Sunday, July 5, 2009

Automated summaries and excerpts

You've seen them around the web: they're the blocks of text appearing under the headlines, giving you a little more information on what the linked article is about. If the headline didn't tell you enough, the summary or excerpt is supposed to serve as a sort of fall-back mechanism to tell you a little bit more.

The websites of such longtime print titans such as The New York Times, The Economist, and The Wall Street Journal tend to have good summaries beneath their headlines. They know how summaries should be written — by hand. They are, after all, the professional producers of such content and have a vested interest in getting that content viewed.

Those who are a few steps removed from the production of such content, on the other hand, make it readily apparent that they just don't care. Two of the most surprising offenders are Google and Apple — ironically, two media darlings who appear regularly in feature articles.

A quick jaunt over to Google Finance gives us an example of their automated summary text. If news articles could get circumcised, then Google Finance summaries would be the equivalent of the unwanted foreskins.

By Brian Love PARIS, July 5 (Reuters) - World leaders are bound to express the hope that the worst of the global economic crisis is passing when they meet this week, but they are under pressure, too, to manage a Chinese challenge to<br />decades of dollar ...

Where do we even begin when trying to enumerate what's wrong with this? First of all, the summary text is much too long to reasonably hold the attention of the typical reader who is scanning the page; it's unreasonable to even call it a summary.

And even if you read the entire thing, it doesn't make sense because it trails off. Decades of dollar what?

And then there's the extraneous information: the name of the author, the location where the article was wired from, and the name of the wire service. I'm sure the author is a swell guy, that Paris is a lovely city, and that Reuters is excellent at what it does, but what is all that doing in the summary text?

Oh, but Google is all about automating stuff, even if it turns out a little ugly. Let's look at what Apple does; we know they've got a better sense of design than anyone around.

Sorry to be the one to tell you this, but the headlines on the Apple start page don't fare much better. Here's just one entry from the entire embarrassing list.

If you'd like to try something both delicious and healthy this holiday weekend, this about serving your guests Barbecue Glazed Alaska Salmon with...

Well, if anything could be called an improvement, I suppose we can give this one a little bit of credit. The summary is at least short enough to read and doesn't contain extra cruft. But it still doesn't make sense when considered as a self-contained sentence. Barbecue Glazed Alaska Salmon with what? Why should I click “Learn more” when you can't even bother taking the time to edit the summary to give me a coherent sentence up front? Is this entire article made up of sentences that are going to trail off and leave me hanging just like the summary did?

Why, in the middle of 2009, do we still see these amateur-looking automated excerpts on websites run by widely respected technology companies?

I'm not the one making decisions inside the conference rooms in these companies, so I can't say for sure. But I can venture a guess. These two companies are highly respected, to a large extent, because they're also highly profitable. Assuming that people don't really care whether the summaries are readable or not, it certainly makes sense to cut costs and forgo hand-editing of summaries. And being technology companies, of course they'd want an automated, technological solution; when all you have is a hammer, everything looks like a nail.

So these world-class companies knowingly produce these barely passable, poor excuses for excerpts. What does that mean? What's the implication? From this, we can conclude that these excerpts aren't meant to be read. If they aren't meant to be read, then they aren't meant to be taken seriously. They're merely filler text, a cheap imitation of what big-boy news organizations have done for years. Maybe in the age of continuous partial attention, this makes sense for a new kind of audience.

As for me, I'm going to remain stubborn and demand only the best from my news sources — including summaries and excerpts. Is it really too much to ask for a complete sentence these days?