Twitter  The newsonomics of the Kyiv Post’s embattled work nie.mn/1l87jD2  
Nieman Journalism Lab
Pushing to the future of journalism — A project of the Nieman Foundation at Harvard
zebras-herd-personalization-cc

Who should see what when? Three principles for personalized news

There’s too much news for anyone to consume. Three key words should determine who gets served what: Interest, effects, and agency.
Email

I really don’t know how a news editor should choose what stories to put in front of people, because I don’t think it’s possible to cram the entire world into headlines. The publisher of a major international newspaper once told me that he delivers “the five or six things I absolutely have to know this morning.” But there was always a fundamental problem with that idea, which the Internet has made starkly obvious: There is far more that matters than any one of us can follow. In most cases, the limiting factor in journalism is not what was reported but the attention we can pay to it.

Yet we still need news. Something’s got to give. So what if we abandon the idea that everyone sees the same stories? That was a pre-Internet technological limitation, and maybe we’ve let what was possible become what is right. I want to recognize that each person not only has unique interests, but is uniquely affected by larger events, and has a unique capacity to act.

If not every person sees the same news at the same time, then the question becomes: Who should see what when? It’s a hard question. It’s a question of editorial choice, of filter design, of what kind of civic discussion we will have. It’s the basic question we face as we embrace the network’s ability to deliver individually tailored information. I propose three simple answers. You should see a story if:

  1. You specifically go looking for it.
  2. It affects you or any of your communities.
  3. There is something you might be able to do about it.

Interest, effects, agency. These are three ways that a story might intersect with you, and they are reasons you might need to see it.

But turn them around and they say: if a story doesn’t interest me, doesn’t affect me, and there’s nothing I could do anyway, then I don’t need to see it. What about broadening our horizons? What about a shared view of unfolding history? The idea that we will each have an individualized view on the world can be somewhat unsettling, but insisting on a single news agenda has its own disadvantages. Before getting into detailed design principles for personalized news, I want to look at how bad the information overload problem actually is, and how we came to believe in mass media.

Too much that matters

A solid daily newspaper might run a couple hundred items per day, just barely readable from cover to cover. Meanwhile, The Associated Press produces about 15,000 original text stories every day (and syndicates many times that number) — far more than one person can consume. But the giants of journalism are dwarfed by the collaborative authorship of the Internet. There are currently 72 hours of video uploaded to YouTube every minute, which now houses more video than was produced during the entire 20th century. There are 400 million tweets per day, meaning that if only one tweet in a million was worthwhile you could still spend your entire day on Twitter. There are several times more web pages than people in the world.

All of this available information is a tiny fraction of everything that could be reported. It’s impossible to estimate what fraction of stories go “unreported,” because there is no way to count stories before they’re written; stories do not exist in nature. Yet from the point of view of the consumer, there is still far, far too much available. Ethan Zuckerman has argued that the limiting factor in foreign reporting is not journalistic resources, but the attention of the consumer. I suspect this applies to most other kinds of journalism as well; raise your hand if you’ve been carefully following what your city council is up to.

Compared to the news, there is simply very little attention available.

For the single-issue activist, the goal is attention at any cost. But editors have a different mission: They must choose from all issues. There is a huge number of potentially important stories, but only a tiny fraction can be “headlines.” Most stories must languish in obscurity, because you or I cannot hope to read a thousandth of the journalism produced each day. But even the flood of global journalism is a tragically narrow view on the world, compared to everything on the Internet.

How, then, should an editor choose what tiny part of the world to show us? Sometimes there is an event so massive, so universal, it demands attention. Natural disasters and revolutions come to mind. For all other stories, I don’t think there is an answer. We can’t even agree on what problems are important. No single set of headlines can faithfully represent all that matters in the world.

There is more than one public

The Internet is not like broadcast technology — print, radio, TV. But the routines and assumptions of journalism were formed under the technical constraints of the mass media era. I wonder if we have mistaken what was possible for what is desirable.

The first technical limitation I want to consider was this: Everyone had to see the same thing. This surely reinforced the seductive idea that there is only one “public.” It’s an especially seductive idea for those who have the ability to choose the message. But there’s something here for the rest of us too. There’s the idea that if you pay attention to the broadcast or read the daily paper, you’re informed. You know all there is to know — or at least everything that’s important, and everything everyone else knows. Whatever else it may be, this is a comforting idea.

Media theorists also love the idea of a unified public. Marshall McLuhan was enamored with the idea of the global village where the tribal drums of mass media informed all of us at the same time. Jürgen Habermas articulated the idea of the public sphere as the place where people could collectively discuss what mattered to them, but he doesn’t like the Internet, calling it “millions of fragmented chat rooms.”

But the idea of a unified public never really made sense. Who is “us”? A town? A political party? The “business community”? The whole world? It depends on the publication and the story, and a few 20th-century figures recognized this. In The Public and Its Problems, written in 1927, John Dewey provided an amazing definition of “a public”: a group of people united by an issue that affects them. In fact, for Dewey a public doesn’t really exist until something affects the group interest, such as a proposed law that might seriously affect the residents of a town.

We can update this definition a little bit and say that each person can belong to many different publics simultaneously. You can simultaneously be a student, Moroccan, gay, a mother, conservative, and an astronomer. These many identities won’t necessarily align with political boundaries, but each can be activated if threatened by external events. Such affiliations are fluid and overlapping, and in many cases, we can actually visualize the communities built around them.

The news isn’t just what’s new

There was another serious technical limitation of 20th-century media: There was no way to go back to what was reported before. You could look at yesterday’s paper if you hadn’t thrown it out, or even go to the library and look up last year on microfilm. Similarly, there were radio and television archives. But it was so hard to rewind that most people never did.

Each story was meant to be viewed only once, on the day of its publication or broadcast. The news media were not, and could not be, reference media. The emphasis was therefore on what was new, and journalists still speak of “advancing the story” and the “top” versus “context” or “background” material. This makes sense for a story you can never go back to, about a topic that you can’t look up. But somehow this limitation of the medium became enshrined, and journalism came to believe that only new events deserved attention, and that consuming small, daily, incremental updates is the best way to stay informed about the world.

It’s not. Piecemeal updates don’t work for complex stories. Wikipedia rapidly filled the explanatory gap, and the journalism profession is now rediscovering the explainer and figuring out how to give people the context they need to understand the news.

I want to go one step further and ask what happens if journalism frees itself from (only) giving people stories about “what just happened.” Whole worlds open up: We can talk about long-term issues, or keep something on the front page as long as it is still relevant, or decide not to deliver that hot story until the user is at a point where they might want to know. Journalism could be a reference guide to the present, not just a stream of real-time events.

Design principles for personalized news

If we let go of the idea of single set of headlines for everyone based around current events, we get personalized news feeds which can address timescales longer than the breaking news cycle. Not everyone can afford to hire a personal editor, so we’ll need a combination of human curators, social media, and sophisticated filtering algorithms to make personalized feeds possible for everyone.

Yet the people working on news personalization systems have mostly been technologists who have viewed story selection as a sort of clickthrough-optimization problem. If we believe that news has a civic role — that it is something at least somewhat distinct from entertainment and has purposes other than making money — then we need more principled answers to the question of who should see what when. Here again are my three:

Interest. Anyone who wants to know should be able to know. From a product point of view, this translates into good search and subscription features. Search is particularly important because it makes it easy to satisfy your curiosity, closing the gap between wondering and knowing. But search has proven difficult for news organizations because it inverts the editorial process of story selection and timing, putting control entirely in the hands of users — who may not be looking for the latest breaking tidbit. Journalism is still about the present, but we can’t assume that every reader has been following every story, or that the “present” means “what just happened” as opposed to “what has been happening for the last decade.” But for users who do decide they want to keep up to date on a particular topic, the ability to “follow” a single story would be very helpful.

Effects. I should know about things that will affect me. Local news organizations always did this, by covering what was of interest to their particular geographic community. But each of us is a member of many different communities now, mostly defined by identity or interest and not geography. Each way of seeing communities gives us a different way of understanding who might be affected by something happening in the world. Making sure that the affected people know is also a prerequisite for creating “publics,” in Dewey’s sense of a group of people who act together in their common interest. Journalism could use the targeting techniques pioneered by marketers to find these publics, and determine who might care about each story.

Agency. Ultimately, I believe journalism must facilitate change. Otherwise, what’s the point? This translates to the idea of agency, the idea that someone can be empowered by knowing. But not every person can affect every thing, because people differ in position, capability, and authority. So my third principle is this: Anyone who might be able to act on a story should see it. This applies regardless of whether or not that person is directly affected, which makes it the most social and empathetic of these principles. For example, a politician needs to know about the effects of a factory being built in a city they do not live in, and if disaster recovery efforts can benefit from random donations then everyone has agency and everyone should know. Further, the right time for me to see a story is not necessarily when the story happens, but when I might be able to act.

These are not the only reasons anyone should ever see a story. Beyond these principles, there is a whole world of cultural awareness and expanded horizons, the vast other. There are ways to bring more diversity into our filters, but the criteria are much less clear because this is fundamentally an aesthetic choice; there is no right path through culture. At least we can say that a personalized news feed designed according to the above principles will keep each of us informed about the parts of the world that might affect us, or where we might have a chance to affect others.

Photo of zebras in Tanzania by Angela Sevin used under a Creative Commons license.

                                   
What to read next
INNlogo_blue
Justin Ellis    April 15, 2014
Chalkbeat, Southern California Public Radio, InvestigateWest and others are awarded over $236,000 in micro-grants to support events programming, collaborative reporting, and a “native underwriting” pilot program.
  • http://storyflow.net/ Martin B.

    Great article and really nice that it touches a bit on story guidance and navigation as much as personalization.

    What’s amazing is that even some 20 years into online news reporting, most of the major news sites suffer from exactly the same issue as print news did in the 20th century, namely that news can only be consumed on the day of its publication (in many case not even the whole day). Once a story is off a news site’s home page (and the corresponding section’s landing page), it’s basically impossible for the reader to find the story, even if he knew that it existed.. And in most cases the reader won’t really know what’s been posted yesterday or the day before (unless he’s read all the stories).
    In a way, this lack of a proper navigation mechanism on major news sites is exactly what’s preventing online news media from creating deep stories, threaded throughout time, without the need for actually linking every preceding story to the current one. What is needed is basically a “related stories” mechanism that doesn’t just give the reader three random links, but shows the evolution of a story, provides context and motivates to keep on reading all of the quality content that newspapers produce, both online and offline.

    Disclaimer: I’m working on http://storyflow.net which aims to achieve exactly that, though to be honest there should have been numerous decent solutions to this problem a long time ago already, it’s really amazing how open the space is.

    Martin

  • Tom Stites

    Fascinating piece.  I’ve already read it twice and intend to come back at least once more.  I’m accustomed to seeing McLuhan cited in journalism essays (i’ve been guilty myself) but I can’t remember seeing another journalism piece that also draws on Habermas and Dewey!

    I see the piece as having two significant strengths:Time flex — The future-of-journalism community pays huge attention to big breaking events and the role of Twitter and crowdsourcing in the moment.  But such events are few and far between — and in-the-moment coverage, no matter the tools used, tends to be significantly inaccurate for several hours at minimum.  What tends to be missing from the discourse is how people actually use other news, which is to say almost of it, so hooray for Jonathan Stray for understanding the need for journalism time to flex to fit readers’ needs.  This topic is worthy of way more discussion.Definition — The word journalism has lost fixed meaning, and however people work out their definitions they tend to end up with something they don’t like or trust much.  So Interest, Effects and Agency provide a fine framework for thought and discussion.  The definition I’ve been using in my Banyan Project work for the last three years is that journalism is the reliable information people need to make their best life and citizenship decisions.  This certainly doesn’t clash with Stray’s idea, and there’s important common ground:  Both approaches start with the needs and role of the readers, which is often lost in the conversation about what journalism’s about.  

  • TSteichen

    This article covers a lot of ground, so I won’t try to address  all the points.  Rather I’d like to share some thoughts on
    personalized news.

    The objective of a “personalized” web interface is to tailor a web
    page’s presentation to the interests of a particular viewer.  It seems
    more logical than it really is.

    The personalization concept seeks to infer a reader’s future interest
    from observations of that reader’s past behavior, and to use the data
    from those past observations to adjust the content and its presentation
    in a manner that will connect more strongly with that reader’s interest
    during subsequent visits.

    This concept is based on a number of questionable assumptions, including:

    (1) It assumes that the targeted reader will
    subsequently adopt the “personalized” page as a gateway point of
    departure from which that reader will access other content.  That’s
    increasingly rare.

    (2) It assumes that the visitor’s past interest can be
    inferred by observing that visitor’s past behavior.  That’s not
    uniformly or consistently true.

    (3) It assumes that the visitor’s present interest is the same as the visitor’s past interest.  That’s often not true at all.

    A far more effective approach is to focus not the readers, but on the articles themselves.  This approach works on the simple assumption that the reader has (for whatever reason) some interest in the article he or she is reading. 
    And by extension, it assumes that the reader of any particular article
    is reasonably likely to to read other articles (which come to his/her attention) appealing to the
    same type of interest as is associated with the original article.

    Implementing this approach starts by identifying characteristics of an
    article that appeal to a particular type of interest.  Other articles
    (and related materials) that exhibit these same characteristics are
    identified and linked to in the original article.  Each particular set
    of characteristics comprises a particular theme, and that theme serves as the means of organizing content related to it.  The
    theme-based links would be presented (as embedded links) to any reader of a particular
    article, regardless of the path that brought them to the article. 

    It’s admittedly difficult to effectively implement a theme-based link structure. 
    But doing so is, IMHO,  far better than spending a lot of time creating a
    personalization strategy that (because of the assumptions discussed
    above) is unlikely to be very effective.
     

  • 49ersGO49

     
     
    Thank you for this excellent article.
     
    I would like to take this to a further point, as it is similar to when FM came on board in radio to AM (more channels), as well as the original intent of cable television, which has taken perhaps 50 years or more to become what it is today (more channels)…and more coming.
     
    The internet is similar in that there are more “channels” and therefore more groups, rather than one large audience, and therefore a desire for news that interests the various audiences.  It is eliminating reference books, cutting in on paper fiction and non-fiction (perhaps also increase more books for more audiences) as their is profit in doing this.  And that is the problem in the news business.  It never was the subcriptions that made the profit, but advertising.  Community printed newspapers know this more than any other, and perhaps the web news is in the same marketplace.
     
    That stories are more entertaining, or need to be more entertaining,is as old as “The World” newspaper’s marketplace.
     
     
    The editor chooses the stories that he believes the readers will read. There may be some that he knows will not be as well read, but included for many reasons ( the editor’s decision).  The editor
    is to know his readers, and not only chose the stories but the employees who put them together, or direct them (news reporters, columnists are different, to a degree). 
     
    Right now we have “mixes,” as those who download music.  They take from various albums single songs and make their own album. Google, Yahoo, Newswhip, and others do this with news reporting. Neither originate, they “steal” — the editors, reporters, and the owner of the newspaper don’t receive a fee for their work—actually, they didn’t “copy it,” but provide a brief
    description with link—so they never published it!
     
    As in the music industry, users of the material on the interest should be paid for the use of their material.
     
    One day the internet sites like Google and Yahoo will wake up and realize they need their own newsroom to generate stories for their various offices.  Perhaps then reporters and editors
    will get paid for what they produce.  Hopefully that happens soon to replace those laid off in the printed news business.

  • http://www.facebook.com/profile.php?id=1407319 Taylor Wray

    I really like your subject matter but dislike your normative approach to the question.  Your criteria of interest/effects/agency definitely seem relevant to personalizing news feeds, but why frame it as some sort of grand moral theory when there is no single, objective value of ‘news goodness’ against which to weigh your criteria?  Will there be some high church of journalism that condemns those sinners who trespass against the criteria when they read irrelevant news stories? 

    I think an empirical approach is more beneficial here – one that recognizes that news feeds have ALREADY been personalized:  If I’m a lefty lib, I can tune in to MSNBC on TV, NPR on the radio, DailyKos and HuffPo online, NYT and the New Yorker in print – compare this to a Tea Partyer’s media diet of Fox News, Limbaugh, Drudge, National Review, WSJ and you have two VERY different news feeds.

  • http://twitter.com/eolsencreative Eric Olsen

    I can’t stop thinking about this concept. Whether or not like-minded humans are currently (or forever will be) better than Pandora of other recommendation engines for helping us find great new things. Here were my thoughts you inspired.
    http://www.allopinionsarenotequal.com/2012/07/finding-your-roger-ebert-human-filters.html 

  • http://arnereport.net/ Arne Kaufmann

    I think news sites should just wait a bit before publishsing their news story. It is so often that right after the first news story of one new thing has been published, something new happened too. Then there comes the second news article, and the first goes without a note to the next update – means it gets a big chaos of articles nobody can arrange right, as you described too. So why don’t they just wait a bit and deliver content understandable for everyone. Sure they lose their, “We wrote first about that”, but I think people value a single article explaining the topic good more than dozens articles better on time but without context and badly explained.