Skip to main content

Sony Reader for Android’s New Support for Epub3 Makes Me Think Epub3 is DOA

unnamed No one really knows why Sony is still in ebooks (not even Sony) but in spite of the fact they are more disorganized that your average group of 6-year-old girls they do still manage to pull off surprising work.

Sony rolled out a new update yesterday for the Sony Reader for Android app. according to the listing in Google Play, the new version of the app now supports Epub3.

You can download the updated app and the extension from Google Play. Be sure to get the extension; that is what (theoretically) enables the support for audio and video, interactive content, and other Epub3 features.

Or at least that is what the listing says. I have downloaded the app (and the required extension) and I’m not really sure that the Epub3 support is completely functional.

Oh, it works about as well as on the iPad (and definitely better than Helicon’s Gyan reading app). And the embedded audio and video work just fine and the videos can even play full screen on my HTC smartphone.


But video and audio isn’t Epub3; various reading apps supported it since 2010. No, I want the interactivity and MathML to work, and they don’t. Some of the embedded image and language support is also effectively nonfunctional.

What’s more, the formatting of the Epub3 ebooks looks like shit on the 4″ screen of my smartphone (htc x710a).

2013-02-23_08-33-33From what I can tell it’s not entirely the fault of the app. I’ve downloaded several ebooks from the IDPF collection of Epub3 sample files, and a couple are frankly unusable. It looks like they have the page margins hard-coded margins built-in, and the ebook designers clearly did not think about how their ebooks would look on screens smaller than the iPad.

You can see the worst offender at right, but there are a couple others that are nearly as bad. That screenshot shows the ebook with the interactive test widget (it doesn’t work, BTW). You’re going to have to click on the image in order to really appreciate how horrible the margins are.

I’m sorry that you had to do that. I really am.

If this app is any indication of what Epub3 support is going to be like for mainstream reading apps then it is dead on arrival.

I’m running the Sony app on a smartphone with a dual-core CPU. In spite of the decent hardware and the fact that 15 months have passed since the Epub3 spec was finalized, most of the cool new features of Epub3 don’t work. What’s more, the ebooks look ugly.

If this is what the commercial Epub3 ebooks are going to look like then that format is dead on arrival. It completely fails to live up to one of the goals of a common format: it doesn’t work everywhere.

Music has standard formats because users know that any mp3 will work in any mp3 player. Video has standard formats and we know the file will work. And of course images have standard formats that work in any web browser.

If you compare Epub to the standard  image, video, and audio formats, you’ll agree that it does not behave nearly as consistently as jpeg, mp3, mp4, or what have you.

Don’t get me wrong, Apple’s implementation of Epub3 works well enough. But that’s just one vendor who also put a lot of work into developing their own version of the Epub3 ebook format, iBooks. So far no one else has been able to match Apple, and that renders support for Epub3 inconsistently non-standard.

At this point it looks to me like Epub3 is going to be about as standard as Epub, which is actually available in at least 7 different flavors. That’s not a standard format – not by any user’s definition.

When I last posted on the multitude of Epub ebook formats in February 2012, Felix left a comment that explains this point. His remarks are still true today and apply equally well for Epub ebooks:

Try this:
What is UNIX? Is it a product? A specification?
What is LINUX? A product? A specification?

Neither is either.

They are simply terms that refer to a variety of individual products that share some ideology and toolkits but are neither a coherent specification not a specific product.

People don’t buy or use LINUX; they buy/use Red Hat, Mandrake, Debian, Ubuntu, MacOX, Android, etc.

There is *no* LINUX as a product. There is *no* UNIX as a product.

At the product level, which is what we are discussing about ebooks, whether the plumbing is derived from the historical UNIX codebase or the various LINUX mutations is irrelevant to consumers.

Just as the internal plumbing of ebooks is irrelevant to ebook buyers.

Like Linux, "Epub" is more of a marketing label than anything. That has to be embarrassing for the engineers who originally developed it.

Similar Articles


Mike Cane February 23, 2013 um 10:02 am

So… how do Kindle A/V books work on your phone?

Nate Hoffelder February 23, 2013 um 10:11 am

From what I can tell the Kindle A/V features are dependent on certain parts of iOS. That’s why Amazon has not expanded support to other devices.

Mike Cane February 23, 2013 um 12:12 pm

My mistake. I thought Amazon had expanded K A/V to Android. This mistake based on Amazon refusing to let me "buy" a free eBook because I didn’t have a "compatible Android device"(!).

Alberto Pettarin February 23, 2013 um 10:18 am

Sorry, but I think that taking the EPUB 3 files from the IDPF test suite is NOT a good way for judging how EPUB 3 *commercial* eBooks will behave.

The test suite files are intended to "stress" (the support of) various parts of the EPUB 3 specification, they are not meant to be "consumed" by the public.

Plus, remember that you are testing a particular rendering app. Hence, you should criticize the combination of those book+that app, not EPUB 3 in general.

Nate Hoffelder February 23, 2013 um 11:22 am

First, if these files weren’t supposed to be used by the public then why were they released into the wild?

And I was using the files to stress-test the Sony Reader app. It failed under the pressure. If I’m not supposed to use these files to check for compatibility then what files should I use?

There has to be something I can use to test reading apps otherwise there’s no way to confirm Epub3 support. And if I cannot confirm support through evidence-based testing then Epub3 is not a technical spec so much as it is a marketing label.

And one thing you don’t seem to understand is that when I point out that one of the mainstream reading apps can’t support Epub3, it weakens the claim that Epub3 is an industry standard. The thing about standards is that they are supposed to work. You know that if you download a random mp3 file off of a website it will work in your mp3 player. It doesn’t matter what tools were used to make it or which mp3 player you are using. You will hear music.

The same cannot be said for Epub3, so it is not an industry standard ebook format (as of 23 February 2013).

Alberto Pettarin February 23, 2013 um 2:44 pm

Perhaps my previous comment was not clear.

My point was the following: each of the EPUB 3 test suite files focuses on some specific aspect of the EPUB 3 specification, and probably they have not been "polished" enough as if they were going to be released commercially, for any one to use.

Said that, I agree that the code inside some of them is atrocious and it defies the purpose of having them in that suite.

On my web page you can find some "commercial grade" EPUB 3 files, that you can use for testing purposes. (And please let me restate that my comment about the fact that you are testing the combination of file+app is still valid.)

fjtorres February 23, 2013 um 11:58 am

Whoa! If you can’t trust the reference files from the *creators* and caretakers of the spec, what kind of standard is epub3 going to be?
Oh, right…
Its. Not. A. Standard.
It is just a handwave specification that is neither certified by an international standards body nor policed by anybody.
Which is why Sony can implement whatever parts they *can* and still call it epub3 support…
Why Apple can impement whatever parts they want to and sell it as iBooks…
Why everybody can tweak and mutate, implement or deprecate, and literally fork the hell out of the spec to create their own proprietary format. And they can still call it epub.

Standards have licensing with teeth; standards have validation suites and reference files that work everywhere. Standards have known-good reference implementations of both content and "player apps". (In this case, reading apps.)

For epub 2.x Adobe filled the vacuum.
For epub 3.x, nobody is stepping up.

epub3 may or not be DOA but it definitely is a non-starter, so far.

Scott Tuttle February 23, 2013 um 10:38 am

Sony makes great hardware. Their software is what always irks me.

CJJ February 23, 2013 um 12:49 pm

How true.

Richard Pipe February 23, 2013 um 10:41 am

Interesting. The comments about the IDPF test books are correct. They are a load of rubbish (to be honest). In fact they are so bad it is difficult to explain how bad and whether they were uploaded to slow ePub3 implementation or as a number of independent social internet self-promos. With a very few exceptions they hinder reading system development.

ePub3 is not a specification. It is a schoolboy essay and a load of vested interest inputs with perhaps lots of discussion and little value. The classic committee output. When implement a reading system, in our case AZARDI, it was essential to distill the nonsense from the substance.

The Linux/Unix comparison is not an accurate analogy. There is Linux as a standardized, tested, hard-core core OS. We can deploy AZARDI on RedHat, Mandrake, Ubuntu, Centos without missing a heartbeat. MacOS and Android are deliberate commercial deviants exploiting enlightened licenses. They shouldn’t be in the same list.

This is like comparing Windows or IOS to the *.DOC or ODT packaging format. Epub is meant to be a simple interchange package. However the ePub3 package makes it so difficult to do really important things, commercial readers willnever standardize. That is not the western publishing industry today which Adobe is trying to control from the production level.

This is the way forward where books and the web can meet in a complimentary package and exploit their differences Anything can read and open this. The important thing is ePub must abandon backward compatibility to OEB and generational different reader expectations.

We are fortunate that in our developing country markets we can just do that. There are no OEB or ePub2 readers.

Nate Hoffelder February 23, 2013 um 11:39 am

Thanks for the details about Epub3, the demo files, and what not. I need to rethink this topic.

fjtorres February 23, 2013 um 12:03 pm

You know, I’m starting to wonder about Sony.
Yes, they are about as disorganized as a six year old, but compared to other players not named Amazon and Apple they are starting to look almost competent.
They at least know to release their readers in the fall.
They knew to get into the UK before Amazon got 100% of the market and they are actively fighting for mindshare with their 20p ebook promotion.

So, yes, they are way behind the industry leaders and the T2 is a step back from the T1, but at least they sell it on their own web site. 😉

Mike Cane February 23, 2013 um 12:14 pm

For how long? Do you really expect a T3 this year? They shouldn’t even bother.

fjtorres February 23, 2013 um 4:26 pm

If they were to quit out of hopelessness they would have done it already.
So yes, I expect a T3 in august. They’ll stay, if only to save face.
After all, they still make Walkman music players…

Nate Hoffelder February 23, 2013 um 5:14 pm

And they only recently decided to stop making Mini-discs.

Nate Hoffelder February 23, 2013 um 12:52 pm

I too am wondering about Sony. I’m beginning to get the impression that the ebook software folks are running stealth projects that aren’t being supervised by Sony corporate.

Thomas February 23, 2013 um 9:04 pm

Sony is a large enough company that one division often has no idea what others are doing. During the rootkit fiasco, one group made malware to prevent computers from ripping CDs while another sold computers with software to rip CDs included.

Bill McCoy February 23, 2013 um 1:05 pm

"HTML5" is a marketing label too, covering 100+ specifications many of which are still wiggly. But browsers are far more compatible these days than they were just a few years ago. EPUB 3, itself based on HTML5, is not perfect but will definitely help raise the industry to a higher degree of compatibility and alignment with the overall Web. IDPF is focusing this year on promoting adoption and pushing for increased compatibility. An EPUB 3 conformance test suite is being developed and you will see within the next few months objects scores of reading system conformance being published ( ). Meanwhile though the EPUB 3 sample files are intended to illustrate specific point features, not be representative examples of best-practices publications, so it’s no surprise that they don’t for example always adapt to a small screen properly… that’s not really a valid criticism of either the test files or the Sony implementation. For overall best practices, one good source is the new O’Reilly book: . And criticizing EPUB 3 today because some reading systems haven’t upgraded yet is like criticizing HTML5 in 2010 because IE8 didn’t support it yet. Yes it’s unfortunate that major advances like HTML5 and EPUB 3 take time to fully roll out but that doesn’t make them "DOA".

Nate Hoffelder February 23, 2013 um 1:22 pm

This is wrong:

And criticizing EPUB 3 today because some reading systems haven’t upgraded yet is like criticizing HTML5 in 2010 because IE8 didn’t support it yet. Yes it’s unfortunate that major advances like HTML5 and EPUB 3 take time to fully roll out but that doesn’t make them “DOA”.

HTML5 was ratified in November 2012, while Epub3 was ratified in October 2011. There’s no comparison. Also, the major browser developers have been working on HTML5 support for years. For example, Firefox might not have complete support but they’ve had the core features implemented for nearly a year now. Mozilla is improving on that support all the time; can the same be said for Epub3 support?

And BTW, verification tools for HTML5 have existed since at least 2011. I know because I’ve used them. The only comparable tools for Epub3 are the demo files which everyone agrees are shit. How the IDPF could finalize a spec with no way to check conformance is beyond me.

Bill McCoy February 23, 2013 um 2:00 pm

Nate, HTML5 has not yet been "ratified". It’s currently an Editors Draft which states "This is a draft document and may be updated, replaced or obsoleted by other documents at any time. It is inappropriate to cite this document as other than work in progress". W3C has a unique process in declaring standards "done" that leads to the opposite problem of EPUB: by the time a W3C standard is declared done it’s irrelevant to real-world conformance which has by then already either happened, or not. IDPF, along with some other standards groups, finalizes standards farther in advance of adoption. This has its pros and cons for sure but it’s not clear-cut that W3C way is better.

I completely agree that the situation with EPUB 3 have been finalized in late 2011 and several major reading systems not having upgraded yet, more than a year later, is not optimal. But I think it’s not too surprising given the magnitude of the upgrade. As far as verification tools, EPUBCheck was updated nearly simultaneously with the EPUB 3 specs, so content validation has been available all along. We are for sure a bit behind in creating reading system verification tests but IDPF depends on volunteer efforts. And if you think the EPUB 3 samples aren’t good then by all means help fix them or contribute more! Or, at least submit bugs in the issue tracker rather than just issuing vague complaints.

Richard Pipe February 23, 2013 um 2:18 pm

The ePub3 spec is so backward looking it is DOA. The test suite will do nothing to advance ePub3 reading systems because the specification is so loose the actual required behaviours of reading systems for essential features are not strictly defined. Curiously only the requirements for non-essential items are provided in detailed. The mechanical required vs. optional issues are moot points because most of them are arbitrary. The specification is flawed.

The issue of developing appropriate reading systems is not what is done, but how it is done. Because the 'spec' is a "what" document with no "how" context it is doomed. Yes ePub3 reading systems will limp into the market with the big EPUB3 banner, and do little more than an ePub2 except at higher production costs and few new features. The e-retailer reading systems AAAKG will just choose which sub-set of the specification they support and continue the incompatibility madness. Production and distribution costs go through the roof.

The real issue that has not been addressed in 2013 is the ability for publishers to create compelling digital content for consumers. Ultimately the current ePub3 specification deals with prettification of linear fiction and non-fiction, a theoretical set of features for academic content that will never work, and copies of Apple’s sad fixed layout nonsense. There are 100 companies out there turning InDesign files into fixed layout for goodness sake!

It took precisely one programmer (admittedly an amazingly talented programmer) with a red-hot testing team just two months to produce AZARDI 17 with more fixed layout features than anyone could imagine. The secret was over 80 defined and controlled test cases to work against. The test cases took six months to create and are ongoing. Some of them are online on our site as exemplars of what test cases should be like.

Everyone seems to make creating a Javascript ePub3 reader in 2o13 some big deal. It’s actually simple if you have a vision and direction… and real test cases. The IDPF provides words and absolutely no direction except for an montonous dialogue on manga and fiction.

My primary criticism of ePub3 is that it completely fails education, learning, and training content. the markets it could have addressed and made a real difference in. Our implementation of the ePub3 reflowable and fixed layout specs delivers asymmetrical panels, and reflowing panels, all within the specification as detailed on our various sites specifically to address the Unit/Lesson/Section nature of K-12 education with self-paced learning.

We think education. We think education in developing countries in many languages. I guess we have a different viewpoint to those who sit on the IDPF committees. To us it is about creating solutions were people connect with content and publishers making money delivering value.

What is reflected in every part of the ePub3 specification is total disregard for the education market. It’s a shock. EPub Zero addresses the requirements for the rest of the world where backward compatibility with Adobe, Sony, Apple something doesn’t matter. All the brains at the IDPF tables should have been able to work that out and how little backward compatibility matters. It is 2-3 month software development exercise to make a reading system that can handle ePub2 and a significantly better forward looking non-compatible package.

Every book should adapt to every screen fixed or reflowable. That is the USP of AZARDI and we will be putting up some demonstrations to show how rich and interactive education content works from 4in to 10in tablets up to 21″ hi-res work-stations. The industry just has to shake paper out of its brain.

My apologies for the rant. My buttons have been pushed! But things are moving and shaking right now in developing countries in ways no-one has seen.

Bill McCoy February 23, 2013 um 8:00 pm

Richard, you are right that education market is critical. And education needs features like accessibility that are built-in to EPUB 3, which has a superset of the functionality of the DAISY format which was specialized for accessibility: all the DAISY efforts are now focused on EPUB 3. You write that "Our implementation of the ePub3 reflowable and fixed layout specs delivers asymmetrical panels, and reflowing panels, all within the specification as detailed on our various sites specifically to address the Unit/Lesson/Section nature of K-12 education with self-paced learning". So what’s not to like? Fragmenting the market with some poorly thought-out alternative HTML5-based format, that doesn’t meet education market needs like accessiblity, would put things further behind.

You’re also right that the developing world education is moving fast. Publishers like Vibal (#1 in Philippines) are already delivering enhanced digital learning materials via EPUB 3. Vibal didn’t wait for any "AAAKG" companies, they just made it happen. I understand and share the frustration that things aren’t moving fast enough but just because a plant isn’t growing as fast as you’d like isn’t a good reason to rip it up and start over. And fragmenting into multiple incompatible formats certainly won’t help (anyone but Jeff Bezos that is, who’d be pleased as punch if the industry can’t get its act together and establish a unified open standard).

As far as benefits of retaining EPUB 2 compatibility in EPUB 3, that’s let O’Reilly already shift all their direct distribution to EPUB 3 files, which also work on EPUB 2 reading system. So it seems like a pretty reasonable tradeoff to me, given how much EPUB 2 adoption there is and that the resulting warts are relatively minor.

Ben February 23, 2013 um 9:30 pm

This is a good reason why Amazon dont care for the open format. epub format is pointless as long as it is DRM. This myth that epub 3 will increase ebook sales and save publishers against Amazon is really funny at this point. It’s funnier to watch publishers pay thousands to attend conferences where people are still telling them epub and and html is the future…one big problem, most publishers out source their ebooks, no one care enough to invest to build an internal ebook division that can actually understand and create epubs. As long as we continues to have this separation of content and skills epub 3 will keep dragging it’s feet…2013 will be another year with more epub3 rumors…maybe 2014 will be different but dont hold your breath

Nate Hoffelder February 23, 2013 um 9:43 pm

Epub3 does present a good argument for going it alone, yes. Amazon announced KF8 in October 2011 and officially released it 3 months later. How long have we been waiting for Epub3, exactly?

Bill McCoy February 24, 2013 um 11:10 am

Nate, as you know Amazon released KF8 on only a limited portion of their installed base. And KF8 is a very limited "spec" (I put that in quotes because there is no clear specification documents, just a list of tags) with features more equivalent to an EPUB 2.5. Comparing the time one vendor took to do one proprietary thing in some of their products to the time it’s taking dozens of vendors and retailers to upgrade an entire ecosystem already using a pretty successful standard (EPUB 2) isn’t entirely fair. And we have relatively complete EPUB 3 already shipping in Apple iBooks, Kobo, and VitalSource Bookshelf plus partial support in Google Play Books and now Sony Reader. And folks like O’Reilly already delivering their content in EPUB 3. Yes we are still waiting for some others and it’s unfortunate that it’s taking as long as it is (one reason no one is voicing out loud for this is that a certain SDK vendor reneged on their publicly communited roadmap for delivering EPUB 3 support, so they could push a proprietary digital magazine solution). But the alternative to pressing on would be much much worse (fragmentation, less interoperability for publishers and consumers, more market share for Amazon and other proprietary vendors, less accessibility, …).

Nate Hoffelder February 24, 2013 um 1:07 pm

Kobo does not support Epub3; so far the only features they have added are the ones needed to support Japanese language text. Amazon, on the other hand, has rolled out KF8 support to everything but the long-dead K2 and the iPad app (even that app has partial support).

And so far as I know O’Reilly is not actually adding anything to the Epub3 files for many of their current titles that wasn’t already there in earlier Epub files. They are arguably offering Epub3 more for marketing than technical reasons – more so that they can say they’re offering it than because they actually need the abilities in the format. BTW, they first adopted KF8 for much the same reason.

But I do agree that KF8 is closer to being Epub 2.5. It’s been suggested that Amazon copied the Epub3 spec, deleted the privacy-violating and CPU-intensive sections, and then developed KF8 based on what was left. That was arguably a smart move considering that KF8 is deployed and functional.

Delphi March 3, 2013 um 5:02 pm

…one big problem, most publishers out source their ebooks, no one care enough to invest to build an internal ebook division that can actually understand and create epubs. As long as we continues to have this separation of content and skills epub 3 will keep dragging it’s feet…

I think this is an interesting and potentially very important point. Are conversion houses likely to invest time and money into exploiting the full power and potential of epub, if they can satisfy 80-90% of their customers with a valid but bare-bones output?

Write a Comment