I think it mostly has to do with provenance. The third shift prints usually don’t make it into contemporary bibliographies [1] and since the printer is untrustworthy, you can’t be sure if it’s a real third shift first edition or the tenth impression of some typesetting that the printer forgot to recycle. If we were talking about the Gutenberg Bible or say, Origin of Species, it might be worth doing original research and chemical analyses because even a third shift copy could be worth more than its weight in gold, but for >>99% of books it’d be a losing proposition since few books are worth tens of thousands of dollars.

That said, if the paper and binding quality are the same between shifts and there’s no identifying mark like unique numbering on each copy, there’s no telling how many first editions in circulation are really third shift printings. So few of the books survive the centuries that I’ve never heard of more first editions showing up than the publisher admits to printing to bibliographers.

Edit: They’re not really “worthless” but old and rare books might see one genuine first edition sold every few years to some rich person with a hobby. Many if not most of these books are bought and sold by knowledgeable collectors so if a client is going to drop thousands on a book, they're going to buy the copy that is most likely to be genuine, making any also rans effectively worthless because there’s not enough of a demand to force buyers to settle for riskier copies.

[1] because they’re unauthorized and the publisher doesn’t talk about them to the bibliographers

To be fair, it’s already odd that first editions are worth something. Collectors are just weird persons who will pay extraordinary sum for things of marginal interest. From there, it’s not surprising that the way value is ascribed itself doesn’t make much sense.