When books ain't books 
An encyclopedia contains many lessons, but the form of the encyclopedia itself is a useful lesson. Its modern shape was drawn in happy innocence at the dawn of the Enlightenment; after two centuries of the primacy of science and inquiry in the West, it had swelled to consume entire bookshelves. It became ponderous and uncompelling, and actually quite hard to use. So, in the early nineties, encyclopedias were among the vanguard of that great gold-rush: multimedia, and specifically, CD-ROMs. Of all the publishing ventures of the CD-ROM age, multimedia encyclopedias (namely Britannica and Encarta) might be the one example of something that approached critical mass.
In the process, their vaunted pages lost something of their power: the very heft of human knowledge you felt pulling a volume from the shelf was not easily translated to dropping a disk into a tray, double-clicking an icon, waiting, and pecking out a search term.
That lost sensation was mourned, but it was not such a bad thing. People came to see an encyclopedia for what it was — not necessarily an intimidating (and quickly dated) academy in printed form, but actually just a database of information about stuff. You could copy and paste a paragraph out of Encarta and into your essay — like a lot of not-so-smart students did in the CD-ROM era and still do now. Of course, smarter kids copied, pasted, then judiciously modified the text in places to cover their tracks. In losing its printed form (and no-one, surely, buys volumes of encyclopedias anymore), it gained a truer form: it left the ivory tower to mingle in the bazaar. In a hop, skip and jump, we had Wikipedia.
How heavy is Wikipedia? How many metres wide is it? Who wrote and edited it? We don't ask these questions. We do ask how reliable it is — but as a kid in the eighties, my family had Britannicas from 1964. According to that arbiter of truth, the moon was a distant and untrammelled body, a virgin of the night sky. This is why Wikipedia is the most reliable encyclopedia ever written.
The point is, the encyclopedia is an idea that was necessarily limited by its form. It had to iterate through a number of forms to arrive at its present (and perhaps not final, but still infinitely better) form: a web application.
Don't think that makes it a special case. Though it's a reference, it is a particular kind of reference — one that suits the model of sitting at a desk and making epistemological leaps of faith. Another kind of reference, the cookbook, has hitherto been poorly suited to a web incarnation, because nobody wants to coat their keyboard and mouse in flour and butter. A book, a book that holds its page when placed on a bench (so a weighty tome and a particular stock), is a much better form. But actually, the clip folder and plastic sheets are a better form again, because the collateral damage of an ordinary kitchen requires some preventative measures. (The internet serves that form very well — with the aid of a printer, of course.)
But technology promises a still better form, because recipes are what a geek might call well-normalised data. Imagine that I have a few items in the fridge that I want to cook up. With some legwork in the indices of various cookbooks, I can locate a recipe for those ingredients. If a cookbook were a database, I could enter those ingredients and quickly see a list of all possible recipes. Or I could look at the date, discover what fresh food is cheap in mid-autumn, and compose a shopping list for one or two specific recipes. Or I could do many other things besides — such as tap to double the quantities, or whatever. The work to do is to bring the technology into the kitchen, and kitchen-proof it, but that's not hard.
If you were developing an ereader, and you had digitised a few novels and made the experience good, you might come to a cookbook and wonder how best to display "that kind of book" in your system. You would probably come up with all these possibilities and more. And your answers might be clever, but the question is wrong. The cookbook is moderately well suited to bound paper, but poorly suited to a Kindle. You aren't inventing the future, you're just inventing a stopgap until someone comes along and does it properly — by giving the collection of recipes and folk wisdom a truer form. Your stopgap is like Encarta, if you're lucky.
I am travelling through San Francisco for a few days, and I have the Lonely Planet application on my phone. The app pays obeisance (too much obeisance) to the bookish origins of the data, but it does at least realise that it isn't a book at all. It's an application. Though I curse the poor implementation of it in the Lonely Planet app, a travel guide is wonderfully well-suited to contextual, database-driven assistance in the form of an app. At any moment I can pull out my phone, and consult the map without looking like a tourist, finding at a tap my precise location, and the directions to the nearest decent bar. If I had the bound paper version, I would get by, but thirstily.
Travel guide publishers, in fits and starts, are getting it. Their asset is not their books, but their commissioned knowledge. And their knowledge, freed from the contraints of a linear page order and clunky, painstaking indexing, is much richer and more valuable in another form.
It's not the same adventure as for encyclopedias, whose patrons are mostly sedentary and browsing on a large screen, but it's the same conclusion. Linearity is a liability. Heft is unhelpful. Books sometimes, not always, but discernably, are not books. For reference publishers, the future shift is much more dramatic than it is for common fiction, opinion and narrative writing. Don't pander to them, don't try to quell their anxiety by pretending their books are still books, you're doing them no service and wasting your time. Concentrate on the things for which the last page must come after the first — that problem is plenty hard enough.
Sorry, comments are not available on this post.