Structure as a key to … everything?
About 13 years ago, when I worked on the online news product of The Washington Post, we struggled with two time-consuming challenges: classified ads and story categories. They were not the only challenges, of course, but they proved particularly difficult to manage.
The classifieds already had a highly structured system of categories (houses for sale, houses for rent, apartments, etc.), but there was absolutely no structure within the categories. If you wanted a house with a fireplace, you couldn’t search for that (except by using your eyeballs on the text). Why? Because fireplaces were represented by firepl, fpl, fp, frplc … you get the idea. I spent a lot of hours with the czar of classified technology, but the more he taught me about the legacy system, the more impossible it seemed to translate it properly to digital media.
I remember at least one lunch conversation (maybe several) where I explained the classified system to my online colleagues. There was amazement all around at the idea that the people who “took” the classified ads over the phone could just type anything they wanted (firepl, fpl, fp, frplc …) instead of choosing a single consistent term from a drop-down menu. How were we going to deal with that unstructured mush — in our digital world? How would people find the house with a fireplace?
The story categories were a bit different. Each story came with a section designation, such as metro, national, foreign, travel, food, sports. That is, each story bore a label (today we call them tags) — one label. We online folks quickly realized that we needed a story to have at least one more label, because some national stories are about health, or elections, or law, and we wanted to be able to serve up content in a variety of useful bundles.
We devised a system, whittled it down to about 70 content labels, and then tried to persuade the copy editors to choose and apply a label to each story — just one, because that was all the newsroom editing system could accommodate. It was hit-or-miss; not everyone remembered to type the label into the box, and some people just downright resisted using our list of labels.
So it was déjà vu all over again when I read The Structured Web – A Primer — an excellent discussion of why structure is so important in the kind of information environment we all live in today.
It also reminds me of the talk Adrian Holovaty walks about data. Every journalism story has people, places, dates, and other pieces of data that could serve as tags — if only the journalism field would realize that finding stories is no longer something that people do with their eyeballs alone. Tagged stories can be deconstructed and reconfigured in ways that make the product of a news organization more long-lived, more valuable, and (best of all) more usable by people who want it.
All those years ago at the Post, I worked with a small group of wonderful people who understood these things. One of our dreams was that if you were looking for information about, say, a country where you’d like to go for vacation, you could get a menu of recent and older stories that were actually about that country — not merely stories that happened to contain the name of the country. We struggled with the traditional separation between the restaurant reviews and the food pages. Foodies love recipes and restaurants, so why should these things be divided? If you searched for Asian noodles, why couldn’t you find both where to eat them and how to make them?
It was a long way off (the 1994 Netscape beta had not yet been unleashed), but the searching, sorting and tagging we take for granted today were things we already understood, because we wanted them.
We wanted them back in 1994, and in some respects, we finally have those things today. But the business I love, journalism, has been very slow to recognize the importance of adapting to these new ways. Very, very slow, when you consider that it was perfectly clear 13 years ago that the existing classified system had to be scrapped and replaced with structured data. Clear to some people, but not to others.
Thanks to Shane Richmond for the link!
Hey Mindy — a very interesting post, especially since i am now discovering geotagging … which seems to be well developed on your side of our mutual ocean but not here. how widespread is geotagging among online journalists, in your experiences?
November 10, 2007 at 5:42 pmMindy, I’ve mentioned this several times on my blog, and I usually get a few responses from industry veterans about how my suggestions are impossible — that the legacy systems just can’t support my ideas of making everything into a database item, tagging everything and making it infinity searchable.
And I ask why? Why is this so? And it’s quite simple really: Newspaper Web sites are still print first, Web second. When newspapers finally become Web first, they’ll understand why tagging stories, photos, videos, etc is so important.
But for now, we’ll still be stuck with excuses. How it’s just not “possible.” How is such a simple concept so daunting?
I think the answer is equally simple. It’s going to take an epiphany by top editors and publishers that this is important. Newspapers could probably make a lot more money off of the same content if it were just easier to find. People don’t thumb through Web sites. Not sure why editors think that people would.
A great example of a horribly hard to search Web site is my papers, and it kills our page views. We, and most newspapers, need more ways for people to find stories than we currently allow.
If people at my paper don’t get it with me working there and harping on all these things, I can only imagine other papers without people saying that they should try new things.
November 10, 2007 at 10:54 pmSilo-based content is a hard nut to crack because it was the only distribution method many of us ever knew. The Web’s biggest challenge — and its biggest asset — is its the ability to separate content from its container and reorganize the material on the fly. That’s a *huge* shift away from traditional story-slot concepts in print and broadcast. Prior to the Web, the thought of publishing one story in multiple areas was anathema because it wasted page space or broadcast time (and it sucked up valuable resources). But now, a simple tagging system can cause one story to show up in a variety of categories, and those categories can be cross referenced and manipulated in hundreds of ways.
Those of us who deal with Web content on a day to day basis understand the Web’s fluid nature, but I think it’s going to be a while before more folks can understand taxonomies, folksonomies and other data structures. Of course, if a distinct return on investment can be connected to data structure, then the discovery process could accelerate dramatically.
November 14, 2007 at 11:39 amHi Mindy,
This post resonates with me, as I spent some time between grad school and teaching practicing Information Architecture for marketing and government clients.
These fields of communication were realizing the real power of interactivity – that websites are more than just pamphlets on the screen.
It does seem that journalism on the whole has been fairly slow to actualize this same insight – there are however bright spots.
This is all apropos to the class we’ve been discussing that I am teaching in the Spring. We want to try to cover three ideas, all through practical application:
1. Online journalism in context of media convergence, and the practice of information architecture
2. Reporting and producing multimedia news stories using Flash
3. Distribution and syndication of content in a real media marketplace
#1 speaks to the point of your post exactly. #2 is based on your book. #3 is going to be very very interesting, and worth further discussion.
Thanks for the great resources, including this blog. I can see some really rich dialogue as we go down this road.
Best,
Jason
November 20, 2007 at 6:58 am