The exclusive focus on the functional capabilities of software doesn’t just overlook usability, it also overlooks affordances.
You can imagine Tim Berners-Lee circa 1990 telling a hypertext-illiterate friend about this World Wide Web thing he'd just invented
"Have you seen my new invention?"
“what is it?”
“There's this special language, called HTML, which you can write documents in. It allows you to put these things called 'links' into your documents, so that when you view them, you can click on bits of the text -- the links -- to load up another document stored on the network.
Oh, and to view these documents, you need to use a separate program called a browser.”
“why would I want to use that? With <my favourite document editor> I can read and write documents, and with the programs I already use it’s easy to load another document off the network.”
The fact is that there's not a single functional capability in the WWW that can't be achieved in other ways; in fact, the web has (and still does, to a certain extent) less functionality than is available by other means.
What it does have is different Affordances.
I was introduced to the notion of affordances by Donal Norman’s book The Design of Everyday Things. It’s about the design of the everyday objects you interact with, like door handles, alarm clocks and car steering wheels. Why are some easy to use and intuitive, and others awkward and frustrating? Norman looks at how designs do and do not match with our tendancies, and our capabilities for perceiving and interacting with the world.
I’d highly recommend it to anyone, even if the topic isn’t something you’re specifically interested in. It’s a real eye-opener, and it's subject matter is relevant to everybody’s lives.
A hinged door that just presents a flat metal panel at hand height suggests that it opens away from you by pushing it. While a door with a metal bar in the same position that you can grasp suggests that you grab hold of it and pull it. These designs exhibit affordances – they are suggestive of function.
In general, I’d say you have an afforadance when properties of the object or system shape and influence the ways that you use them. This is a bit broader than the original meaning, but I think the term still fits.
Lets go back to the World Wide Web. Links are easy to embed into documents -- you can embed them seamlessly into the text of the document, rather than having to figure out where to put them in relation to the flow of the text -- and it's easy to use a link. You just point at it, click the mouse and then the linked page appears.
Without links, you'd have to find another way to describe the location of the other document, and then the user would have to either cut and paste that location or remember it, and then go to whatever program they would use to access it. And they won't have any forwards or back buttons.
In principle, there isn't a huge difference between these two alternatives. But in practice there is. We’re all very sensitive to how much effort it takes to do things. It doesn't take much for the effort to exceed the threshold where we just won't bother. Trivial amounts of extra effort can break the camel's back. Yes, “in principle” we could make the extra effort if we really wanted to. But the amount of effort involved means that in practice we don’t want to.
So it encourages linkages and jumping between documents. Blogs wouldn't make sense unless this was easy to do.
Or take SMSs or twitter messages (not that I’ve used twitter myself, I've just read about it). You could in principle write exactly the same way you would in emails or blog posts, and just split the text up into multiple SMSs or tweets, but you don’t. SMSs and tweets have differents sorts of affordances to email/blog posts, so you write in different ways.
That’s the thing about affordances. You can have multiple different systems that, in principle, you could do exactly the same things with, but you don’t, because they have different affordances.