It was often brought up in conversations with universities and their libraries - having them made things more convenient. The books were selling for a year or so without ISBNs, no problems either way!
ISBN is intended for the marketing of new books and publishers are allowed to reuse the ISBN if a book goes out of print.
I used to work at the library at my Uni and did some analysis of what we had in our catalog and found quite a few books that shared the same ISBN from South End Press which I thought was funny because I had a friend who grew up next door to Noam Chomsky and was friends with the people who ran South End Press. We were talking with them about web publishing in the the early 1990s and found the people there were really excited about something called Futuresplash which eventually became Macromedia Flash.
I think they didn't want to pay for new batches of ISBN numbers and maybe it was colored with a desire to "stick it to the man".
Also at home in my collection I have a lot of books that are from the 1950s and 1960s which have an LCCN but don't have an ISBN so the ISBN would not be a good primary key for a personal book database though I think the LCCN would be better.
Books not published in the US often don't have a LCCN, so not great as a unique ID.
Also if you're actually building a database, never use a meaningful data element like ISBN or LCCN as the primary key. What happens if you have multiple copies, for example?
That's a good point. I should probably print a QR code that is equivalent to a type 4 UUID or a snowflake id and stick it on each individual book. Now that I think of it, I have two copies of Nixon Agonistes which I mentioned in a comment here yesterday.
