Over the weekend, I read Anne Gentle's book Conversation and Community: the Social Web for Documentation. It's quite good, although since I'm writing a blog, and you're reading a blog, we may not be the demographic to extract the most value from it. But that's not important right now.

In it, she talks about "content curation". After asking around, this turns out to be this month's Tech Comms buzzword, and behind it sits the idea that just maybe, publishing shouldn't be a fire and forget affair. Dandy. What surprised me was just how familiar this seemed - both the turn of phrase, and the sentiment.

The turn of phrase, of course, reminds me of Warren Ellis, and his short blog post: The patchwork years. He may or may not have coined the term, but this idea of the internet as a system of informed curation and creation is one he's been knocking around for a while. For anybody who doesn't know, Warren Ellis is a writer, web pundit, and exemplary cranky old ***. Information saturation, and the difficulty of sifting and working with knowledge are recurrent themes in a lot of his work (see particularly Transmetropolitan and Lazarus Churchyard), as well as being pretty well worn to anyone who reads sci fi, or just happened to drift into a screening of Minority Report. It goes to the whole Kurzweil shtick of technology and information galloping away.

Funnily enough, this is also what sat under the more interesting remarks in Peter Anghelides' keynote address at TCUK09. He talked about IBM's Smarter Planet idea - managing ("leveraging" if you're that way inclined) the output of an increasingly instrumented and networked world. In English: as more of the things around us (and, crucially, it doesn't really matter what they are) can not only produce but consume and manipulate data, the world could become exponentially more confusing if this data isn't tidied up into useful information. Your house could tweet when your laundry's dry, but how do you integrate that into better running your life? That kind of thing.

Being the keynote at a technical communications conference, Peter was kind enough to suggest that technical communicators might be best placed to address these questions. The paper manuals crowd almost looked placated. Indeed, as stereotypes of a profession with (at best) a thing for taxonomy, they might  even relish the fact that IBM now refers to us all as "Information Developers".

I sound horribly dismissive there, and initially I was, but now I'm not so sure.

At the Cambridge leg of the Stack Overflow dev days, Christian Heilmann introduced YUI and YQL - bundles of APIs, passably usable, for mashing up not only data, but already-manipulated data. The idea of progressively complicated layers of APIs interacting to make the web smarter is not an astoundingly new idea, and it sure looked cool. But as I got to wondering what YQL might be for, I came back to Peter's speech, and the idea of content needing to be purposed, and. um.

Well, yeah, maybe the need for curation (coupled with informed creation) is getting more pressing, and maybe Information Developers (which sounds a little less pretentious each time I think about it) are well placed to engage with that.

Actually, when you get down to it, we're doing a bit already, at Red Gate and elsewhere. Not perfectly, but we're making a start. To pick a Red Gate example, the Future of Monitoring site is a community conversation around our users data and information needs, driven by technical communications an usability. Equally, across our products, we spend a fair bit of time collating data to try and make sure we have a thorough initial (creation) and progressive (curation) understanding of what our content is for.

Ok, so this may not always be spectacularly granular. For V1 of a brand new product, it's likely to be speculative. Oh, it'll be based on user research, talking to folks, feedback from usability trials, and informed by close involvement with the development process, basic stuff like that; but it can still go live with fairly limited validation. We know, in fact, what we think it's for, and even if it's the seventh re-draft of a help topic for version two hundred and thirty of SQL Compare, we need to make sure it's still up to scratch.

So every few months, we take a product, usually one that hasn't had a bit of love in a while, and check that its user assistance is working. It's a similar piece of thinking to those we do in content creation, but this curation step has the added benefit of accumulated data. There are web analytics, logged support calls, survey responses, subsequent user research from UX, or product management, and the direct, front-line impressions of our sales force. It's all big, juicy, enticing data, and quite a lot of it is also information. Simple example: if users are still asking questions I think I've answered, then either my content or my information architecture is broken. If users are asking questions I hadn't anticipated, I need to offer them answers. If the answers they need are fiddly to explain, thorny, and tricky to formulate, we may also have a product design or usability issue.

Validating user assistance makes the help better, sure. It lets you remove redundant content, optimise what you have, and lets you learn about how to do it better next time. It also feeds back into the development cycle. In short, it's spiffy.

It's also the beginnings of Content Strategy.

There are approximately twenty squillion blogs out there defining content strategy. For a good one, see Kristina Halvorson's article at A List Apart: The Discipline of Content Strategy. (She's also written a pretty decent book) Cliff notes: it's making sure your web content (and I mean all of it) constitutes a meaningful, consistent experience, and fits the needs of your users as well as your business goals. You can throw in the word "holistic" too, if you went with "leverage" back there. It includes making sure your content continues to behave like that. And if it's still true and useful, and if it's still spiffy, even just in our help, might this kind of structured knowledge stewardship have benefit elsewhere, as the web gets smarter?

I guess we've got a handle on some of where to start. And we know about some tools to be going on with (contnet inventories, usability work, analytics, etc). But to ponce it up for a bit of a soundbite: information structures with changing substrates demand curation. This feels like it's only going to get harder as the smarter planet, or the "emergent web" or what you will, means that more information is changing information, and more objects are information objects.

The concept of information curation has been borrowed largely from the arts, from museums and galleries. As anybody who's been to a particularly good (or particularly bad) exhibition will recognise, the curator's act of presenting existing and new resources in arrangements that elicit a complex, cohesive response is incredibly powerful. My thoughts on this aren't very formed yet, but I'm sure we're gonna need some of that.

(I also confidently predict that somebody with a robust business case for doing so will assert that DITA is the answer within four or five comments)