The same thing is happening with entertainment. Seeing a video used to require going to a movie theater. The effort was only worth it for a movie that lasted 1.5 hours or more. Maybe it was preceded by a newsreel or a cartoon to provide even more value. Then TV came, and most shows went down to half an hour. Now even that is often too long – we can use YouTube to get five minutes of something interesting.
This is not a random change or a pendulum swing. With better technology, we have better ability to search for content. This means that instead of getting a package which will hopefully include what we want, we can get exactly the thing we want. And with so much content available at our fingertips, we don't have the time to deal with fluff.
The unit size is affecting the financing of content creation. Any time you are asked to pay for something, you have to evaluate if it is worth paying for. This mental cost is going to happen, even if the financial cost is very low. The mental cost also has a lower bound. It's worth thinking if a book is worth $30 or not. But it is not worth thinking whether a Web page would be worth 5 cents. The total cost of purchase is going to be too high for the value, regardless of how low you make the financial cost.
So how do we get content creation financed (or, from another perspective, how do we get people to pay us for writing)? I can think of several models.
0. The volunteer model. Authors working for fun and glory, without an expectation of profit. If this sounds ridiculous, remember that Plato's dialogues were an unpaid labor of love. So is Wikipedia. But this model leaves something to be desired, such as food on the table of those of authors who aren't Athenian aristocrats.
I think it's obvious why this model gets numbered zero. Great things can be done with it, such as Wikipedia, but it tends to get biased results. There is only so much work you can do for free.
1. The patronage model. If you're the Emperor Augustus and you want people to believe Rome has a divine mission to rule an empire, you pay Virgil to write the Aeneid. But this isn't limited to propaganda pieces. If you're IBM and you need people to find out how to do things with Tivoli Identity Manager, you might pay Ori to write short demonstrations to put online so your customers won't have to call support. Hopefully, the cost of getting these demonstrations written is lower than the support costs IBM saves.
It's just a hunch, but I suspect we'll see a lot more of this model. The fact that information can be copied so easily does not hurt it. It merely means more viewers, which is what the patron likes.
2. Storyteller's bowl. It's very difficult to evaluate the value of content in advance. It's a lot easier to do it after reading. If enough people feel obligated to donate after writing, you can let them decide if the material is worth supporting (and how much). This is how Dave Freer's Save the Dragons works. It's also the business model for Torah.Org, and very similar to the model for Kingdom of Loathing.
3. Business as usual. If you have really good content, you might be able to charge for it. You're competing with a bunch of free material out there, but if you have great material, such as Baen's Webscriptions, you can still get people to pay you. But as units of content get smaller, this gets harder.
4. Aggregation and subscription. Business almost as usual. If you can aggregate enough units of content, then access to the whole collection might be worth enough to decide to pay for it. This is the way Books24x7 works, for example.
I'm not very optimistic about this model, because I don't think it produces enough per unit of content. I once wrote a book about RACF. It's on books24x7, and we (the publisher and all four authors) get about $200 per quarter. At $800 a year for a full book, I don't see enough value coming out of aggregators.