I've got 1050 books all catalogued in LibraryThing. I find it quite worthwhile – I can search my library, useful if you have an extensive reference collection on a particular topic. And if I'm out at a bookshop wondering if I have a book already, I can check! :)

A true collector would buy the book regardless. The duplicates are part of the character of the collection. If you like a book that much, gift the extra.

With a barcode scanner it shouldn't take you long to catalogue even thousands of books (disregarding the pre-barcode ones).

The barcode on paperbacks is usually UPC, not ISBN (unless it has changed in recent years). And therein lies a big problem: UPCs get recycled. I was bitten by this when I scanned a bunch of old Sci-Fi books, only to discover that totally different titles got assigned to those UPCs later.

By coincidence yesterday evening I decided to start scanning my books with an Android application I'd come across during random surfing.

I figured if I picked a bookshelf at random and could enter all the books without a problem I'd do the whole house. As it turned out a random bookshelf had about 60% success rate.

I knew that I have some books without ISBN entries, and some that have got barcodes covered in stickers, etc, but I didn't expect it to be such a hard job. I guess every decade or two I try auditing books and I never get very far.

I should just pay a student to enter them into an online CRUD application and be done with it..

When I scanned my books with LibraryThing (more than a decade ago now) with one of those CueCats I had lying around as my barcode scanner at the time (now the LT iOS and Android apps can act as a barcode scanner), the primary data source was (and still is, I believe) Amazon's database which has a very good idea of paperback UPC codes to ISBNs from their used book sales. In the average case you'd just have to pick between two or three alternatives, and in the worst case that Amazon had very few used sales of that book just text search some other source (and LibraryThing is rife with all sorts of really cool catalog data sources).

Unfortunately, ISBNs also get recycled. They shouldn't be (I think it's actually against the ISBN rules), but it happens.

That's mostly older commercial paperbacks that have UPC instead of ISBN barcodes. I think nearly everything published since the 90s has an ISBN barcode.