Bit of a thorny one - we equally want to encourage the transfer and restoration of old works - and owning a few digitally restored pieces of 1920s cinema, I can appreciate there's a cost involved in that - and it's questionable whether the public purse would pay for restoring all work in the public domain into a usable state, before it decays.
Previously, the fact that these costs could typically be recovered through selling a physical edition - i.e. it doesn't matter if work is out of copyright if only one firm is actually printing a book.
Remove the ability to make money back, and we're back to the primary question - how do we fund this? Obviously, the academic community has failed (lacking the time/money) to scan these works ahead of a private publisher (for whom the ability to extract money from academics is acting as a motivator).
[Given that eventually the money is going to come from the same place, of course we'd be better off funding the universities to do the scanning directly]
Is it better for a work to be in the public domain but not actually available, or vice versa?