3 ms·
It seems to me as though an interlibrary alliance could pool resources to create their own digitizations of print-version books. They already pay extra for the
by logfromblammo 7y ago
It seems to me as though an interlibrary alliance could pool resources to create their own digitizations of print-version books. They already pay extra for the library edition of a print book. For that higher price, they are supposed to get more durable covers and paper. For a more expensive e-book, what do they get? Self-destructing DRM. Why would anyone agree to that--especially a library?
If it is legal to lend out that physical copy, what is the difference, philosophically, between taking that copy off the shelf and placing it in a reader-proxy robot that can turn pages and focus its camera on them? What is the difference between that and flipping through static photographs of the pages while the physical copy is kept in a "digital lending" vault? What is the difference between that and flipping through pages of OCR-digitized text that correspond to the photographs? And if the physical copy is permanently placed in a defunct mine and no one ever has physical access, what prevents the mine from lending the digital text to any partner library, so they can in turn lend the single digital copy to a person?
If the publishers want to play stupid games, award them stupid prizes. Buy the paperback/cheapest edition of the book, shear off the spine, scan both sides of each page in a machine at 1200 dpi, re-bind the pages (plus paper cover) into a corrugated plastic cover, catalog it, and put it in permanent storage. Now OCR the 1200 dpi images, downsample them to 96 dpi, and package the text and low-res images into an e-book. When a user thinks the OCR made an error, the low-res image can establish a page and location, and a rectangle of the high-res image can be delivered from a web interface. The user's guess at the text content can then be verified or rejected by AI--easier than OCR from scratch--and used to update the book file.
All the lending software needs to do is verify that no more than N natural persons are looking at the digital files for N real copies at the same time. The physical copy could always be retrieved from storage and shipped to the borrower, after all. The legal copyright model has nothing to do with the time and money costs of physical shipping. A property title can move faster than the speed of light, even when the property itself is immobile. Even if the lending model doesn't work, sell the beneficial interest in the book for $20.00, and then pay $19.99 when they sell it back, the $0.01 representing physical-book storage fees and bandwidth costs.
Whether or not the end-user can copy the files they receive is immaterial. I can already borrow a library book and scan every page at home. That would be me violating the copyright, not the library. The burden on the library is to say, "please don't do that, unless it is for legal, fair-use purposes". And the burden of policing it is on the owner/assignee of the copyright. If going along with the copyright system voluntarily is reasonable, people will go along with it voluntarily. If not, people will pirate. This may seem tautological, but increasing forced compliance will reduce voluntary compliance.
- shantly 7y agoYour scheme requires copying the work, probably several times, and distributing some of those copies, so it won't fly. I'm pretty sure the fact that digital lending necessarily makes tons of copies of the work—consider live server disk, backups (possibly many copies there), copies in cache across who knows how many devices on the Internet, the copy on the user's device, copies in RAM all over the place, et c., such that there might well be twenty copies of a book before it reaches the eyes of a borrowing reader—is why publishers are able to put license terms on ebook lending in the first place, since technically the whole process, practically speaking, involves copying the book a bunch of times.
- logfromblammo 7y agoWhat we know to be a copy by common sense, real world interpretation of the word "copy" is not necessarily what copyright law considers to be a "copy" for the purposes of enforcing copyright laws. Is holding a book up to a half-silvered mirror making a copy of it? Is looking over someone else's shoulder as they read making a copy? Do the activation levels in a person's visual cortex count as a copy? It may be useful to keep hairs unsplit, forget about the quantum mechanics of copying and focus on the observers. If there are two files with identical content on a filesystem such that only one user is authorized to read them, how many copies exist? To me, that's two copies in fact, but one copy by copyright law, in that the second real-life copy is not separable from the first for the purposes of commerce. The act of changing the access permissions on the second identical file would be what makes it a copyright-copy, not the act of duplicating the bits.
- shantly 7y agoIn that case why are libraries bothering with jumping through these hoops and paying extra money for expiring ebooks anyway?
- logfromblammo 7y agoThey want to be "good citizens" in the book-publishing ecosystem. They follow along with the publishers' schemes, to ensure that the channel for public access to the material remains open. They don't want to risk the possibility that there might be knowledge out there that an ordinary person might be barred from accessing via the public library. And as long as the publishers are reasonable, even when charging a higher library price, that's cheaper than libraries setting up a different infrastructure. The mission of most libraries is about access and public service, rather than raw price. The publishers can abuse that.