Final Paper
The Journalist Machine
Ben Rogers
November 30, 2001

Journalism is a slave to news.  Without news there is no journalism.  Moreover, without means of gathering relevant, timely stories and retelling them to massive audiences, there is also no news—only hearsay, only conversation.  Journalists turn happenings into news.  “News isn’t a Supreme Court decision or a hurricane or a landing on the moon; it’s the newsman’s record that the decision, the disaster, or the ‘giant step’ took place” (Charnley, 1975, p. 6).  Technology makes news possible. Without technology, news reaches only as many people as can gather inside the range of a human voice. 

Journalism is also technology’s slave.

The tools of journalism have redefined and reinvented what it means to be a newsman or newswoman throughout history.  In many cases the tools supplant the journalist.  The Gettysburg Address was transcribed by journalists and later reprinted in newspapers for people to read.  Today, the same speech would be carried live on numerous national television stations.  People could listen to each intonation in Lincoln’s voice and watch each pensive pause.  And while the speech, if given today, would still be published in newspapers and magazines, printed copies of what Lincoln told a mourning nation would be no match for the immediacy and fidelity provided by a camera and microphone. 

The transcriber becomes a cameraman. 

Technological innovations force journalists to acquire new skills or risk becoming obsolete.  If he applied for a job in a newsroom today, Ernest Hemingway’s mastery of the simple sentence and his Pulitzer prize couldn’t fill the holes in his resume left by his ignorance of search engines, databases, word processors, fax machines and other modern journalistic ubiquities. 

In addition to reinventing journalism, technology can also provide mass audiences alternatives to journalism, and to journalists themselves.  Some specific examples of this: Cable television means New York Yankees fans living in Los Angeles need not rely upon sportswriters’ accounts of Derek Jeter’s game-winning homer; they can watch the real thing.  The internet means skiers don’t have to wait for a weatherman to regurgitate a National Weather Service forecast to see if snow is expected in the mountains;  they can check the Weather Service’s Web page on their own.  And cameras mounted in the back of city council chambers give handicapped veterans interested only in the proposed renovation of a local VFW hall the opportunity to vicariously sit in on that portion of the meeting instead of having every agenda item later summarized by a reporter.

This is not to say journalists are no longer necessary.  Far from it.  The question worth addressing is how journalists are necessary.  This is the goal of this essay.  Assuming journalists are handy with modern journalism tools, what can they as human beings contribute to the news?  The modern mass media audience has far more options than it ever did in terms of where to get news and far less time to digest it.  This audience has grown accustomed to being spoiled by specifics.  Yankee fans want only Yankees news;  skiers want only the updated ski report for a favorite mountain; and the vet doesn’t care about new parking meters at the library, but hangs on every word about the new VFW hall. 

Altruistic notions of what the public should know aside, technology lets picky, time-strapped audiences skip the middleman, the journalist, and go straight to the source for news and information relevant only to them.  They no longer have to wait until the evening news to be told what is important.  “Journalists cling to the conceit that we're at the center of the media universe. But the harsh reality is that, for many, the press is expendable. Increasingly, citizens are bombarded with news and information from all directions: morning news shows, talk radio, magazines, newsletters, tabloid TV. And now, the Net” (Lasica, 1996, p. 23).

The audience can boil it all down on their own.  Or can they?  In “What Are Journalists For?” Rosen writes:  “The point of having journalists around is not to produce attention, but to make our attention more productive” (1999, p. 295).  This would indicate that a journalist’s job is to filter information down to what the we need to know and then communicate it very quickly.  How does the journalist decide what we need to know?  Could a machine make these decisions?  Such questions are among those here addressed.

Technology is giving shape to journalism in the year 2001.  New machines are encroaching upon journalistic territories, including those as sacrosanct as writing and editing.  Technology is also giving non-journalists a chance to do journalists’ jobs.  Newsmen and newswomen must adapt the way they approach their profession in order to remain necessary and important. 

Transcribers need to become cameramen. 

Building the machine

If it were possible to build a journalist machine capable of all the same things as a  modern journalist, what would the machine do?  Just as engineers must first define the specific tasks a new product must perform, we too can use a journalist machine as a way of specifying exactly what journalists do, or need to start doing.  “Now is a good time to ask what journalists should be doing, in the pragmatic tradition for which Americans are well known.  A good pragmatist takes a look at the present state of things before coming to working principles; and so it is for journalism” (Rosen, p. 300).  Journalists need to change the focus of their efforts if certain tasks inherent to their jobs can be done quicker or better by a machine.

The dictionary defines journalism as “collection, writing, editing, and dissemination of news through the media” (Webster’s II New Riverside University Dictionary, 1988).  A journalist’s skill set, it can then be deduced, must first include the means of acquiring news, which in turn means the ability to seek news out, wherever it may be occurring (or hiding).  These conclusions may sound mundane, fodder for introductory journalism courses.  But only with such an elementary treatment of a craft such as journalism is it possible to grasp its core elements.  These elements must be made excruciatingly clear in order to design a machine that can perform tasks a person takes for granted.

So can a machine be taught to seek out news?  First it must know what news is.  News, according to the dictionary, is “recent events and happenings, especially those that are notable and unusual” (Webster’s II New Riverside University Dictionary).  Machines are certainly capable of pattern recognition and can detect a deviation in a series, for instance a man biting a dog after a long string of dog bites.  Computers coupled to cameras can detect visual pattern deviations as well, such as a car driving on a sidewalk.  A software program could be written to retrieve all of the daily police reports from a database and pick out a trend or anomaly, such as an unusually high number of rapes in a certain zip code or the first rape ever in another.  It could also crosscheck the same police reports with a list of notable people or compare the number of African-Americans  pulled over to the number of whites.

Journalists already use existing versions of this imagined software to analyze information in databases in search of newsworthy items on their own.  “Using a computer and a modem, reporters can now…do the following:  access vast amounts of information not otherwise readily available; check references in more dictionaries, encyclopedias, gazetteers, almanacs and glossaries than they ever believed existed; locate and contact sources and experts; search archives of news stories; store this information; speedily manipulate it; load information into a database; and search and analyze the database” (Randall, 2000, p. 86).  The evolution of the technology journalists use to do all these things will be programs that can not only pore over data, but know what to look for on their own. 

In The Principles of Journalism, Casper Yost writes: “There is always before the editor not only the question, what is news, which he answers instinctively, without any need for precise definition…but also the question, what is news from the standpoint of journalism…and to answer this requires the exercise of judgment as well as of instinct” (1924, p. 23).  (Worth noting is that in the margin beside this statement it is scrawled “wrong use of instinct.”  The handwriting most probably belongs to longtime journalism professor A.J. Higginbatham, who owned the book in 1924 before it was given over to the University of Nevada library.)  If identifying news indeed requires instinct, that doesn’t necessarily rule computers out.  After all, what is instinct but a pre-programmed behavior, an innate manner of reacting to circumstances?  By such a definition, computers operate solely on instinct, so long as all possible circumstances the computer might encounter have been identified.  Feed a computer a string of parameters by which to judge newsworthiness, even assign relative weights to these parameters, and it could whittle a day’s worth of happenings down to a few newsworthy stories based upon its own programmed instincts. 

Moreover, stories created by a machine might be less prone to the simple mistakes and lapses of objectivity humans can’t avoid.  “Reporters are…flawed.  As a group, they have a more soiled reputation than most; for enough of them routinely exaggerate, simplify and contort the truth to have made parts of the trade a by-word for calculated dishonesty” (Randall, p. 1).  Randall also makes note of limits in the process of journalism, including “the errors that journalists make when working under pressure”  (p. 14), and goes on to say that the rampant innumeracy of journalists is a “fatal weakness” (p. 72).  Machines on the other hand know the difference between “affect” and “effect” every time, are designed to handle numbers without error, and are immune to pressure.   

None of these observations are novel; machines long ago proved to be on a par with people when it came to anything formulaic.  Factory robots now assemble cars once pieced together by hand, vending machines sell snacks once peddled by people, and word processors check our grammar and spelling.  Yet machines have outshined their warm-blooded creators in more cerebral challenges as well.  In May 1997, IBM’s Deep Blue computer defeated the world’s preeminent human chess player in a six-game match.  “It was a sensational story that rippled worldwide, cast as the frightening symbolic denouement that humankind had been dreading ever since our bucolic rural existence began ceding ground to the dark, infernal mills of the Industrial Revolution:  the machine dominating man in the one area where he was unique and unchallenged—his intelligence” (Chelminski, 2001, p. 100).  If machines can think, they must certainly be capable of handling a few names, dates and quotes.  After all, “as fact assemblers and carriers, reporters are in one sense passive agents” (Charnley, p. iv). 

Still, there is only so much we can expect of a machine.  Scanning databases in search of patterns and playing games with finite possible outcomes is one thing.  Interviewing city council members, streetwalkers, and rock stars is another entirely.  In order for a machine to locate, meet with and talk to real people and do it as well as Barbara Walters, it would have to be many things today’s machines are not, including personable.  While a journalist machine might outpace humans in many journalistic newsgathering duties—including information recovery, data manipulation, and anomaly recognition—it fails miserably at others.  “To cover the news is to judge what is newsworthy; such estimates of worth involve ideas about politics and power, images of democracy, views of culture and the individual, all of which are subject to change as common conditions remake themselves.  In these ways and others, journalism requires imagination: some way of conjuring with the world so as to station yourself within it and draw news from it” (Rosen, p. 283). 

For now, imagination exists solely in the minds and souls of the warm blooded.  Machines can be programmed to identify and gather news, but they are capable of neither creativity nor inspiration in their search.  “A news story is more than the sum of the facts that compose it: it is the sum of the facts plus the form a reporter gives it” (Charnley, p. 27).

Imagination seems to separate man from machine in all disciplines where the two overlap.  This will be discussed later.

Returning to the definition of journalism, news collection must be followed by news writing and news editing.  This essay will focus primarily on writing, as opposed to other means of presenting news such as radio and film.  (Both radio and film usually require stories to be written down before they are broadcast anyway.)

Samuel Johnson believed that “to write news in its perfection requires such a combination of qualities, that a man completely fitted for the task is not always to be found” (Brown, 1957, p. 17).  Perhaps a machine is up to the task?

A pair of computer scientists at North Carolina State University in Raleigh are already providing preliminary answers to this question.  Charles Callaway and James Lester have created Author, an intelligent system capable of writing convincing fairy tales.  The software was developed to help children conquer literacy problems, but could soon be applied to journalism.

Here’s how it works:  Author is supplied with a story plan that includes characters, scenes, props and an order of events.  Using a plot outline to arrange the events chronologically, Author strings facts together into sentence-like chunks, then applies language rules to create grammatically correct sentences.  “The software can already tell a mean version of ‘Little Red Riding Hood.’  But given a different knowledge base, says Calloway, it could just as easily write newspaper stories, short stories and movie scripts” (Graham-Rowe, 2001, p. 20).  Author belongs to a family of systems known as automated summarizers, which could one day be coupled to data mining algorithms that extract information from text.  “This would make it possible to dispatch automated roving hounds to scan news wires or government papers for the bare bones of a story” (p. 20).  This appears to be the aim of Artificial Intelligence (AI) efforts already underway.  “Right now, all efforts point to the Web, where programmers hope to unleash bots that can troll for information, learning the ticks of language and meaning along the way”  (Menduno, 2001, p. 82). 

All of this is logical enough.  After all, roaming the Internet for specific information is getting easier and easier…for humans.  Yet, in order for computers to move freely on their own through the vast and eclectic galaxy of information that is the World Wide Web, the Web must first be organized differently.  It must be organized by computers for computers, rather than by computers for people.  The latest effort to do so is called the “Semantic Web” and its proponents include one of the Web’s founders, Tim Berners-Lee. (McDonald, 2001, para. 1).  “Today’s Web is basically a ‘publishing medium,’ a huge warehouse where text and images are stored.  The Semantic Web wants to turn it into a more interactive place, where information can be interpreted and exchanged and where software agents roam from page to page, performing sophisticated tasks for users.  Instead of merely displaying information on their screens, computers will ‘understand’ what they are displaying” (McDonald, para. 4).  Still, the reorganization of the Internet could prove a more massive and daunting task than did creating the Web in the first place.

Journalists, fear not.  Not yet.

In terms of the journalist machine, we still have a long way to progress before reporter-bots are writing their own stories.  For starters, software programs such as Author are still unable to distinguish fact from fiction.  What Hemingway referred to as his “built-in bullshit detector” has always been a tenet of journalism and is not yet found in any algorithm.  When it comes to writing the news, reporters are probably better augmented by a journalism machine than replaced by it.  This sentiment is echoed by Douglas Lenat, AI pioneer and former Stanford professor.  Lenat believes that machines should work for people, not as people:  “Anthropomorphizing a computer program isn’t a useful goal” (Menduno, p. 82).  As founder and president of Cycorp, Lenat is working to create a machine endowed with computerized common sense.

While fresh, natural-sounding writing may prove difficult for machines, the more formulaic task of editing might not.  Word processors already check grammar and spelling automatically and could no doubt evolve more intelligence.  Why couldn’t computers check stories for the nitpicky items?  Checking state capitals, crosschecking ages, helping with word choice, correcting dates, automatically generating relevant art or graphics, suggesting Web pages where further information can be found:  these are all within the realm of possibility for computerized editors and could prove a time-saving, preliminary filter for human copy desk editors. 

Computers have already become quite adept at picking stories off wires and matching them to meet specific audiences, especially when the sheer volume of stories makes human editing impractical.  Database developer NewsRx (NewsRx Press Release, 2001) uses proprietary information-management systems known as Component Aggregation Technology (CAT) and Journalist Assisted Technology (JAT).  Very little information is available about how either system operates; NewsRx says only that “CAT and JAT assist human journalists in performing vast publishing and news reporting tasks” (NewsRx Press Release, para. 5)  With a news database containing over 100,000 health and science articles, NewsRx generates weekly reports on 25 different medical and bioscience topics.  It also produces custom reports daily, weekly and quarterly on myriad niche health and science topics.  Alan Henderson, NewsRx director of business development and technology, believes that preparing such a quantity of material without the aid of artificial intelligence “would be comparable to space exploration minus the rocket.”  He adds, “the future of information distribution clearly depends on advancing technology.  What we have here may be the Gutenberg Press of the twenty-first century” (para. 6).

Another related program is the Columbia Newsblaster.  Developed by the Columbia University Computer Science Department, Newsblaster is an online news summarization tool still in its experimental phase. (John V. Pavlik, Executive Director of The Center for New Media, Graduate School of Journalism, Columbia University, personal communication, Nov. 13, 2001) Using “natural language processing” the program automatically tracks news events and groups stories from multiple sources dealing with one event.  It then writes daily summaries of the stories based on a computerized analysis of text reports from 12 online news sources, including ABC News, CBS News, CNN, FOX News, Reuters, the New York Post, the Washington Post, and USA TODAY. 

How do technological developments like NewsRx and Newsblaster affect journalism?  By making journalists work harder for original content.  For years, media outlets have used wires such as the Associated Press to generate stories.  If a reporter snatches an Associated Press story off of the wire, tweaks the lede a bit and adds a few localized quotes, that reporter can then slap his or her byline at the top of the touched up story, even though 90 percent of it was taken directly from AP. 

A combination of Author and Newsblaster can now do almost exactly the same thing. 

Similarly, many newspapers fill their pages with stories picked from the various newswires based on the stories’ appeal to the newspapers’ regional audiences.  NewsRx, which boils thousands of stories down to a few that appeal to a specified niche, can already do this as well.  Journalists who make a habit of rewriting news stories or whose job it is to pick stories off newswires might not want to make a career out of doing so.  Technology is already nipping at journalism’s heels in these areas.  However, as already discussed, those journalists who can dig up and reveal fresh news, who can find their own information and then relate it in an inspired format—these are the people who can still outperform any algorithm.  News must be new—hence its name.  Journalists would do well not to forget that.

Machine-enabled amateurs

“Why don’t the newscasters cry when they read about people who die?

At least they could be decent enough to put just a tear in their eye.”

--From “The News,” by Jack Johnson

Machines are blurring the line that once separated journalists from their audience.  In certain situations technology gives amateur journalists a chance to upstage the pros.  In doing so modern media technology serves as an equalizer, an enabler.  Instead of looking to the evening news, people might now turn to an online discussion forum for breaking news coverage of catastrophic events.  Such was the case during the 1994 earthquake in Northridge, Calif., the arrest of the Unabomber and the Oklahoma City bombing. 

This phenomenon is also evidenced in recent coverage of the Sept. 11^th, 2001 terrorist attacks on the World Trade Center.  Armed with cameras, keyboards, and modems, the everyday people unlucky enough to be caught in the shadow of the tumbling towers proved proficient newsmen and newswomen.  Technology, this time, empowered non-journalists.  This is much like the example of the Yankee fan who forgoes the sportswriter’s report to watch the real thing, only in the opposite direction: this time, witnesses skipped the middleman, the journalist, and gave a mass audience their stories directly.

In many cases the result was unfiltered information, bursting with emotion, rife with rumor, and subject to subjectivity—all things journalism students are taught not to fall victim to.  It was just what an American audience desperate for information needed in the hours after the first plane struck.  Moreover, it was exactly what the mainstream media, busy regurgitating official announcements, failed to provide.  Technology made it possible.

In “Amateur Newsies Top the Pros,” (2001) reporter Leander Kahney recounts the outpouring of outstanding news coverage from ordinary people.  “The terrorist strikes prompted an avalanche of personal accounts, photographs and videos of the terrible events from people on the ground.  And not only was the citizen-produced coverage sometimes more accessible than professional news organizations, it was often more compelling” (para. 6).  The article compares the many powerful photographs emerging from the tragedy, noting that “some of the most intimate and unexpected shots came from outside the community of professionals” (para. 9).

Haunting images weren’t the amateurs’ sole contribution to the coverage.  Written accounts, too, were intimate and informative.  Read one such account, written by Usman Farman, a Muslim man (para. 14):

I was on my back, facing this massive cloud that was approaching, it must have been 600 feet off, everything was already dark.  I normally wear a pendant around my neck, inscribed with an Arabic prayer for safety; similar to the cross.  A Hasidic Jewish man came up to me and held the pendant in his hand, and looked at it.  He read the Arabic out loud for a second.

What he said next, I will never forget.  With a deep Brooklyn accent he said, ‘Brother, if you don’t mind, there is a cloud of glass coming at us, grab my hand, let’s get the hell out of here.’  He helped me stand up, and we ran for what seemed like forever without looking back.

“And while the New York Times, the old, gray lady, was full of bloodthirsty war-mongering, ordinary citizens gathered online to present a wide range of opinions.  Depending on where you looked, there were measured calls for restraint, investigation and thought before action.  Of course, there was also plenty of bloodlust, chest-thumping and stupidity, but online there was at least a debate” (Kahney, para. 18).