Give online news stories a relative importance rank

Written by Adrian Holovaty on January 5, 2004

While reading a crime story on the Seattle Times' Web site earlier today, I wondered how important the Times' editors thought the story was. Not knowing the Seattle area, I wasn't sure how newsworthy parking-lot shootings are in Des Moines, Wash., so I wanted to find out.

Problem was, I'd stumbled upon the story via a section index page that appeared to be ordering stories by date and time alone. That didn't give me any clues as to how important the story was, relative to the rest of the day's news. And the story wasn't on the site's home page; either it had already been replaced by more recent content, or it had never made it there in the first place. I didn't know, and I couldn't find out.

That got me thinking. It'd be useful if news sites made stories' importance more obvious.

A well-refined news-judgment scale is one of the few things print newspapers still have going for them. When you pick up a newspaper, it's easy to figure out what was important when that dead wood went to press: Just flip the thing to page one. And, for the most part, the rest of the stories fall in order of priority. (With exceptions, of course, for pages deep within a section, where story placement becomes more dictated by ad space, story length, photo availability, etc.)

On a news Web site, though, there aren't many good ways for editors to communicate how important a story is. Stories sit alone, in templated obscurity, with no hint of how much more or less newsworthy they are than every other story that day, that week, or that month.

But it doesn't have to be that way. Why not assign metadata to each story that explains how important the story is? Such a system needn't be complex; producers could assign a simple, one through 10, "importance value" to each story. With that data, a site could dynamically alter a story's appearance -- using a different sized headline, for instance -- based on its importance. Or maybe it'd be more straightforward to include a small sidebar saying: "This was the third most important story on our Web site on Thursday. Click here to see stories that were more important."

This extra metadata lends itself to a few other cool possibilities: Dynamically generated "Most important stories of 2004 (so far)" pages, search results sorted by story importance, section pages ordered by story importance, breaking-news e-mails sent out automatically whenever a story has the highest importance...And the most-important story list could be cross-referenced with the most-e-mailed story list to create some sort of greatest-hits collection of editors' picks and readers' picks.

I'm certain existing content-management systems have "priority" functionality; this idea is nothing special. But why aren't news sites using it?

Related: Nathan Ashby-Kuhlman's Can your CMS do 150-point headlines?


Posted by Sara on January 5, 2004, at 9:36 a.m.:

Importance is a subjective thing. Rating stories on a scale of 1 to 10 to determine importance is a tricky prospect. Some stories are more important on some days than others - and when that's extrapolated to a Web archive comprising years of articles, the importance assigned on a particular day isn't necessarily still accurate.

As a copy editor, I think the story itself should be able to explain how important an event is. Headline and photo size (and maybe a rating system) help, but that's gravy to the content itself. If the story doesn't give enough context for you to judge its importance, the reporters and editors haven't done their job.

Posted by AgentKen on January 5, 2004, at 10:46 p.m.:

On our family of sites, stories are prioritized for presentation on the page, but that priority ranking is not shown to the viewer in any other way (it is, however, hidden in the source code). Should these rankings be shown to readers in some fashion?

Think of the opposite way of doing this: Yahoo! News lets readers rank stories and then posts "top stories" links based on reader preference.

Which approach offeres better ROI for the site?

Posted by Julie on January 6, 2004, at 4:09 a.m.:

I have to chime in and agree with Sara. And I'd add that determining that 1-10 (or similar) ranking is not that simple a task unless ratings are applied without much deliberation (in which case they become less meaningful). Think about how much disagreement over story importance (especially as relative to other stories) takes place in daily editorial meetings. Much of "importance" is relative to the reader's personal interests and viewpoints. I think assigning a quantitative value to something as non-quantitative as news value, while well-intentioned, is a slippery slope. It blatantly devalues those stories with low rankings which I'd think would only serve to demoralize the reporters and subjects of those stories and, as a reader, if I'm looking at a story with an importance rating of say "2 of 10" I might find myself thinking well why are you wasting my time. I think the story itself and accompanying info (placement/display on index pages, number of photos, related stories, comments on message boards) should carry the burden.

Posted by Nathan Ashby-Kuhlman on January 6, 2004, at 9:02 a.m.:

I agree that actually displaying some kind of numerical importance ranking is a bad idea, primarily because of the problem Sara mentioned, which is that data would just be fairly irrelevant info-junk since it wouldn't be accurate over time. For United States news sites, on Feb. 1, 2003, clearly the breakup of the space shuttle Columbia would have merited the top-of-scale ranking. It was the biggest story of the year at that point. But if you already used your top ranking, what would you have done two months later, when the United States was at war in Iraq -- a bigger story? (My ranking of those two events is just an opinion, of course, which is a big problem Julie mentioned.)

All of that said, I have to disagree with Sara. I think the presentation in print, specifically the page number, speaks to its importance more than the text of the story itself does. In print, we can quibble all we want -- because of our different opinions -- about where a story belongs on page one relative to the other stories on page one, or whether another story belongs on page four or five, but those are fairly trivial distinctions. The main point I'm seeing in Adrian's blog entry is that in print, there IS a big distinction between a page-one story and a page-twenty story, but news sites (especially on automated archive pages) do not always do a good job of conveying that distinction. Readers in print would be very confused if a random selection of stories were put on the front page each day. So online readers probably get confused too if the section index page orders stories without some sense of relative importance.

Personally, I'm very much in favor of putting a list of "other articles in this section" in the rail or footer of article pages, with the current article included in its ranked position in the list. It's not really what Adrian was talking about, but I think showing a list of actual headlines that are "more" or "less" important than the one you're viewing gives you a better sense of relative importance than a number would.

Posted by Paul on January 6, 2004, at 1:28 p.m.:

I'd hate to be the editor forced to choose to give the Britney Spears story a numerical value higher than the Mars rover, or the latest combat toll in Iraq.

Posted by Derek Willis on January 6, 2004, at 5:24 p.m.:

As a first step toward achieving this goal, newspaper websites could provide lists of stories within a section for the past day/week/month broken down by what page of the paper the story appeared on. You could then see, for instance, all the page 1 political stories or page 1 crime stories, etc. Some sites include this information anyway, but don't provide a breakdown.

Posted by Vin Crosbie on January 6, 2004, at 11:04 p.m.:

You don't need to invent a ranking system for story importance; that would mean reinventing the wheel. A story prioritization system is already in use everyday in the news industry. It has been for more than a century. Moreover, its latest version is an accepted XML standard. The problem is that most newspaper Web sites haven't yet realized it exists.

The system originated back when the importance of a story set the priority for sending it over telegraph lines to newspapers. Its priorities were 'Flash', 'Bulletin', 'Urgent', 'Routine', and 'Deferred'. Whenever a telegraph operator at a relay station received a message that started with a 'Flash' heading, he'd shout that word to all other operators in the room, who would immediately stop sending other telegraphs and would keep their wires open and ready to relay the 'Flash' message. Western Union originated this system in the 19th Century, and the Associated Press, the United Press, and the International News Service (which merged with UP in 1958 to form UPI) adopted it early in the 20th Century.

Twenty years ago, when manually sent telegraph code (7-level TTY or 8-level TTS) finally gave way to computer-to-computer ASCII communication, the news industry adopted an ASCII version ( Its priorities were changed to 'Flash', 'Bulletin', 'Urgent', 'Rush', 'Daily' (for all other spot news items), 'Release at Will' (deferable), and 'Advance' (for pre-written copy being sent in advance of its publication date). Originally known as ANPA 1312 [American Newspaper Publishers Association], this coding scheme was adopted worldwide by the International Press Telecommunications Council. All wire services use it. All newspapers' and broadcasters' news front-end computer systems nowadays operate with it.

And now that ASCII hard coding is giving way to XML, the IPTC has been adopting XML versions of this schema (, such as NewsML, NITF, and SportsML:

There's even ProgramGuideML for TV and radio schedule guides:

These XML versions have existed for a few years. Reuters and some other news organizations are already using these for print and online editions.

Why aren't more newspaper Web sites using it? My guess is because most online production systems were designed primarily to shovel <i>local news</i> stories from print to online. Hardly any print newsrooms bother placing any priority code on local stories because they instead use print layout placement as their prioritization schema, something that doesn't quite translate online.

So, if you want a coding schema for online story prioritization, don't reinvent the wheel. One already exists in XML and and it's accepted worldwide. It's just that most newspaper Web sites haven't realized that it's there.

Posted by kpaul on January 7, 2004, at 1:58 a.m.:

I want to see Google (or someone! ;) invent AI of some sorts to auto-rank stories based on past user interaction with the site - could something be coded complex enough to create a unique homepage for thousands of users?

IIRC, the BBC was messing around with something like this (the section colors changed depending on how the visitor had moved through the site previously.

You'd still have human editors of course, to pick which stories get throw out there, but I think eventually it will be scripts or subroutines deciding which ones to play up on the page.

I'd love to see some inside numbers on Google News traffic up til now. I imagine it's growing, but I wonder how fast.

And yes, NewsML and such are very cool when you can get the newsroom to populate desc fields.



Posted by Peter on January 7, 2004, at 2:14 a.m.: already does something like this: they assign important stories a special icon that indicates that it's an important story with a wide impact (I forget the name of it - there's none right now).

Posted by Peter on January 7, 2004, at 2:17 a.m.:

I just found an example:

Scroll down a bit to see the HIGH IMPACT icon. The icon is also shown on the article overview pages. I kinda like this approach - they don't overuse it.

Posted by Dave on January 7, 2004, at 3:05 a.m.:

To some extent, web news editors already assign priority to news stories.

They sort them into lists, top to bottom, for presentation on home pages and inside section home pages. So we're 80 percent of the way towards a priority-assign system on the editorial side.

But I would argue that there are multiple measures of priority. Timeliness -- which underlies the comments before mine about Flash, BUlletin, Urgent, etc -- is one. Popularity is another. The Editors' judgement of importance is a third. And one can subsort by location. Washington Post online readers may be more interested in suburban coverage where they live than in the Beltway stuff.

All of this argues for a system where the reader determines first priority, but makes other judgement systems available to him.

Every home page and section home page should allow the reader to sort by:

> Editor's Picks

> Most Read Stories

> Most E-mailed Stories

> Newest Stories First

There must be other data sorts available here.

Posted by Anthony Moor on January 9, 2004, at 7:54 p.m.:

Timeliness changes not only by day, but also by the minute. A living and breathing Web news sits may have a story of huge importance at 4:30 p.m. dive to zero at seven: Say, a big accident on a major artery.

Posted by the house of real estate on January 12, 2004, at 5:43 a.m.:

It can't be done. The importance of a news story is relative to the person reading it. If there is a car crash and someone dies, I really don't care. It happens many times every single day of my life. Now if the accident killed my mother, that's the most important story ever to me.

I think having an easily searched news archive is of far greater value, because each individual can find stories that are important to them.

Comments have been turned off for this page.