Phil Sim

Web, media, PR and… footy

Google News fatally flawed

By now if you’re a web afficianado you’d know that Google News is out of beta . Big deal. The real meat in this story is Google’s launch of a personalised news service , which it seems the company has been waiting to complete before it ripped up the beta sticker.

As I mentioned in a previous post, we’re building a content aggregator as our very first “labs” project so I’ve spent a lot of time analysing what I feel are the strengths and weaknesses of the various aggregators out there. So where do I see Google News and what impact will its new recommendation make?

I use Google News every day. MediaConnects produce a daily column each day in which we analyse the day’s Australian tech media coverage, giving credit to journos for breaking news stories. A Google News search is invaluable for us in terms of trying to determine if a story was exclusive, how it might have been sourced and who broke the story first. For me, that’s always been the power of the service – searching. It’s a great “journal of record” but I tried using Google News as a gatekeeper for barely a day before giving up on it.

The fundamental flaw with Google News is that its based primarily on clustering, not ranking. It’s ability to rate a story is limited to analysing how many times a particular story is reported. So it has enough intelligence to pick out the most reported story. Woop-de-doop. That’s equates to almost zero value add because by the time a story makes it to prominence on Google News, it’s already been reported by every man and its dog and subsequently Google News can only ever be a follower. Can you imagine a newspaper editor saying “I’ve got this really great concept. We’re going to concentrate on reporting all the news, that everybody else has already reported!”

At best, the Google News news pages serves a purpose as a backup source, enabling a reader to ensure they haven’t missed any big news stories. Again, there’s that journal of record aspect.

The other related flaw with Google News is it has no ability to rate stories within a cluster. So where memeorandum gains the intelligence of ranking stories in a thread, by looking at the number of links a post receives, Google News has no such intelligence. I presume it makes a decision based on how long ago a story was posted and by the credibility of its various news sources it tracks. Hardly, a sophisticated way of picking out the best coverage of a particular story.

Now Google is also recommending news stories based on your search history. So how does this change the game for Google? It improves the service in one manner. Google News has large overarching topics which are really too broad to be of any use as an aggregator. So at least, you’ll be able to keep tabs on narrow subject areas. However, Google’s got a better killer app here. It’s Google News Alert. I subscribe to all stories related to the football team I follow and have those emailed to me. The value-add here is that I’m alerted to new stories even if I’m not checking the web or my newsfeeds.

The problem for Google News is the flaws I’ve already outlined, are only exacerbated when it comes to the effectiveness of its recommendation engine. According to the Google News help page, (hat tip to Search Engine Watch ) this is how the recommendation service works.

By signing in to personalized news and keeping Personalized Search enabled, you allow Google to track and save your news selections. Then, Google News can automatically recommend relevant stories just for you by using smart algorithms that analyze your selections. The algorithms compare your tastes to the aggregate tastes of other groups of similar Google News users. Simply put, we recommend news stories to you that have been read by many other users who’ve also read similar stories as you in the past.

So you get served up other popular stories. But chances are, a story is only popular because Google’s flawed rating system has put it there in the first place. It’s very much a case of the blind leading the blind.

I’d be very surprised if the people behind today’s leading aggregators are shaking in their boots over these latest developments. Google News will always have its place and the fact that its a Google product means it will always be an influential aggregator. However, as a gatekeeper that points people to new, interesting content, I’m afraid it needs a fundamental redevelopment. That’s highly unlikely based on the fact the service is now out of beta.

So content aggregators rejoice. Squash’s bet is that this is one corner of the Internet that Google won’t be dominating anytime soon.

Filed under: Content Aggregation, Google, memeorandum

13 Responses

  1. Gabe says:

    Phil: I’m happy you don’t work for Google.

  2. “That’s equates to almost zero value add because by the time a story makes it to prominence on Google News, it’s already been reported by every man and its dog and subsequently Google News can only ever be a follower.”

    I don’t agree with this statement. The far majority of people don’t need or want realtime or even near-realtime news. Most people still rely on relatively old-world ways to get news such as through the evening news on the tv or listen to news radio on their commute home. Those that do use the Web for news, use sources such as Y! News, G news, cnn.com, .com, etc. No, these sources are not good at providing real-time news because they incorporate either algorithms or human-editors (or even some combination of both) to find which news items are most newsworthy and only present those to their readers. For the typical person, though, this is exactly what they want: A high-quality snapshot of the day’s news (or past 6 hour’s news..whatever)
    Also, you bash clustering but is Memeorandum’s use of back-links really *that* different in concept. Many newspapers use Reuters, AP, etc as news sources and even though they have journalists which create fresh copy for their respective publications, in a sense the article is a backlink to the same news event. Thus what you call clusters, many articles on the same news event, is essentially the same as many bloggers backlinking.

    I do agree with you that the use of search history to customize news is a flawed concept.

  3. Phil Sim says:

    Rishi, what I’m saying is that for the type of news consumption your talking about, I would prefer a human-edited site than a computer-generated one. When you consume media you have favourites, because you like their “voice”. Even when you look at newspapers, they’re voice is far more specific than Google News. I buy a newspaper that is targetted at my city and at my demographic. The best Google News can offer me is an aggregation based on the country I live in.

    So Google News’ isn’t really doing anything AS A GATEKEEPER than human-edited mainstream media isn’t already doing. And when your local newspaper goes online, and also starts clustering like stories, probably using Google News as a source why would you go through their front page?

    Where aggregators come into their own, is where they allow you to produce media that you couldn’t easily do with human editors. I think Google News will need to do much narrower subcategories and try and go down the long tail route, but even then can they aggregate better than say a niche blogger. Probably not. Especially a niche blogger, who makes good use of Google News Alerts anyway.

    Memeorandum clusters, but it also ranks the stories it clusters, which is where the value exists. Google News clusters and shows you the biggest clusters but you then need to wade through that cluster to find your news. Did you try doing a Google News search during MacWorld? You’d get a couple of hundreds results and no way to pick out the most interesting stuff. That’s whats different. In some respects the term content aggregator is misleading, the real game is about content ranking.

  4. Google News finally out of beta (updated)

    Harray, Google News is finally out of beta, but it still has a fundamental problem. Google clusters news stories from a range of sources, so if something appears with a dozen references and a link such as “all 415 related »” then you know it’s a bi…

  5. Eric Berlin says:

    “I presume it makes a decision based on how long ago a story was posted and by the credibility of its various news sources it tracks. Hardly, a sophisticated way of picking out the best coverage of a particular story.”

    That’s about the best description I can imagine for how GNews works.

    I wonder if the future of news search lies in personalized search (a la the new GNews feature) or intelligent search (a memeorandum model enhanced with human [editorial] and technical developments)? Perhaps a combination or something no one (or at least me!) has even thought of yet.

    Eric Berlin
    Executive Producer
    Blogcritics.org

  6. kashmir says:

    I would only like to comment on the Personliased News.

    The news it recommended to me was what I would have never wanted to read, I only wonder how did it analyse my previous news search! I stopped using Google News after seeing it’s recommendations.

    Google alert is good, but I would still prefer Yahoo alerts as Yahoo provides much better alert content than Google. My focus of interest is Kashmir (a south east Asiana region) Yahoo news have a category for Kashmir Conflict and the news is properly aggregated while as Google simply searches for the word Kashmir in news feeds.

    Just my two cents.

  7. […] For more details read [this] review […]

  8. Thoughts on Google News Recommendations

    Greg Linden, founder of Findory, recently weighed in on Google’s announcement about recommended news stories via personalized search. Greg has obviously given much thought to the subject, as the goal of Findory is to build a personalized newspaper for…

  9. Hace says:

    I’m wondering what Google’s long-term plans are with GNews or news delivery in general, since I’m certain they recognize GNews’ shortcomings as a method of finding breaking news. Currently they seem to haveessentially developed a nifty search/personalization algorithm and are now desperately trying to find more applications for it.

    Google has been all about personalization (or gathering personal information, rather…) for a while now, and while a level of that is certainly useful on it’s own in GNews to provide what Rishi Khaitan aptly called a “snapshot of the day’s news”, it still doesn’t address this lagging behind-issue you were talking about in the first place.

    In my opinion they shouldn’t change GNews itself, since it has it’s aforementioned uses, but add, say, GNews Now! (replace with your own Xtra Cool trademark…), which would work on the same basis that digg.com does. Certainly it’d need alot of work, but I’m sure with the personalization Google already has the user-added links to breaking news could be filtered to people who’d read them and just click a “This is good news”-button. Enough clicks raise the news item to the personalized homepage of people who are logged in etc.

    I’m leaving issues with duplicates, false news etc. to the Google engineers 🙂

    Just my somewhat lengthy two cents. Sorry, I can’t write short…

  10. Flaws?? says:

    So can news stories be linked bombed, to boost their prominence?

    from where?

    Which news links are being monitored?
    I noticed A-list bloggers are only scanned every 2 days,

  11. free online credit report…

    predicts protecting finalizing!inquisitiveness Ralston:compromisingly Bartlett free credit report http://www.secured-credit-report.com/

  12. kip says:

    I’m love this great website. Many thanks guy

Leave a comment

Categories