(Edit: I misread the original claim, my fault. So I'm fixing the text here.)
I think it's important to look at authoritative documents, especially since copyright duration is absurdly complicated. You can learn more about copyright duration in the US here: https://www.copyright.gov/help/faq/faq-duration.html
Circular 15a "Duration of Copyright" from the US Copyright Office ( https://www.copyright.gov/circs/circ15a.pdf ) says in heading "Automatic Renewal and Voluntary Registration" that:
* "Mandatory Renewal Works originally copyrighted between January 1, 1950, and December 31, 1963. Copyrights in their first 28-year term on January 1, 1978, still had to be renewed to be protected for the second term. If a valid renewal registration was made at the proper time, the second term will last for 67 years. However, if renewal registration for these works was not made within the statutory time limits, a copyright originally secured between 1950 and 1963 expired on December 31 of its 28th year, and protection was lost permanently."
* "Works originally copyrighted between January 1, 1964, and December 31, 1977. Congress amended the copyright law on June 26, 1992, to automatically renew the copyright in these works and to make renewal registration for them optional. Their copyright term is still divided between a 28-year original term and a 67-year renewal term, but a renewal registration is not required to secure the renewal copyright. The renewal vests on behalf of the appropriate renewal claimant upon renewal registration or, if there is no renewal registration, on December 31 of the 28th year."
So there is a cutoff before 1964. However, it's not at all clear to me how many times the copyright holders didn't renew their copyrights. I would be unsurprised if the big publishers did typically renew them. If a book was published in 1950 and it was renewed, then the copyright would continue until 2045 (1950+28+67).
I've only listed 2 cases (just before & just after 1964), but there are actually many more cases and it's all quite complicated.
I think copyright duration is grossly overlong, far in excess of what is needed to get people to create works.
> So there is a cutoff before 1964. However, it's not at all clear to me how many times the copyright holders didn't renew their copyrights. I would be unsurprised if the big publishers did typically renew them.
That doesn't sound that surprising to me either, but I don't think that matters for the claim that most are PD now. My intuition would be that big publishers wouldn't be responsible for a large percentage of the published titles in that time period. Certainly, from the standpoint of units sold they would be, but not distinct works.
There's a lot more niche, technical, academic, low-end, etc works even though most people consume the mass-market popular stuff. I would expect that all those long-tail things to be the ones that never got renewed, and to be in the majority as far as titles go.
I think it's important to look at authoritative documents, especially since copyright duration is absurdly complicated. You can learn more about copyright duration in the US here: https://www.copyright.gov/help/faq/faq-duration.html
Circular 15a "Duration of Copyright" from the US Copyright Office ( https://www.copyright.gov/circs/circ15a.pdf ) says in heading "Automatic Renewal and Voluntary Registration" that:
* "Mandatory Renewal Works originally copyrighted between January 1, 1950, and December 31, 1963. Copyrights in their first 28-year term on January 1, 1978, still had to be renewed to be protected for the second term. If a valid renewal registration was made at the proper time, the second term will last for 67 years. However, if renewal registration for these works was not made within the statutory time limits, a copyright originally secured between 1950 and 1963 expired on December 31 of its 28th year, and protection was lost permanently."
* "Works originally copyrighted between January 1, 1964, and December 31, 1977. Congress amended the copyright law on June 26, 1992, to automatically renew the copyright in these works and to make renewal registration for them optional. Their copyright term is still divided between a 28-year original term and a 67-year renewal term, but a renewal registration is not required to secure the renewal copyright. The renewal vests on behalf of the appropriate renewal claimant upon renewal registration or, if there is no renewal registration, on December 31 of the 28th year."
So there is a cutoff before 1964. However, it's not at all clear to me how many times the copyright holders didn't renew their copyrights. I would be unsurprised if the big publishers did typically renew them. If a book was published in 1950 and it was renewed, then the copyright would continue until 2045 (1950+28+67).
I've only listed 2 cases (just before & just after 1964), but there are actually many more cases and it's all quite complicated.
I think copyright duration is grossly overlong, far in excess of what is needed to get people to create works.