14 Comments

Summary:

A new iPad app called Zite is the latest entrant in the race to build the “Daily Me,” a personalized newspaper that learns what you like. No one has really won this race yet, although Flipboard appears to be leading — and where is Google?

Zite-screenshot

Ever since the web first started to become mainstream, there have been attempts to build the “Daily Me,” a personalized newspaper that learns what you like or are interested in (does anyone remember PointCast?). But as I noted in a recent post on the topic, many of these efforts are lackluster at best, and irritating at worst. They either require too much fiddling to tune them, or they don’t show any intelligence at all (or both). But that doesn’t stop companies from trying — and the most promising entrants in this race so far are those that try to build their recommendations on top of the social signals coming from Twitter and other networks.

The latest to join the field is a personalized magazine app for the iPad called Zite, whose name is a play on the German word “zeitgeist,” meaning “the spirit of the times.” The company behind the app is based in British Columbia, and has been funded by angel investors and research grants from the Canadian government. CEO Ali Davar says Zite has been working on its recommendation engine for several years. An earlier version of the project, which is based on technology developed at the University of British Columbia’s Laboratory for Computational Intelligence, involved a browser extension called Worio that suggested related results when users did a Google search.

The Zite app pulls in your Twitter account and your Google Reader feeds (if you have them), then suggests topics based on your interests. This was the first place where it fell down for me — it said that it didn’t have enough information about me, which I thought was odd, since I have been on Twitter for about four years, have posted more than 35,000 tweets and follow over 2,000 people. I’ve used Google Reader for years as well, and am subscribed to about 600 feeds. Although Zite got some of its suggestions right, it recommended Barcelona as a topic, which was totally out of left field — in fact, I can’t recall ever mentioning the Spanish town before.

Although Robert Scoble says Zite doesn’t feel as slick as Flipboard, I thought the app worked quite well in terms of usability — you can swipe to move through articles, click to read them in a built-in browser, and share them easily (although you can’t save them to Instapaper, which is a shame). And you get asked with each one whether you like the content or want to see more of it, which is something other apps and services such as Flipboard are missing. It requires some effort on a reader’s part to do this training, and many probably won’t do it, but it is crucial for learning likes and dislikes.

One glaring omission from Zite is the lack of Facebook integration. Davar says Facebook tends to provide sources that are too heterogeneous (that is, too diverse) to be a source of good recommendation data, and that might be true, but it’s still a giant social network and a huge part of many people’s online news consumption, so it seems odd to leave it out — especially when the data coming from the billions of “like” buttons scattered around the web could be a source of so much data on what people want to read (Yahoo Labs has just released an interesting survey of what that information shows about the popularity of news at some major websites).

There’s one nagging question that keeps jumping out at me as I look at all of these apps and services, however, and that is: Where is Google? The combination of smart aggregation and algorithm-driven personalization seems like something the search engine should be all over. Google News has added some personalization aspects, but they are anemic at best, and one of the original customized news-readers — Google Reader — hasn’t really capitalized on that opportunity much at all (although it does provide some recommendations for readers related to new feeds).

The reality is, the RSS reader has been eclipsed (for the small proportion of the population who even used one) by Twitter and Facebook and other social news sources, or smart aggregators such as Techmeme and Mediagazer. Google has more or less failed to take advantage of that transition at all when it comes to news reading, although it is trying to add social signals to search. Why not take FastFlip and try to make it a Flipboard or Zite or News360 competitor?

And apart from the Washington Post‘s new Trove project and the News.me spinoff from the New York Times that Betaworks is close to launching, newspapers — who should know a thing or two about filtering and recommending the news to people — are virtually nowhere in this game.

If there’s one thing that web users need more than ever, it’s smart filters to help them navigate the vast tsunami of information that comes at them every day. (The big problem isn’t information overload, says Clay Shirky, but rather “filter failure”.) Someone is going to solve that problem, and if they do it properly, they could wind up capturing a significant share of the online news-reading market.

Related GigaOM Pro content (sub req’d):

Post and thumbnail courtesy of Flickr user Retinafunk

You’re subscribed! If you like, you can update your settings

  1. Alberto Nardelli Wednesday, March 9, 2011

    Intelligent filtering is definitely an exciting space.

    At Tweetminster (www.tweetminster.co.uk) we’re trying to tackle the opportunity slightly differently – by identifying influencers around any topic and analysing their networks to see what they’re paying to.

    We then use that data to aggregate content. What you see featured on Tweetminster is content filtered by networks of experts. The layout and size of the story on the page is controlled by data. There is no human or editorial intervention.

    The goal is to provide relevant content to users who may not even be on Twitter or have the time to create lists, set up feeds and build networks of people to follow.

    Our current focus is on UK politics & current affairs, but will soon be expanding into new topic areas.

  2. A few comments:
    First, I do not, personally, use Facebook for any news consumption. I think this is based largely on my age but I find all the links on Facebook are to dumbass YouTube videos and Mashable articles. On the flip side, I find Twitter a great source of news because I have personally vetted every person I follow and thus know they’re likely posting useful content.

    With respect to the daily me, I think someone needs to build a personal recommendations algorithm. Just like Google has search algorithms and Facebook has network algorithms, someone needs to identify all the ways to rate content and assign weightings to that. The ways to rate content might include page rank, Facebook Likes, Tweets, number of comments, reputation of publisher, etc. That algorithm then needs to outline that a Facebook Like is worth 10% of the recommendation while a Tweet might be worth 15%. Using that algorithm, if the content is likely of interest to the reader (based on previously read articles) and has a high enough ranking then it would be presented.

    I’ve randomly tried different content recommendation engines in the past but have yet to find a useful, usable one.

    1. I agree, SL — that’s pretty much the holy grail of recommendation, what you’ve described, and I think a bunch of different people are coming at it from different perspectives. Even Google is supposedly trying to develop something called “author rank” to determine authority on social networks.

  3. Great article Matt. User centric experiences are a must given today’s real-time content environment. At kikin we are solving this problem with better contextualization and personalization based on your social sources, favorite sites and of course search. I’d love your opinion on kikin and how we are making the web more enjoyable for consumers.

    1. I will take a look — thanks for the heads up.

  4. Great post. There’s a critical need for the same “smart filtering” solutions in the enterprise space as well.

    There are lots of inputs that can be used to drive an information recommendation engine. In the enterprise Facebook isn’t a useful driver for a recommender.

    Capturing and using the implicit actions and behaviors of a given user provides a treasure trove of input the process however, and allows the system to continually learn to provide better recommendations to each user. What do you read? How long do you read it? Do you share or comment?

    From this information we can build a powerful, comprehensive profile of each user’s interests. This profile drives recommendations that are higher quality and importantly, much quicker to react to changes in your interests.

  5. I don’t use twitter, but wouldn’t following 2,000 people consume all of your time?

    1. I don’t follow everything that each of those people say, Ken — that would be overwhelming. I mostly use lists that I have put together of people who tweet about different topics, and then I kind of dip my toe into the main stream from time to time.

  6. The idea of intelligent agents filtering and finding information of interest has a lot of value. On the other hand, I find that there are many different techniques that the filter and clipping approach doesn’t touch in value or surprise factor (delight).

    RSS is very efficient for my needs.

    For example, I use iGoogle to organize a single view of my favorite news sites, blogs and magazines using RSS widgets. It’s both convenient and it allows me to scan and pick what I find relevant or of interest at that moment. My information vocabulary is dynamic and rarely static unless it’s project or subject focused. By using the one start page, I can tune it as needed but it’s always there.

    I have also found that I prefer viewing the New York Times as a web site on my iPad versus their custom app for one simple reason; I scan the front page in a zig zag pattern and then landing on those articles that “grab” me because they’re on the front page and in editorial context. Do the digital newspaper gurus get that?

  7. “there’s one nagging question… where’s google?”

    google, unlike countless others, realizes they are in THE ADVERTISING business.

    i think they are quite happy going after the income side of the business while leaving the expense side stuff to others.

  8. Totally agree there’s a need for “smart filters” to curate news and my company YourVersion has its own approach. You can add any topic you want to your profile and YourVersion always bring you the latest relevant news, blogs, tweets, videos, and Quora Q&A.

    The YourVersion product suite spans our website, browser tools, and free apps for iPad, iPhone & Android. We’d love to hear your feedback on our iPad app http://bit.ly/yv-ipad

  9. Matt,
    Great article. I noticed that Zite like Flipboard grabs the entire article from it’s source, this doesn’t appear to follow fair use etc. What in your opinion seems to be the approach for publishers when apps do this over say web based news aggregators. From what I am seeing there is a difference in what is ok for apps vs. Web aggregators.

    1. Good call on that one Jeremiah. I’d be interested in Matt’s take on this as well.

      Generally I think the party line here on the technologist’s part is that their scraper gives the pageview to the publisher so it’s all well and fair. I’d question whether that’s really a good thing for publishers, however, since it comes at the expense of some of their advertising customers and erodes data quality by, for example, watering down conversion metrics.

      Moreover, if Zite is indeed scraping (i.e. “grabs the entire article from source”), it would be a pretty brittle approach technology-wise as all I would have to do is rename the ID or class selector in order to break that content extraction. The somewhat slow release cycle on an iPad would have a hard time keeping up with any sites that did this and sites could break for an indeterminate amount of time.

  10. Thank you for the update on Zite ,Mathew.I found it very interesting.

Comments have been disabled for this post