Google News Goof-Up
Posted by Harry McCracken | Monday, September 20, 2004 12:47 PM PT
Back when Google News debuted, there was a flurry of articles saying that this automated news-gathering service might be bad news for those people who (ahem) make their living as journalists--because it showed that a computer with an Internet connection could make pretty intelligent decisions about what the world thought was newsworthy.
I think Google News is terrific, and visit it several times a day. But I never understood the notion that it was a replacement for other news sources--what it's gathering is news reported, written, and edited by real live journalists. In other words, a world without journalists (and media companies willing to pay them) would mean a world without Google News.
And as good as Google News is, it sometimes shows signs of its automated nature. (Maybe that's why it's still labeled as being in beta, eons after it became an indispensable tool.) We've seen this with items from PC World's own Weblogs--when Google News picks them up, it consistently thinks their title is "Enter Your Email:," which is actually a phrase that appears on the left-hand side of the page in a sign-up box for our newsletters. (We're working on fixing this.)
Google News also gets confused by stories on the New York Times site, which display the first character as an oversized "drop cap" rather than an ordinary ASCII character. The drop cap is a graphic, so Google News can't understand it; it thinks the story begins with the second character.
And that's why it's currently showing (see below) a Times story on the Bush National Guard document flap that appears to call the TV news organization involved in the controversy by the, um, unflattering name "BS News." Oops! Talk about unintended editorial comment...
Harry,
I think one of the most interesting shortcomings of Google news is its choice of sources. It would seem a geographically close source would provide more depth than a distant one, yet I often see US stories - stories with local interest in one city - being cited by Google using Asian news services. That's nuts. I don't care how good Xinhua is, I'll get better coverage on a Cleveland, OH story from a Cleveland, OH source.
I also noted that when stories about Bill Clinton's heart surgery appeared, they were almost invariably accompanied by Hillary Clinton's picture - I guess Google's computer sees the name Clinton and assumes Hillary is the one making news these days
when you use offshore programmers you get the chaff, or was the translation shaft? The US has yet to reap the sad benefits it's infrastructure is accumulating with non-English speaking programmers and IT personnel. Your insurance companies, like Nationwide now has very nearly unsupportable software, but media is first in line for the deviation since it is "sound byte" required for fast turnaround.
Google News is interesting and valuable because it is different, quirky, international. Most of the proposed improvements would simply make it like every other news site.
I love the variety, even in the mistakes, which require
you to actually use judgement and investigation to
establish the truth. It's almost like having to place
analysis before belief, instead of afterward
(...gasp!!!)
What trully telling is how the news gets distorted. You'll frequently see headlines of the same story being reported in *completely* different ways - sometimes actually being diametrically opposed.
There's been an incident of 'BS News' appearing on the Google News home page today (viewable at 9:53AM ET on 23-SEP) that isn't so easily explained by the drop-cap thing. I grabbed a screen shot; if they update the site before you can see it I'll send a copy.
Here's the text:
NR Editors
National Review Online - 1 hour ago
Twelve days after airing a story that questioned George W. Bush's National Guard service - a story built on forged documents BS apologized.
Geoff Fox's comment is an interesting one. However, the news story (for example) on Ohio, written by an Ohio journalist, could well be picked up by one of the newswire services - indeed, it might have even originated there - and re-reported in an Asian paper, verbatum. The advantage with Google pulling results from other parts of the planet is the *time differences*. The Asian publication may be 12 hours ahead in real time than the Ohio one. Same source, but earlier delivery and access.
Reacting to Geoff Fox's comment: Google News doesn't try to give you the most appropriate article for a story. It's just trying to give you the most recent. Often this means something that's just completely repeating a better, earlier article. Many times this earlier article was written days previously, and more recent information is available.
This effect could be lessened by the algorithm including something to spot extensive quoting; possibly tracking both repeated text and mentions of other news sources.
Excellent, that was really well explained and helpful
online casino