Hacker News new | past | comments | ask | show | jobs | submit login
Blind people won the right to break eBook DRM, but they'll have to do it again (wired.com)
318 points by pabs3 on Nov 8, 2021 | hide | past | favorite | 69 comments



> Although the exemption protects people who circumvent digital copyright protections for the sake of accessibility — by using third-party programs to lift text and save it in a different file format, for example — that it's even necessary strikes many as a fundamental injustice.

This is not only an injustice, but also a stupidity. What I do with something I paid for should concern only me as long as it's within my personal bounds (no redistribution, no profit, no business, etc.) Breaking DRM for personal use should not and cannot be illegal.


Yeah. This is what happens when laws are bought and paid for by corporations through lobbying. Their concerns just override all common sense.


Nobody really cares what people do locally on their own. But companies do care when people share this local work with others. And any tool for accessibility can be used for this in one way or another.

This is simply a problem of two opposing forced where both have legitimated reasons for their demand. And so far the weak side is losing. It would be better to find some compromise. Having tools and readers with official support for blind people should be the obvious solution. Why do they even need to fight against the publishers, instead of working with them together?


If companies care when people share the work with others means that the act they want to prevent is breaking copyright this way, and it becomes a question of enforcement.

I understand its basically impossible to enforce all cases once they have the raw digital text, but does that mean we should ban OCR software and limit the access to high quality scanners in case they are used for copyright infringement? That seems an absurd cost to protect the existing business model. But is also the conclusion to banning DRM circumvention efforts. I think instead we can just go after people selling large number of copies like we always have and publishing will survive.

If your business model requires you to break other people’s property rights to something that you ostensibly sold them, in order to enforce it, it usually isn’t a viable business model! Except when you have a massive army of lawyers apparently.


> Breaking DRM for personal use should not and cannot be illegal

Agreed, including fro other people's personal use.

Companies should be allowed to protect data by copyright or DRM, but not both, i.e. encasing something in DRM would remove its copyright.


Much like one cannot have a patented trade secret. The grant of patent requires disclosure, and using the court system to publicly resolve disputes.

If one uses technical means to protect information, parent poster proposes that courts should decline to protect it under the copyrights regime as well. Certainly, they should not protect both the information and whatever technical device is used to do private enforcement.

You can either use the public protection regime, or your own private protection device, but not both.

The issue at hand is that copyright implicitly includes an agreement to allow the work to pass into public domain after the copyright term expires. Technology that prevents that is anathema to the agreement. At the least, such schemes should fail-open, such that archived copies may be read as plaintext when the copyright term expires.


I don't really track this argument. DRM generally exists to make it easier to protect copyright, no? They go hand in hand.


The argument doesn't really track but it's very much worth examining DRM and its relationship with/effects on copyright and the rights of consumers, anyway.

DRM is frequently used not just to protect copyright, but to also completely control access beyond the purpose of copyright. See the recent Denuvo snafu for an example of this being done (presumably) by accident, noting that it can also be done on purpose to e.g. plan obsoletion. It doesn't just control copyright, it controls your actual rights to access content you legally own and hinges that access on running a service in perpetuity, which is not the least bit realistic. It's digital rights management but treats those rights like a privilege instead.


Think about patents. If you get a patent, you are legally protected from people stealing your patented idea for the lifetime of the patent. However, in exchange for that protection, you must make the idea public. In exchange for the protection of law, a patent holder sacrifices the protection of secrecy.

A simmiliar model could be applied to copyright, in exchange for legal protections, you must forgo technical protections.


This is terrible, and while IANAL I’m confused why it doesn’t fall under ADA requirements to provide accessible versions of materials - especially school materials.

Also, can people really claim with a straight face that running a book through text-to-speech is the same as a real audiobook (that’s often a voice actor, right?)?


I listened to all of Worm (https://parahumans.wordpress.com/) with TTS and while it wasn’t ideal, it worked out well enough that I got a full understanding of the story. I’ll note that I did this before the fan podcast to read the whole story lol. That’s a much better option now.


> Also, can people really claim with a straight face that running a book through text-to-speech is the same as a real audiobook (that’s often a voice actor, right?)?

TTS is quickly improving. While it's not quite there yet, there are already ways to have any text read with anyones voice.

There also comes a point where even if it isn't perfect, it doesn't matter to the consumer who just wants the book read to them. Meanwhile, the publishers want the revenue from selling both versions of the book.


It's not really all that grating anymore, just a bit dry in delivery. On a separate note, I think it's really dumb that voice actors can't easily record works without explicit, particular permission from an author. They should license it from the author with some sort of profit sharing per copy, but the current system means there is often only one audiobook version unless a work is public domain. And when I say license it, I mean some sort of form contract available to anyone without huge legal fees.


for books popular enough to get that treatment sure, I used to do volunteer reading to make audiobooks but had to quit when someone handed me the most pseudoscientific rubbish to read - it really felt like taking advantage - some book selling a lifestyle advice thing that would lead to immortality!!


LibriVox is a well-known project in this space that by and large focuses on worthwhile stuff, not pseudoscientific rubbish.


Some years ago, I wrote the code specification for our ebook conversions. I made a priority of specifying good semantic structure, employ ARIA tags where needed, etc. I work for a university and there was yet no mandate for it, but it seemed only a matter of time. In the long run being a little ahead of the curve aid paid off, but now I need to go through and revise things to account for changing standards, and my own improved (I hope) understanding of what's needed.

I'm also reminded of an incident years ago when I was working in a New York, and an agent (for a popular, well-paid writer) called our contract department from my desk phone to insist they remove the clause allowing the LOC to make audiobooks for the blind. As agent crimes go, it might not be the worst, but it always struck me for how egregiously petty it was.


Fuck having to ask for the "right". If they bought their digital books, they should absolutely be free to do whatever they like with them for personal usability. I can understand the legal rights position here, revolving around bringing attention to the issue of these idiotic rules, but if I buy ebooks, i'll just go ahead and strip their DRM without giving a damn who says no. The absurdity of DRM laws is really made visible again.


> but if I buy ebooks, i'll just go ahead and strip their DRM without giving a damn who says no

Yeah, dream on. How long do you think this option will stay available to you? I sometimes get the impression that intelligent technical people on HN can't see ahead 5 minutes into the future, because being a peon without rights is something that conflicts with their self image as some technical sophisticate not subject to the same insults by our overlords as lowly non-techies. But your relative technical sophistication just means you'll lose some basic control over your privacy, data and possessions at t+ε rather than t.


> How long do you think this option will stay available to you?

This is a significant concern. We're seeing more and more non-free hardware in the market. Hardware designed to stop us from having access to data and from running our own software. Our computing freedom must be sacrificed in order to have any hope of enforcing these antiquated laws.


> Yeah, dream on. How long do you think this option will stay available to you?

For as long as my reader can decrypt ebooks locally.


Fahrenheit 451 style, right?

Because local vs cloud doesn't matter squat if all the devices your have locally are locked down completely and controlled by someone else.


For books the analog hole is probably widening with better ocr now than ever.


The last remaining analog hole may soon be people. Recording devices are basically already digital, and there is no fundamental difficulty in embedding enough pattern recognition to ensures that you don't replicate material that's either thoughtcrime or the valuable IP of Elsevier, RIAA or some similar benefactor of humanity. Photoshop reportedly already refuses to work on scenes that contain money.


Money is one thing, but you can’t exactly autoblur all text without a mass uproar. And at least for now I don’t think it’s feasible to keep enough updated storage and processing around photo devices to recognize temporary pics of all copyrighted texts either - even with heavily optimized hashes. But, I could be wrong in this assumption.


As long as the content is presented to a human for consumption, I think it will be possible to duplicate it, even if it is just recording the screen with a camera or something.

Even "unpublished" works can be recreated by those with talent, like Mozart and Allegri’s Miserere.


Until you have a camera that detects watermarks and refuses to record. Its already there with cinavia which I guess has functional workarounds now, but it took, what 3-4 years (?) before a remover showed up that didn't garble the audio.

Its not at all hard to believe that in 20-30 years when all the non-drm'ed version are sitting in landfills the analog "hole" will be closed because capture or playback devices will simply refuse to function when they detect watermarks when interacting with unsecured channels.

I acted when ultra-hd was "cracked" and grabbed one of the friendly drives, but its not been particularly useful because there seems to be far less effort in the public to assure that ultrahd disks are readable (look up the details) so its more likely to "miss" these days. The point being that its a giant game of back and forth, and while the "crackers" win everytime they crack something, they are mostly hobbiest fighting against corps with billions to employ engineers who spend their entire day coming up with ways to make it more difficult. Given few of us have electron microscopes and giant supercomputers, it seems really likely they will eventually win given that they learn to compensate every time something is cracked.

Plus, in the past it was pretty easy to buy PC's/etc that gave you full control, these days even with secure boot disabled, you might not be able to get into the lower levels of firmware because the machine simply refuses to accept unsigned updates. Flash it by hand and the bootstrap code integrated on the CPU die refuses to accept it.


> As long as the content is presented to a human for consumption, I think it will be possible to duplicate it,

Why do you think so? If you can tell the difference between a family scene and a screening of a Hollywood film, why can't your camera? It doesn't have to perfect, and you can gradually tighten things down, which is exactly what has been happening for some time.


> If you can tell the difference between a family scene and a screening of a Hollywood film, …

That's a big "if". I wouldn't guarantee to always be able to tell the difference. And there is a lot of non-Hollywood content which looks a lot more like something out of a Hollywood film than just "family scenes". (People do make their own films, you know, not just recordings of family events.)

> … why can't your camera?

That's asking a lot out of the kind of computing hardware you can reasonably fit into a camera. Even if you assume something like a smartphone with much more computing capacity than a standalone DSLR or mirrorless camera (and a correspondingly higher power budge / shorter battery life), you're not going to be able to train it to recognize every bit of Hollywood content and reliably distinguish it from non-Hollywood content offline. (Online recognition where the camera can upload content to the server for pattern recognition is a different matter, but something like this would presumably be done offline anyway.) The information requirements would be on the order of having all that content downloaded to the phone—never mind the processing involved.


Forever. DRM is created under the cryptographically insane belief that one can treat a person as an eavesdropper at any time after they have been an authorized recipient.

That is the unsolvable problem that guarantees DRM will never be able to create a perfect protection device that allows the company to stuff the genie back into the bottle, or the cat back into the bag, at will.


Asking as a non-sophisticate who has a mental block on most math education beyond 2nd grade: what does t+ε mean?

edit: HN would rate limit me if I tried to reply to each explanation with thanks, so: thanks!


Current (t)ime plus some hand-waiving (ε) Epsilon of change in time.


t = some point in time

t+ε = some point in time a tiny fraction later


Time + Elapsed Time. AKA not now, but eventually


Epsilon stands for a "arbitrarily tiny bit" in calculus.


Well, I can't predict the future, but so far I've managed to bypass every bullshit DRM control I've come across for text content and ebooks without too much trouble. It's mainly a question of choosing your mediums and hardware a bit more selectively. This doesn't make me blind to anything beyond the next 5 minutes and personally, i'm far from a technical person. Yes, the tech and media giants will keep looking for ways for screw people into becoming captive consumers who "buy" things that they don't even really own, but technology has so far done a remarkably good job of counteracting much of these attempts if one bothers to use it with a bit of creativity.

edit: clarity


Indeed. What an indignity. People "buy" books only to be forced to beg authorities for access to the items they paid for. It should be the other way around. It should be illegal to prevent this access.


Where can I get these DRM free eBooks?


The Calibre ebook managemet app has plugins available that can strip both Amazon and Adobe DRM.


Or it used to. Last time I tried it wasn't able to do Amazon anymore.


It still works for Amazon but it doesn't really work for books in their newer KFX format. It's best to use an older version of Kindle for PC or use their download and transfer to USB if you have a physical eInk Kindle to get an AZW3 file which should unlock no problem.


Installers for older versions are becoming increasingly elusive, I've found. Anything newer than Kindle for PC v1.26 and DeDRM can't break the encryption. Installers for v1.17 or v1.19 can still be found, but on Linux those versions can't connect to Amazon because one of the SSL certs used by the app has expired.

That leaves v1.21 as the "sweet spot", but most of the sites that used to host that installer EXE have mysteriously disappeared.


You can get the missing certificate from the links in https://forums.linuxmint.com/viewtopic.php?f=47&t=342186&uid... .


That workaround gets you the text but it's missing a lot of the recent typographic improvements. It's fine for extracting text for TTS, but isn't ideal if what you are trying to do is maintain a DRM free archive of books you've purchased.

It's like saying you can break BluRay DRM by recording the screen with your phone camera.


What kind of typographic improvements do you have in mind?


Stuff like drop caps, kerning that isn't terrible, hyphenation, and better word spacing. None of it matters if you are feeding the text to a speech synthesizer, but it can make a difference when reading.


Some of that stuff you can get back by using an ePub version and reader from the AZW3 file which can typically be done using KindleUnpack. Some of it is less the file format and more the rendering engine used and they simply haven't bothered to backport it to their AZW3 renderers. Amazon does AFAIK precompute some stuff like hyphenation but not everything.


The point is the KFX format does have all that natively.


And my point is that some of it not native (it's handled by the renderer) and it's possible to get back or similar enough (precomputing hyphenation).


If you're willing to do a bunch of work, you may be able to recreate something close to the original KFX. AFAIK you won't get kerning though. That seems to be tied to the renderer used for KFX files.

The other part is that there isn't a single definitive KFX file for a title. If you have an Oasis and I have a Paperwhite and we both download the same book, I believe we get different KFX files. There's no reason to doubt that there will be future Kindle devices that are even better with respect to typography and so you will need to re-download your library for that device.

Today, KFX can sometimes be De-DRM'd but it's gotten to the point where it's a cat and mouse game. It works for a while then Amazon changes something and breaks it. I don't doubt that essentially uncrackable DRM is achievable.

The real fix is legislative.


Also does a very good job of metadata management and format conversion. I don't know of a better solution. With Calibre you can buy wherever you like and maintain the format and metadata system you prefer.



Libreture maintains a list of DRM free stores [1].

[1] https://www.libreture.com/bookshops/


That's a good list. I've found some stuff I was looking for at Bloomsbury. Tor is probably good if you're into sci-fi/fantasy. But if you're looking for a specific book in English it's typically very hard to find a place to pay for a DRM-free e-book.

A search engine that covers all the little sites selling DRM-free books would be nice!

(Publishers of Norwegian books fortunately tend to have just do watermarks, which is fine – you can read the book on whatever device you want, and your copy has a little note saying "ex libris Your Name" or something)


Kobo [1] and Google Play Books [2] also sells some books DRM free. You have to check the book details to see if it says 'DRM free'.

[1] https://www.kobo.com/ [2] https://play.google.com/store/books


the pirate bay & friends.


libgen for books. The pirate bay theoretically covers everything, but good luck finding books there.


Try these Libgen proxy/mirror links:

http://libgen.gs/ (NEW – November 2020)

http://gen.lib.rus.ec

http://libgen.rs – (NEW – October 2020).

https://libgen.is – (Working Again – May 2021).

http://libgen.st – (Re-directs to www.libgen.rs).

http://libgen.li

https://libgen.lc

http://libgen.io/ (No longer working).

https://libgen.be/ (NEW – September 2020 – Search proxies to www.libgen.rs).

https://libgen.nl/ (NEW – September 2020 – Search proxies to www.libgen.rs).

If you are having trouble reaching with above Libgen URLs, try http://93.174.95.27 or http://185.39.10.101.


I've literally always just grabbed the latest working URL from the Wikipedia article's sidebar.

https://en.wikipedia.org/wiki/Library_Genesis


Libgen is great for older literature and non-fiction books of all kinds, but for modern fiction it sucks.


Really? What have you struggled to find? I've found libgen enormously helpful.


Maybe it's just my pop fiction preferences but if I search for any sort of modern horror or sci fi fiction, I just can't seem to find any content. Pirate bay on the other hand? No problem. Like I said though, almost any modern non fiction and a great deal of older fiction literature is definitely there, so I'm definitely not criticizing libgen. It's mostly wonderful, though one worries about malware luring in downloads, but on this maybe im just being paranoid.


Besides TPB, is there a libgen for audiobooks? Audible blocks lots of books here and the prices are honestly quite steep. TPB often doesn't have what I was looking for, not even something obscure.


Not actually a joke: audiobookbay.ws


The best source for ebooks if definitely eMule


Z-Library is the best.


>Even ebooks that are formatted correctly for TTS can have other issues. Math and science are the worst. Textbooks may be formatted for 100 percent accurate text-to-speech, only to falter when it comes to formulas, equations, charts, and tables. Those are typically rendered as images in an ebook, which would require publishers to take an additional step to record "alt text" for each individual figure—audio that would describe the image once encountered by a screen reader.

My now wife and I spent hundreds of hours during undergrad converting text and graphs to plaintext for a grad student. Truly nasty formulas, with multiple integrals, fractions with many factors above and below the line, exponents all over.

While the legality was questionable at the time, I broke the DRM on the PDF of the book, copied everything to Sublime, removed line breaks, formatted footnotes with [1] formatting, labeled graphs.

Then I fired up GIMP and translated the images. When embossed on paper, braille is much larger than normal text, so I had to greatly simplify things.

The frustrating part was, on math books, there's a good chance it was LaTeX to begin with. The format we converted formulas to was almost identical to LaTeX.

Additionally, surely for undergrad level classes, multiple colleges were each paying undergrad students $1,000+ per course to make the materials accessible. A website where School 1 could see that School 2 had already converted Calculus 1 to braille and allowing payment to reimburse a proportional amount of the cost would be incredible helpful and maybe legal. In essence, $1,000 would be paid by School 1 to do the initial work, School 2 reimburses $500, then School 3 pays $333, split to the other two schools so ultimately, they're all in it for $333 rather than $1,000.

----

I also worked with another student to build a molecule generator for 3d printers. Extensive use of the Blender Python API. Each bond type was a different shape connector, and the periodic table was divided by row and column to form a unique combination of scale and shape.

MacMolPlt [1] generated plaintext files that contained the position of each atom, bond connection types, and a list of coordinates for the electron clouds which could also be generated on top of the molecular structure, all exported to .STL for 3d printing.

It was a neat project, definitely opened my eyes to disabilities, and I became a more empathetic person because of it. Through some weird stuff with funding, it's sort of in copyright gray area. The disability office was paying to make things accessible for the student, not necessarily to build a tool that would be released to the greater world. It wouldn't be hard to replicate though. Back in 2011, I knew every 3d printer on campus and was kinda begging to print stuff while we tried things out. Now printing is trivially cheap and capacity is abundant - it would be 10x as easy and helpful.

[1] https://www.swmath.org/software/7039


If a book is printed, and someone asks for a braille version, are you required to provide it?


Great. Now how about the rest of us?


Legally, I guess. Because thankfully, it's not hard to create PDF copies of any ebook even in 2021.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: