits-all. The duration would be variable based on multiple dimensions of legible value given and received. Everything gets, say, 5 years. Then perhaps a bidding process to extend 5 or 20 years (with additional later bids if the full term isn't bought), with some discount or advantage to the current holder.
Dagon, yes that seems like a reasonable setup. Its pretty amazing that world and life altering inventions gets a protection for a maximum of 20 years from the filing date where as if someone doodles something on a paper get a protection that lasts the life of the author plus 70 years. But... maybe the culture war is more important to win than the technology war?
Anyways, with the content explosion on the internet I would assume that pretty much every permutation of everything that you can think of is now effectively copyrighted well into the foreseeable future. Will that minefield prove to be the reason to reform copyright law so that it fits into a digital mass creation age?
Note that patent and copyright limit different kinds of use, and apply to different things, so there's no reason to compare them, and no real reason they should have similar terms. In both cases, though, there are very different circumstances, different investment levels, and different re-use (and expansion) possibilities for the idea or expression, and there should be variable terms. Oh, and variable enforced-licensing and fair-use regimes.
Dagon thank you for follow up on my comment,
yes, they are in some ways oranges and apples but both of them put a limit on your possibility to create things. One can argue that immaterial rights have been beneficial for humanity as a whole, but it is at the same time criminalizing one of our most natural instincts which is to mimic and copy what other humans do to increase our chance of survival. Which lead to the next question, would people stop innovate and create if they could not protect it?
Sure, the point is to legibilize and monetize ("protect") the creation of valuable expressions and inventions. The fact that some of the motive is to make it visible enough to tax transactions around it is kind of irrelevant. I can't think of a way to do that except by limiting the ability of others to copy/reuse them.
I don't have a strong opinion on when or whether it's a good mechanism for fostering innovation. I suspect there are better ways, but probably not enough better that we'll find and implement them through the headwinds of inertia and currently-mostly-OK.
I DO have a strong opinion that we overgeneralize and apply the same rules and limits to very different types and scales of thing, and the current setting is too much in duration and coverage.
Is the slow forward march of copyright terms the optimal response to the massive changes in information technology we’ve seen over the past several centuries?
Of course not! Even without the detailed analysis and assorted variables needed to figure out anything like an optimal term, economics tells us this can't actually help much in increasing idea, tech, patent, or creative work output. Market size for copies of a given work changes over time (speed and direction vary), but to a first approximation assume you can get away with holding it steady. Apply a 7% discount rate to the value of future returns and by year 20 you've gotten roughly 75% of all the value you'll ever extract. By year 50 you're over 95%.
Even without copyright, Disney must still own the trademark rights to use Mickey in many ways, so that part is really about copyright. Honestly outside of an occasional TV special I can't remember the last time there was an actual new creative work about Mickey at all. Who can claim that copyright was still incentivizing anything at all? What court would believe it?
Patents are trickier, because they start at date of filing, and in some industries it can take most of the patent term just to bring it to market, leaving only a few years to benefit from the protection. Something as simple as a lawsuit from a competitor, or any form of opposition to building a plant, or a hiccup in raising funds, could create enough delay to wipe out a significant chunk of a patent's value, in a way that wasn't really true in a century ago. It makes little sense to me to have the same patent durations across software, automotive, energy, aviation, and pharmaceutical inventions/discoveries.
It’s fairly rare for a patent to be granted and only have a few years left, and if it does that’s typically because of patent owner delays rather than uspto delays. The US law specifically gives you extra time post grant based on uspto delays. Also US patent holders have access to pre-issue, post-publication damages for cases where infringers had actual notice of the published patent application.
But even given that, I am 100% in agreement that patent terms should be extended.
Yes, agreed, and just to be clear I'm not talking about delays in granting a patent, I'm talking about delays in how long it takes to bring a technology to market and generate a return once a patent has been filed and/or granted.
Also, I'm not actually sure I'm 100% behind extending patent terms. I probably am. I do think they should be longer than copyright terms, though.
Information goods have private costs and public benefits. Pharmaceutical research requires upfront investment from one firm, but once a certain compound is known to be effective it’s cheap for any company to produce. Writing a novel or filming a movie takes hundreds of hours of work from an author, but once the first copy is stored digitally it doesn’t cost anything to make a million more.
This is information’s greatest advantage and its central challenge. An idea, produced once, can spread and benefit millions without much investment in its reproduction or maintenance. Precisely this easy spread, though, means that the progenitor of an idea has little advantage over copiers. Copies carry almost all the benefits of information and almost none of the costs, so the original investment in creating the information is difficult to recoup.
Copyright attempts to correct the resulting underproduction of information by increasing the cost of copies for everyone except the original creator. This taxes information’s greatest advantage to subsidize its central challenge.
Since the invention of the printing press, the cost of producing and distributing almost all cultural artefacts has fallen rapidly. It is several thousand times cheaper to film a video, record a song, or write a paper and send it to a million people today than it was 50 years ago. Over this same period since the printing press, copyright was invented and repeatedly extended in breadth and length from 14 year bans on reprinting books to `life + 70` year monopolies over all things that look like Mickey Mouse or sound like Marvin Gaye.
Is the slow forward march of copyright terms the optimal response to the massive changes in information technology we’ve seen over the past several centuries?
Copyright before and after the printing press
How would optimal copyright law respond to the invention of the printing press?
The press affects copyright through three channels: Fixed costs of information, marginal costs of spreading information, and the substitutability of author-copies and knockoffs
The upfront fixed cost is easiest to understand. As printed books filled libraries and bookshops it was cheaper to do the background research required to write a new book. Decreasing the required upfront investment unambiguously decreases the optimal strength of copyright, all else equal. This upfront investment is what copyright is made to reimburse so if it decreases, copyright shouldn’t reimburse as much.
In the early years of the press, before printed books had filled all the libraries, the printing press mainly changed the marginal cost of spreading information. The upfront research and drafting that went into a book was unchanged; notes and first drafts were still handwritten. But the marginal cost of producing an extra copy once the draft was done fell by a few orders of magnitude.
This marginal cost decrease shifts the supply curve for books outwards, lowering prices and raising the quantity demanded. This higher quantity of books lets original authors compete more closely with copying competitors because the upfront fixed costs of research can be spread out over more books, decreasing the price difference between original and knockoff copies.
For example, imagine a market for books where author-published copies and knockoffs are imperfect substitutes, so they split the market. Under the pre-press production function and demand, only 100 books are produced. Say 10 come from the author and the other 90 come from copiers. If the author wants to recover their upfront investment, they have to price their books higher than the marginal cost of production by one tenth of their fixed cost. Depending on the elasticity of demand for author-published copies, this price hike might not leave them with any customers.
After the press, books are far cheaper to produce so the equilibrium quantity rises to 10,000. If the market shares of author-published copies and knockoffs stay the same, the author will sell 1,000 copies. Recovering their investment now only requires adding one thousandth of their fixed cost to the price of each book, a much more competitive price compared to the copied version. With smaller price differences, the little advantages that author versions have, e.g authenticity, warm-glow from contributing to public goods, or first-mover advantage, can more easily justify the author’s higher prices.
So decreases in the fixed and marginal costs of producing information seem to lower the optimal strength of copyright: The upfront investment is cheaper and authors don't need as much help competing with copiers when the market size is big.
But the printing press didn't just change costs, it also made author copies and knockoffs better substitutes. Knockoffs could be made faster so the first mover advantage of authors wasn’t as valuable and the printing press decreased transcription errors so copies were more reliable. When two goods become closer substitutes, changes in the price of one have larger effects on the quantity demanded of the other. In this case, both goods are decreasing in price at the same time and the net effect depends on the relative size of price changes and the elasticities of demand for author originals vs knockoffs.
Empirically, this seemed to balance out in favor of authors and massively increased their production of books. Despite the initial lack of copyright protection (or perhaps because of it), the printing press made it easier than ever to make a living writing and selling books without the subsidy of an appointment at a university or monastery.
Copyright before and after the internet
Most of the major information technology innovations after the printing press follow the same pattern of massively decreasing the marginal and fixed costs required to produce information but also making author published originals and copies closer substitutes.
This pattern reaches its zenith in text on the internet. It’s never been cheaper to research, write, and publish something that millions of people can read. But it’s also never been cheaper to copy text with perfect fidelity and post it somewhere else. So should text on the internet have longer or shorter copyright than text in books in the 16th century?
I'm honestly not sure, but I suspect modern copyright is far too long for two reasons.
1.) Empirically, the of exploding rates of content production we saw with the printing press has held with all following information technologies as well, even in areas with weak copyright. It seems that lots of people (including me) are willing to put hours of effort into content that they post on the internet for free. Billions of hours of high effort and quality content are available for free on Youtube. The generosity and patronage of a few thousand true fans is often enough to subsidize the fixed costs of creation for the rest of us.
2.) Our current copyright regime is based on concentrated-benefits-diffuse-costs attacks from special interests rather than anything like this optimal tradeoff so it seems unlikely that we’ve got it right, and the bias is towards overly strong copyright.
More of our economy is information goods than ever and this is only accelerating. Generative AI will continue this centuries long trend. The upfront cost of creating a beautiful video or piece of music will fall further along with the marginal cost of spreading it, while the substitutability of copies will increase. Better understanding the effects of this type of technological change is important and growing more so by the day.
See also, Gwern's useful comments on this post