“Newsless”?
Many people use the terms “news” and “journalism” interchangeably. But what is news? I think most folks would say it means what’s new, accounts of the latest developments affecting some corner of the world.
Until recently, newspaper editors defined news as “important developments over the past 24 hours.” Editors of newsmagazines might expand that time horizon by a few days; Web editors will contract it to within a few hours. But there’s no escaping the time-bounded nature of “news.”
My understanding of journalism is broader. To me, journalism is the constant effort to deliver a truer picture of the world as it is. The “latest developments” provide one lens through which to capture that picture. And as long as journalism was primarily delivered by static media, that lens made perfect sense.
The Web, however, makes possible other ways of delivering that picture of our evolving world. It allows us to shirk the tyranny of recency and place more emphasis on context – the information that often gets buried beneath the news.
The title of this blog is a provocative misnomer. I don’t think news is going anywhere anytime soon, and I certainly think it remains a useful way of hooking our attention into the context surrounding the latest developments. But I do want to end the headlock news has placed on journalism. For all our handwringing and speculation, our conferences, our books, etc., news is as old as humanity and will survive us all. What ails in journalism – and what we have the opportunity to fix – is context.
I want to hear much, much less about the future of news, and much more about the future of context. I want to shift the focus of our books and conferences from how we’ll deliver the latest developments to how we’ll help our audiences better understand the state of our world.
For the next nine months, I’ll be at the Reynolds Journalism Institute at the University of Missouri, attempting to lay out a vision of a news website centered around context rather than time. I’ll be blogging my explorations and discoveries here, and welcoming your insights. Journalism has a moment of great opportunity before it. Let’s figure out how to rise to the occasion.
No related posts.
[...] started a blog devoted to exploring an alternative. He writes in the introductory post: Until recently, newspaper editors defined news as “important [...]
The New AP - Publishing 2.0
8 Oct 08 at 8:10 am
[...] started a blog devoted to exploring an alternative. He writes in the introductory post: Until recently, newspaper editors defined news as “important [...]
The New AP » Publish2 Blog
8 Oct 08 at 10:54 pm
[...] into a coherent, compelling structure — is one of the basic premises of my work at RJI. In the inaugural entry on Newsless.org, I put it this way: “I want to hear much, much less about the future of news, [...]
Ten questions for journalists in the era of overload at Newsless.org
31 Dec 08 at 1:48 am
Do not underestimate the power of print to help create context. What I mean is that the missing element is time and place. The web lives in NOW and the place is the SCREEN. It may well turn out that web and print are joined, each playing to it’s unique media strengths various solutions emerge.
For example.
Consider using the long tail of “snapshots of history” found at most serious newspapers to produce paperback books or special editions about a Place.
Why couldn’t those sold into K-12 to help our kids develop a context that they can take with them for the rest of their lives?
The news organization website then might be freed from it’s need to “advertise by selling “stuff” that people will buy. Another beneficial effect would be to eventually break the strangle hold that textbook publishers on K-12 (not to mentions college) education.
Consider a science reporter curating news for a weekly print edition. Read for free, pay for print. And choose your advertising VERY carefully.
Here’s a post I did a couple of days ago about Newspapers and the Dog Food Business. Maybe you’ll find something useful in there.
http://sellingprint.blogspot.com/2008/12/news-is-niche-market.html
Michael J
4 Jan 09 at 12:20 pm
I just had a chance to read your research proposal Brilliant! IMHO.
What about adding a print component? I did a post on how that might work at my blog at http://sellingprint.blogspot.com/2009/01/printed-...
Michael_J
7 Jan 09 at 2:23 am
I think there are strong possibilities for print, especially in this transitional period. I think the Politico's model is interesting – most of its audience is on the Web, but most of its money comes from advertising in its free, highly-targeted print edition. I don't know how much experimenting has been done on producing regular print editions for news junkies in a variety of topical/geographical niches. It's something I'd be very curious to try.
mthomps00
7 Jan 09 at 11:26 am
I have 30 years of experience in the print industry. But am presently semi-retired. I would be glad to share whatever I think I've learned either here or over at my website. Or get in touch via email. I'm on linked in. I think the tech is welldefined and in place.
Michael Josefowicz
7 Jan 09 at 11:43 am
[...] post you are at, National Explainer, and two from Matt Thompson’s project, Newsless.org (see Newsless? and Ten Questions for Journalists) you will absorb the idea that new information [...]
Online Journalism Rocks » Blog Archive » Ideas for projects - Journalism 453: Online Reporting and Editing
27 Jan 09 at 12:26 pm
Consumers want at least two things that we typically conflate as "news." First, they want to know what important things are going on in the world, or their backyard. Second, they want to know what each of these things mean – how do they fit into the stream of previous similar events.
This second component of news may also be called context.
The "trick" is how to discover that elusive characteristic and how to put it to useful work.
I suggest that one way (actually, the only way that I've been able to figure out) is to (in an automated way) discover the main underlying concept(s) of an article, and then retain that as metadata.
Once you have a common measure of concepts, you can easily use that to collect together related content, thereby creating the Holy Grail of context. With that context you can embed links in articles that, for the first time, provide readers with really useful access to more information on the same concept that had them reading the article itself.
Terry Steichen
4 Feb 09 at 8:39 pm
Terry,
While that AI approach may someday come to pass, my take is that a smart experienced person , supported by a Cloud based knowledge base – a wiki type thing – is the best way to connect the dots for regular people. For me connecting the dots is the operational definition of context. What is necessary is that the journalist needs a pretty articualted view of how the world works. Otherwise you get the patterning problem. Paranoids connect the dots into what can be a very compelling story. The trick is to do with an evidence based approach. Not easy. Hard to scale. But that's why people who can do that are valuable. It's the real value creation of great journalists.
Michael Josefowicz
5 Feb 09 at 2:26 am
@Michael
I don't see what we need as the ability to "connect the dots." Rather, we need a means to consistently and accurately identify and record the concepts underlying an article. The concepts become metadata, but generating it is a lot more challenging and involved than just collecting "entity" tags (persons, places, etc.) that is fairly common today. But if you're going to have linkages that let people track with their interest (and it's not worth much if it doesn't do that), it's essential.
I don't really see how this can be done manually for several reasons: (1) you need to process all previous articles in your archive (hundreds of thousands), (2) all present and future articles in real time, (3) applying precisely the same biases for each article, day-in and day-out.
A manual approach would create a 'single point of failure' that wouldn't be acceptable (should the person involved get sick or quit), and would – as you point out – not be scalable.
The technological solution isn't AI (artificial intelligence), but it is a more sophisticated categorization approach than newspapers are currently employing. I know of at least one implementation and there may be more, should the industry get serious about it.
Terry Steichen
5 Feb 09 at 3:37 pm
@Terry,
Here's where I see the problem: "Rather, we need a means to consistently and accurately identify and record the concepts underlying an article."
For example, this morning I found a piece about IBM joining with Google to make a connection between medical devices and electronic health records at Google. How would you identify the concepts in that story. IBM? Google? Health Care? Information infrastructure? Congress? Republicans v Democrats? Economic stimulus? Fragmented delivery of health care? etc etc etc.
These are just a couple of the concepts that were energized in my head upon seeing the headline. I'm sure that someone else who follows these things would have a different set of concepts. I would connect the dots one way. Someone else would connect them another.
It's not a manual approach. It's more like what Gladwell describes as Blink. I think great journalists blink all the time. i think computers cannot blink.
Michael Josefowicz
5 Feb 09 at 4:39 pm
@Terry,
Here's where I see the problem: "Rather, we need a means to consistently and accurately identify and record the concepts underlying an article."
For example, this morning I found a piece about IBM joining with Google to make a connection between medical devices and electronic health records at Google. How would you identify the concepts in that story. IBM? Google? Health Care? Information infrastructure? Congress? Republicans v Democrats? Economic stimulus? Fragmented delivery of health care? etc etc etc.
These are just a couple of the concepts that were energized in my head upon seeing the headline. I'm sure that someone else who follows these things would have a different set of concepts. I would connect the dots one way. Someone else would connect them another.
It's not a manual approach. It's more like what Gladwell describes as Blink. I think great journalists blink all the time. i think computers cannot blink.
Michael Josefowicz
5 Feb 09 at 4:39 pm
@Michael –
The concepts need to be manually defined and manually associated with the initial set of documents. Then, you use those manually defined relationships to teach (or tune, depending on the technology being used) the classifiers to associate this concept with these documents. Thereafter, the classifiers do the work without manual intervention, other than sampling for quality control and, where necessary, refining the classifier logic.
So, I agree with you that determining the proper set of concepts to use, as well as making the initial association between an article and a concept does require "Blink" and I agree that has to be done by a human. But once that's done, there's very little human intervention needed (if the classification process is well designed).
The classifiers will not only identify which articles seem to be associated, but will also identify just how strong that association seems to be. In many if not most cases, a given article will have some association with more than one concept. That's fine – the links subsequently developed will probably be ranked in order of the strength of the computed association. You can let your imagination run a bit to see what kinds of interesting news products this computer-assisted "context" could produce.
Lest you think this to be theoretical, let me assure you it's not. I personally developed a prototype and, for over three years of continuous operation, tested it using the articles published by a prominent newspaper as the target content. Checking these results just about every day, I determined that the resulting article assignments (using just under 200 discrete concept/topics) reflected between 98% and 99% precision (absence of false positives) and above 95% recall (absence of false negatives). From my market research of news consumers, that was well within the acceptable range. I found that the average classifier could run for weeks or even months before any tweaking would be indicated.
I don't want to consume too much of this blog's "real estate" on this specific point. You have my e-mail if you'd like more detail on how this works, which I'll be pleased to provide.
Terry Steichen
5 Feb 09 at 7:10 pm
@Michael –
The concepts need to be manually defined and manually associated with the initial set of documents. Then, you use those manually defined relationships to teach (or tune, depending on the technology being used) the classifiers to associate this concept with these documents. Thereafter, the classifiers do the work without manual intervention, other than sampling for quality control and, where necessary, refining the classifier logic.
So, I agree with you that determining the proper set of concepts to use, as well as making the initial association between an article and a concept does require "Blink" and I agree that has to be done by a human. But once that's done, there's very little human intervention needed (if the classification process is well designed).
The classifiers will not only identify which articles seem to be associated, but will also identify just how strong that association seems to be. In many if not most cases, a given article will have some association with more than one concept. That's fine – the links subsequently developed will probably be ranked in order of the strength of the computed association. You can let your imagination run a bit to see what kinds of interesting news products this computer-assisted "context" could produce.
Lest you think this to be theoretical, let me assure you it's not. I personally developed a prototype and, for over three years of continuous operation, tested it using the articles published by a prominent newspaper as the target content. Checking these results just about every day, I determined that the resulting article assignments (using just under 200 discrete concept/topics) reflected between 98% and 99% precision (absence of false positives) and above 95% recall (absence of false negatives). From my market research of news consumers, that was well within the acceptable range. I found that the average classifier could run for weeks or even months before any tweaking would be indicated.
I don't want to consume too much of this blog's "real estate" on this specific point. You have my e-mail if you'd like more detail on how this works, which I'll be pleased to provide.
Terry Steichen
5 Feb 09 at 7:10 pm
Fascinating. Given that there is virtually no cost to blog real estate, and I don't think we are sidetracking any other discussion, perhaps Matt, our host, could weigh in to see if it's ok to hang around and keep it right here. i would prefer it because there are lots of really smart lurkers. Terry would this be ok with you if Matt agrees? If not, send me an email "offline"
Matt?
Michael Josefowicz
5 Feb 09 at 7:26 pm
Terry,
Am I correct that the process of tagging can be ongoing as long as previous tags were not eliminated?
Michael Josefowicz
5 Feb 09 at 7:29 pm
@Michael –
By "tagging" I presume you mean the identification of new concepts and teaching/tuning the associated classifiers? If so, the answer is yes.
Over time, you'll probably also want to retire some classifiers, because interest in the concept/topic has waned. And you'll want to occasionally redefine a concept (expanding or narrowing its scope) and the associated classifier.
Please keep in mind that doing these things isn't possible with all classifiers, and much of the detailed information about the technology is quite proprietary. Publicly, the vendors will say "no problem." The reality, however, gets a lot more complicated. (At the same time, the technology is improving all the time, and I've been out of the loop for a couple of years, so there may be additional capabilities out there by now.)
Full disclosure: I am an independent inventor and have a couple of pending patents bearing directly or indirectly on this issue.
Terry Steichen
5 Feb 09 at 8:00 pm
make sense. My interest is to get closer to distribute and print. My other interest is to help figure out how good journalism can benefit from web tech so that more money can be focused on creating and supporting great journalists.
Michael Josefowicz
5 Feb 09 at 8:24 pm
I wanted to share a link I found this am. I think it is another step closer to using the Cloud in the way that Matt has been arguing for since I got here"
The New York Times and ProPublica are looking into doing something similar through DocumentCloud, which would be a place for reporters to store documents they gather during reporting for other newsrooms to use.
The link:http://www.collegemediainnovation.org/blog/2009/0...
FYI, I've been asked to do a bimonthly post as the "Print Correspondent" for http://www.pbs.org/ideala My first column is supposed to go live late in the day today. The FYI is that I put in links to newsless.org and proposed a scheme to raise some money for Everyblock. Keep an eye on your RSS.
Michael Josefowicz
9 Feb 09 at 2:10 pm
@Michael-
Document clouds are certainly interesting, and if they do what we hope they will, it will make reporters' jobs easier. Unfortunately, however, they don't really help us to solve the context challenge.
I think that most participants in this blog generally agree that news consumers want not just to be informed about events happening in the world around them, but also to better understand what that event means – in other words, they want context.
In the pursuit of developing ways to add context to news, it's easy to get pursuit of that goal confused with the related but different task of locating background material. DocumentCloud and other repositories seek to offer a richer set of more easily retrieved background material.
However, no matter how easy it might be to retrieve it, the background for the individual article still has to be written and that takes more effort on the part of journalists.
But even if the background is added, readers aren't going to want to wade through essentially the same background on every article covering a related event. What they want is background that they can access if and when they want it.
That means the background information must be independent from the news articles itself. It also means you need to organize the background information by categories that are meaningful, and you have to determine which categories are meaningful to each article.
And all of this brings you right back to square one.
Terry_Steichen
9 Feb 09 at 3:24 pm
Terry,
We agree. The art of journalism is similar to the art of writing history. How to nest an event within a series of events to make a plausible narrative. I thought the cloud was interesing as one more piece that is falling into place.
Michael Josefowicz
9 Feb 09 at 8:23 pm
[...] It’s a problem that can be solved. [...]
Old Forest, New Trees » Blog Archive » Online news should be replayable
27 Jul 09 at 2:00 am
[...] Yes. Basically my whole goal in creating a site like this is to bring some context to the issue of healthcare [...]
The Shocking Truth About Journalism, Activism, and the Healthcare Reform Debate « The Levisa Lazer
5 Sep 09 at 8:44 am
[...] A Prescription for Journalism Posted in Ethics, Future of Journalism, Journalism by James Rowe on September 10, 2009 The goal Matt Thompson takes for his fellowship at the Reynolds Journalism Institute at the University of Missouri is enormous. He began with Newsless.org. He explains in his own words in his inaugural post. [...]
A Prescription for Journalism « Journalism
10 Sep 09 at 3:23 pm