Filtered for ownership

19.48, Thursday 26 May 2022


Sherlock Holmes, the world’s greatest detective, analytic, rational, aloof, is public domain.

Sherlock Holmes, the world’s greatest detective, analytic, empathetic, sometimes even warm, is intellectual property owned by the estate of Arthur Conan Doyle.

If this movie wants Sherlock Holmes to express emotions, its creators need to pay up (The Verge).

The estate suggests the copyright specifically covers Sherlock Holmes caring if Watson is injured or kidnapped.


In 1974 Sir Roger Penrose, mathematician and physicist, discovered Penrose tiling (Wikipedia), a form of aperiodic tiling with (potentially) fivefold symmetry.

There are two shapes, a thin rhombus and a fat rhombus. Using these two tiles, you can infinitely cover a plane without the pattern repeating.

It looks regular but it ain’t.

(I seem to remember that the tiling pattern is a 2d slice of a regular 5d lattice, but I’ve never been able to get the code to work. The maths is a bit beyond me)


In 1997 Kleenex produced loo paper embossed with the Penrose tiling pattern, and Roger Penrose took them to court and they stopped. The London Science Museum has the illegal Kleenex in its collection.

Why would Kleenex do this?

The story I always heard was to do with luxury quilted paper: any other embossing patterns would risk repeating as the paper winds round and round, and so the loo roll would bulge and become uneven. Smart!

How would Penrose get away with owning the pattern?

That’s what I don’t get… the algorithm can’t be owned, right? Just the output. And with an infinite pattern, you just scroll to a part of the infinite canvas that hasn’t been copyrighted?

Anyway, they settled out of court.

RELATED: who owns the immortal cell-line of Henrietta Lacks – or rather, not of her but of a cancerous tumour? As previously discussed.


Some celeb owns an NFT of a cartoon ape. It has been stolen. Seth Green, whose NFT was stolen, has been pleading on Twitter with ‘DarkWing84,’ who bought his ape from a scammer, to return it.

(Matt Levine, who writes daily about money on Bloomberg, is - in this age of oligarchs and impossible-to-understand high finance - 100% a must read. Levine covers the NFT theft in his column today.)


The source code of that code that implements an NFT contains a mapping that associates an arbitrary ID (representing this particular instance of the cartoon ape NFT) with a particular Ethereum address. This mapping is traditionally called “owners”. Does DarkWing84 own the ape? The code says he does.

Here is a fascinating legal unpacking by James Grimmelmann (Twitter thread). It’s… common sense? But I’d never considered it as plainly as this!

  • Some people might object that an NFT is just some made-up collective hallucination on a blockchain, not an actual thing like a car or a book that can be possesed and owned. But lots of other made-up things can be owned, like patents and corporate shares.
  • This is also a familiar distinction in property law: between ownership and possession. If you pick up my computer and walk away with it, I have ownership and you have possession. I can sue you to get it back. If you intended to keep it, you can be arrested for theft.
  • But just as some changes in possession do not transfer ownership, some changes do. If I give you my computer, you become the new owner upon delivery, i.e., when I give you possession. I cannot now sue you to get it back, because you are now the owner.

Forget about NFTs being a made-up collective hallucination. Ownership is a made-up collective hallucination!

Incredible to think of ownership as being so arbitrary.

Implies that we could have completely different configurations of ownership, moral frameworks around it, feelings around it.


GPT is copyright laundering, says v buckenham on Twitter.

And continues: I think copyright has a lot of problems, but I care about the bundling up and reselling of so many people’s voices and labor.

GPT being the startlingly good AI text generator that I got to play around with in 2020. My takeaway: GPT-3 is capable of original, creative ideas.

If the input to GPT-3 is all the text on the internet, some of which is explicitly in the public domain, but most of which has been written by actual people who could be asked and could assert that they have ownership of their words, and the algorithm is proprietary, what is the copyright status of the output? OpenAI, the makers of GPT-3, act as if there is none. Ownership vanishes. Sleight of hand.

BUT: If the input were a single text, and the algorithm were a simple and automatic translation (say) from English to French, the output text would clearly not escape the original copyright.

So where is the threshold?

Could it be the quantity of input text? The complexity of the algorithm? No. Because if I ask about something really really specific, for example some song lyrics, the GPT-3 generator doesn’t somehow “clean room” the words and make them public domain.

AI plagiarism will be a future problem.


Maybe there is a market for a future GPT-PD, where PD stands for public domain, and the AI model is guaranteed to be trained only on public domain and out-of-copyright works.

And litigiously cautious megacorporations like Apple will use GPT-PD for their AI needs, such as autocomplete and auto-composing emails and how Siri has conversations and so on.

But out-of-copyright works tend to be older.

So my future copyright-unencumbered smartphone will cause me to speak like my great-grandmother, with the cadence and turns of phrase of Victorian novels and newspapers.

More posts tagged:
Follow-up posts:

If you enjoyed this post, please consider sharing it by email or on social media. Here’s the link. Thanks, —Matt.