Fritz 17

[Rebel]

The Fat Fritz NN files are not just random numbers, they were created by an creative human act, by selecting games.
I can see the point that argues that it should be covered by copyright.

Nevertheless I guess that science will get into 'Devil's kitchen' if you can copyright or patent a set of numbers.

Where does the copyright start? 1,2,3,4,5? Is an XOR MLP copyrightable? Do I need millions of weights for copyright? What about self-play NNs without human interaction?

Time will tell I guess....

--
Srdja

Would be an interesting court case. Fortunately for ChessBase is that they don't have direct competitors any longer.

[chrisw]

Okay, here's another one.

Same argument.

I don't really understand what you are saying. You may notice by examination that each page in the preview says, "Copyrighted material".
Individual numbers are not copyrightable for obvious reasons. But a book of numbers is copyrightable.

Now, mathematics itself is not copyrightable. So, for instance, I cannot find a new calculus formula and copyright it.

A similar thing relates to chess games. A chess game score is not copyrightable. However, a collection of chess games is copyrightable.

Copyright requires a human hand.
A chess game is not copyright. (Interesting reasons, connected to adversarial creation of game by TWO humans, neither one can claim to have created the game alone).
Pi, e, prime numbers, random numbers are not copyrightable.
A collection of chess games is not copyrightable. Annotated part of annotated chess games are copyrightable because a human has added the annotation.
A download of lichess and then selection of games by Elo or something does not copyright the selection. Mechanical selection processes do not create copyrightable material.
Just because content is printed onto the pages of a book does not suddenly give that content “copyright status” if it has no copyright status in the first place. Pi does not become copyright because somebody prints it out in a book.
And, according to the only actual legal person on these forums, collections of numbers that form the weights of a neural net are also not copyrightable because the numbers are produced not by human hand but by mechanical process. They don’t suddenly become copyrightable just because ChessBase transfers the numbers into a CD. Not even if Frederic Friedel transfers each bit by one by laboriously inscribing them with his own personal miniature magnet.

[smatovic]

Okay, here's another one.

Same argument.

I don't really understand what you are saying. You may notice by examination that each page in the preview says, "Copyrighted material".
Individual numbers are not copyrightable for obvious reasons. But a book of numbers is copyrightable.

Now, mathematics itself is not copyrightable. So, for instance, I cannot find a new calculus formula and copyright it.

A similar thing relates to chess games. A chess game score is not copyrightable. However, a collection of chess games is copyrightable.

Copyright requires a human hand.
A chess game is not copyright. (Interesting reasons, connected to adversarial creation of game by TWO humans, neither one can claim to have created the game alone).
Pi, e, prime numbers, random numbers are not copyrightable.
A collection of chess games is not copyrightable. Annotated part of annotated chess games are copyrightable because a human has added the annotation.
A download of lichess and then selection of games by Elo or something does not copyright the selection. Mechanical selection processes do not create copyrightable material.
Just because content is printed onto the pages of a book does not suddenly give that content “copyright status” if it has no copyright status in the first place. Pi does not become copyright because somebody prints it out in a book.
And, according to the only actual legal person on these forums, collections of numbers that form the weights of a neural net are also not copyrightable because the numbers are produced not by human hand but by mechanical process. They don’t suddenly become copyrightable just because ChessBase transfers the numbers into a CD. Not even if Frederic Friedel transfers each bit by one by laboriously inscribing them with his own personal miniature magnet.

Maybe as a reminder, we all come from different legislative systems, with
different historical backgrounds and principled judgment...

in Germany there is something called 'Schöpfungshöhe', if I parse all works
by Shakespeare and sort the words he used in alphabetical order, I guess this
list is not copyrightable, cos there is no original authorship involved, if
I pick instead in an creative act words from this list and create a new kind of
work by this, it is probable covered by copyright.

--
Srdja

[jp]

And, according to the only actual legal person on these forums, collections of numbers that form the weights of a neural net are also not copyrightable because the numbers are produced not by human hand but by mechanical process.

Who is that person?

[chrisw]

Okay, here's another one.

Same argument.

I don't really understand what you are saying. You may notice by examination that each page in the preview says, "Copyrighted material".
Individual numbers are not copyrightable for obvious reasons. But a book of numbers is copyrightable.

Now, mathematics itself is not copyrightable. So, for instance, I cannot find a new calculus formula and copyright it.

A similar thing relates to chess games. A chess game score is not copyrightable. However, a collection of chess games is copyrightable.

Copyright requires a human hand.
A chess game is not copyright. (Interesting reasons, connected to adversarial creation of game by TWO humans, neither one can claim to have created the game alone).
Pi, e, prime numbers, random numbers are not copyrightable.
A collection of chess games is not copyrightable. Annotated part of annotated chess games are copyrightable because a human has added the annotation.
A download of lichess and then selection of games by Elo or something does not copyright the selection. Mechanical selection processes do not create copyrightable material.
Just because content is printed onto the pages of a book does not suddenly give that content “copyright status” if it has no copyright status in the first place. Pi does not become copyright because somebody prints it out in a book.
And, according to the only actual legal person on these forums, collections of numbers that form the weights of a neural net are also not copyrightable because the numbers are produced not by human hand but by mechanical process. They don’t suddenly become copyrightable just because ChessBase transfers the numbers into a CD. Not even if Frederic Friedel transfers each bit by one by laboriously inscribing them with his own personal miniature magnet.

Maybe as a reminder, we all come from different legislative systems, with
different historical backgrounds and principled judgment...

in Germany there is something called 'Schöpfungshöhe', if I parse all works
by Shakespeare and sort the words he used in alphabetical order, I guess this
list is not copyrightable, cos there is no original authorship involved, if
I pick instead in an creative act words from this list and create a new kind of
work by this, it is probable covered by copyright.

--
Srdja

Shakespeare is dead 500 years or so. Not good example

Ultimately, Germany is covered by EU law and regulations, as is UK for the moment, so final reckoning on any legal decision is pretty much international nowadays. Nations on the whole work to the Berne convention as far as copyright is concerned.

Okay, firstly, he is known for possible creation of new descriptive words, or maybe he is just the first to record already created words. No matter. Anyone can use a word created by someone else, copyright doesn't apply to new words, else development of language would be against the law.

Secondly, if taking words from a corpus of text and sorting them, or listing them was somehow not allowed by copyright law, there would be no dictionaries, and huge word embeddings databases by Google, Facebook, such as 'Glove' and so on would also not be able to legally exist. But they do, downloadable and free of use.

Texts need to be quite extensive to be copyrightable. A book, a poem and so on. The words need to have been creatively assembled by human hand in some meaningful way, and further, it is not the "words" that are copyrightable, it is the expression of a story or narrative by use of words that gets the copyright. The copyright is not in the medium, it's in the massage/message as McLuhan might say.

[smatovic]

Okay, here's another one.

Same argument.

I don't really understand what you are saying. You may notice by examination that each page in the preview says, "Copyrighted material".
Individual numbers are not copyrightable for obvious reasons. But a book of numbers is copyrightable.

Now, mathematics itself is not copyrightable. So, for instance, I cannot find a new calculus formula and copyright it.

A similar thing relates to chess games. A chess game score is not copyrightable. However, a collection of chess games is copyrightable.

Copyright requires a human hand.
A chess game is not copyright. (Interesting reasons, connected to adversarial creation of game by TWO humans, neither one can claim to have created the game alone).
Pi, e, prime numbers, random numbers are not copyrightable.
A collection of chess games is not copyrightable. Annotated part of annotated chess games are copyrightable because a human has added the annotation.
A download of lichess and then selection of games by Elo or something does not copyright the selection. Mechanical selection processes do not create copyrightable material.
Just because content is printed onto the pages of a book does not suddenly give that content “copyright status” if it has no copyright status in the first place. Pi does not become copyright because somebody prints it out in a book.
And, according to the only actual legal person on these forums, collections of numbers that form the weights of a neural net are also not copyrightable because the numbers are produced not by human hand but by mechanical process. They don’t suddenly become copyrightable just because ChessBase transfers the numbers into a CD. Not even if Frederic Friedel transfers each bit by one by laboriously inscribing them with his own personal miniature magnet.

Maybe as a reminder, we all come from different legislative systems, with
different historical backgrounds and principled judgment...

in Germany there is something called 'Schöpfungshöhe', if I parse all works
by Shakespeare and sort the words he used in alphabetical order, I guess this
list is not copyrightable, cos there is no original authorship involved, if
I pick instead in an creative act words from this list and create a new kind of
work by this, it is probable covered by copyright.

--
Srdja

Shakespeare is dead 500 years or so. Not good example

Ultimately, Germany is covered by EU law and regulations, as is UK for the moment, so final reckoning on any legal decision is pretty much international nowadays. Nations on the whole work to the Berne convention as far as copyright is concerned.

Okay, firstly, he is known for possible creation of new descriptive words, or maybe he is just the first to record already created words. No matter. Anyone can use a word created by someone else, copyright doesn't apply to new words, else development of language would be against the law.

Secondly, if taking words from a corpus of text and sorting them, or listing them was somehow not allowed by copyright law, there would be no dictionaries, and huge word embeddings databases by Google, Facebook, such as 'Glove' and so on would also not be able to legally exist. But they do, downloadable and free of use.

Texts need to be quite extensive to be copyrightable. A book, a poem and so on. The words need to have been creatively assembled by human hand in some meaningful way, and further, it is not the "words" that are copyrightable, it is the expression of a story or narrative by use of words that gets the copyright. The copyright is not in the medium, it's in the massage/message as McLuhan might say.

Hehe, I am not going to argue here for Albert Silver that he did enough creative
work by human hand to create his neural networks...

I only wish to state that if we start to copyright or patent numbers or a set
of numbers science is in trouble.

--
Srdja

[Ozymandias]

Texts need to be quite extensive to be copyrightable. A book, a poem and so on. The words need to have been creatively assembled by human hand in some meaningful way

Computer programs already produce (compose?) music. You can get a drawing out of a picture and have it in the style of a dead painter. How far away is literature? What will happen then? Will those texts not be considered to have meaning, just because there's no human author?

To us, weights are data, because we can't make any other sense out of them, but it doesn't mean weights are data, they have a meaning beyond our comprehension.

[chrisw]

Okay, here's another one.

Same argument.

I don't really understand what you are saying. You may notice by examination that each page in the preview says, "Copyrighted material".
Individual numbers are not copyrightable for obvious reasons. But a book of numbers is copyrightable.

Now, mathematics itself is not copyrightable. So, for instance, I cannot find a new calculus formula and copyright it.

A similar thing relates to chess games. A chess game score is not copyrightable. However, a collection of chess games is copyrightable.

Copyright requires a human hand.
A chess game is not copyright. (Interesting reasons, connected to adversarial creation of game by TWO humans, neither one can claim to have created the game alone).
Pi, e, prime numbers, random numbers are not copyrightable.
A collection of chess games is not copyrightable. Annotated part of annotated chess games are copyrightable because a human has added the annotation.
A download of lichess and then selection of games by Elo or something does not copyright the selection. Mechanical selection processes do not create copyrightable material.
Just because content is printed onto the pages of a book does not suddenly give that content “copyright status” if it has no copyright status in the first place. Pi does not become copyright because somebody prints it out in a book.
And, according to the only actual legal person on these forums, collections of numbers that form the weights of a neural net are also not copyrightable because the numbers are produced not by human hand but by mechanical process. They don’t suddenly become copyrightable just because ChessBase transfers the numbers into a CD. Not even if Frederic Friedel transfers each bit by one by laboriously inscribing them with his own personal miniature magnet.

Maybe as a reminder, we all come from different legislative systems, with
different historical backgrounds and principled judgment...

in Germany there is something called 'Schöpfungshöhe', if I parse all works
by Shakespeare and sort the words he used in alphabetical order, I guess this
list is not copyrightable, cos there is no original authorship involved, if
I pick instead in an creative act words from this list and create a new kind of
work by this, it is probable covered by copyright.

--
Srdja

Shakespeare is dead 500 years or so. Not good example

Ultimately, Germany is covered by EU law and regulations, as is UK for the moment, so final reckoning on any legal decision is pretty much international nowadays. Nations on the whole work to the Berne convention as far as copyright is concerned.

Okay, firstly, he is known for possible creation of new descriptive words, or maybe he is just the first to record already created words. No matter. Anyone can use a word created by someone else, copyright doesn't apply to new words, else development of language would be against the law.

Secondly, if taking words from a corpus of text and sorting them, or listing them was somehow not allowed by copyright law, there would be no dictionaries, and huge word embeddings databases by Google, Facebook, such as 'Glove' and so on would also not be able to legally exist. But they do, downloadable and free of use.

Texts need to be quite extensive to be copyrightable. A book, a poem and so on. The words need to have been creatively assembled by human hand in some meaningful way, and further, it is not the "words" that are copyrightable, it is the expression of a story or narrative by use of words that gets the copyright. The copyright is not in the medium, it's in the massage/message as McLuhan might say.

Hehe, I am not going to argue here for Albert Silver that he did enough creative
work by human hand to create his neural networks...

I only wish to state that if we start to copyright or patent numbers or a set
of numbers science is in trouble.

--
Srdja

We do agree here

[chrisw]

Texts need to be quite extensive to be copyrightable. A book, a poem and so on. The words need to have been creatively assembled by human hand in some meaningful way

Computer programs already produce (compose?) music. You can get a drawing out of a picture and have it in the style of a dead painter. How far away is literature? What will happen then? Will those texts not be considered to have meaning, just because there's no human author?

Such compositions may well have meaning, but as copyright law stands at present, they will not be copyrightable. Copyright is something that lawfully belongs to a person. A machine is not a person. As yet.

To us, weights are data, because we can't make any other sense out of them, but it doesn't mean weights are data, they have a meaning beyond our comprehension.

There is information stored inside the data, and a suitable decoding engine can make sense of the data, and show it in human readable way. Lc0.exe for example.

The story, narrative, whatever we call it, that is represented in the NN weights data is not copyrightable

[Ozymandias]

Copyright is something that lawfully belongs to a person. A machine is not a person

But belongs to a person, a group of persons or a company.