And when you get to incompatible implementations of an allegedly identical format such as ePub, what do you do then? How do you decide whether you should be converting ePub automatically to ePub or the other way round?

Historically, on-the-fly conversion has often seemed an attractive temporary solution but it has generally been superseded by one of the contending formats emerging as a clear winner.

But yes, it’s always worth remembering that if you ask consumers to go away and check what formats their device is capable of reading, they will indeed obey you, and go away.

That puts it in the company’s hands to get the program written, and make it work. If he succeeds, the company makes money. If they fail, the device tanks, and everyone’s on the unemployment line. That’s how consumer-oriented business works.

Is creating a conversion engine to read a file of text and images that tough? Not really. Much tougher conversions have been tackled in the past, and this can be done, too. (Sony has problems because they created an overly-complicated format.) There are devices out there right now that read 6 or more formats, which pretty much says it all.

Sony struggle to display their own format reliably. Every other device I’ve seen has had similar issues (I’ve seen iPhone, iLiad, Kindle, bEBook, Cybook and a few others). If you add all those display bugs together you’ve got rather a lot of issues. Enough to make it unlikely that any one device will deal “well enough” with more than a few formats.

Conversions are always lossy. Every format has features missing from others, some of which can be emulated but some can’t. Dump any book to HTML and see what you get, for example, or in the extreme, text. Until we have a metadata standard no automatic conversion will cover all cases, and it will have to be a format in its own right to cover everything – like OGG, for example). So a “native” version will almost always be superior.

Many formats assume computing resources that are unreasonable for a portable device. Rendering an MS-Word docx file on my laptop redlines the cpu for a few seconds and eats ~10x the document size in RAM, plus the overhead of the docx viewer. My laptop can play MP3 files at ~30x real time and CPU is the limit there. My Sony PRS-505 can play mp3 files at 1x speed, but battery life drops to less than 10 hours. So assuming CPU limiting in both cases and assuming enough RAM (my Sony is ~100x short), the Sony would take a couple of minutes to open a similar DOCX file and doing so would flatten the battery in a few hours. Pre-converting on the laptop would make a lot more sense, if you were willing to allow that. The Sony “library” software does this, for example.

Now, assuming the Moore’s Law pixies continue to work their magic it’s quite likely that within a decade that last issue will be much less important – I’ll have enough hardware power to run whatever software I need for weeks between charges. But unless we make more progress in software engineering in the next ten years than we have in the last fifty, we’ll still have buggy rendering engines making a mess of eccentric ebook formats.

We will also need dramatic reform of a few laws and a big change in attitude from major players. How do you propose to let “my gran” read a book she bought for her Kindle on her new Sony PRS? Does she just pay another rental fee for the new format, backed by anti-monopoly law banning exclusive deals? And what about licenses for all the various formats that require fees for rendering/decoding engines?

Steve’s point is that with a babel of file formats and renderers, most manufacturers and publishers would greatly benefit from massive cross licensing, rather than attempts to lock in customers. Without that, sales of ebooks and readers will remain a tiny percentage of the book market.

Amazon, Sony, etc. can have the biggest slice of a tiny pie, or a modest slice of a giant one. They need to think about this carefully.

Regards,
Jack Tingle

The reason this isn’t a widespread approach today is DRM. If ebooks were DRM free, the ebook format would be much less of an issue because conversion is largely a “solved” problem. Only PDF to reflow conversion is currently hard, and I expect good progress even on this front.

I’m talking about a reading device that already has multiple conversion tools on-board, and does the conversion itself, automatically, without consumer effort. A seamless transition from whatever file you have, in up to a dozen or more formats, to words on your screen. That’s what consumers want, and they will respond to the first device that gives them that with overwhelming support (unless, of course, it is hideous and expensive… in which case, they’ll respond to the first device that converts all files, is pretty and/or inexpensive).

I’ve long since stopped worrying about whether any device I might buy can read the formats I buy in. Instead I only buy formats that I can convert to something open and re-convertible. Mostly because the device I have now is not the device I will have in ten years time. So buying a stupid format or renting DRM’d files might be ok today but is just going to bite me later on.