In these difficult economic times, it shouldn't surprise anyone that employers are looking for ways to save money.
These days, most employers show little or no interest in training people to do a job that needs to be done. They want someone with hands–on experience. If the new hire is a little rough around the edges, well, they figure they can soften those edges — and, in the long run, they'll be better off because they hired someone who could step in and be productive from the first day. They didn't have to devote time and money to training.
In this quest for economic efficiency, advanced technology seems to have renewed a desire in business managers to make one–time investments in machines (or software) that can do jobs people have been doing for a long time.
From a purely business, dollars–and–cents perspective, I suppose that is logical. Machines (or software) don't need to be paid. They don't have bills to pay and families to support. They don't mind working overtime or on weekends.
That can work pretty well in some endeavors — but not in others. And my conclusion, based on years of working in the newspaper business, is that it's the kind of solution that tends to appeal to bottom–liners.
For a long time, the way that publications responded to recessions was to cut back on copy editors. Reporters could do their own fact checking, they reasoned. They could check their own spelling and grammar. Well, they
could do those things — but most of the time they don't.
And, the reasoning continued, they could perform multiple tasks when on assignment — like taking pictures or shooting videos. They might not have much experience at those things, but just give 'em a camera.
Anyone can aim and click, right?
Heck, for that matter,
anyone can write. All it amounts to is stringing some words together in sentences — and making sure those annoying little things like spelling words correctly and checking facts are done.
I presume that is the reasoning behind some new software from Narrative Science. Justin Bachman of
BusinessWeek reports that Narrative Science's software can take the box score from a game and produce an article that is every bit as good as anything that was written by humans.
To prove the point, Bachman takes articles provided by the sports information departments of two schools that played each other in baseball on April 24 and compares them to an article that was
"composed" by the software — and asks the readers which one was generated by a machine.
Then Bachman triumphantly identifies the article that was produced by the software and quotes the software company's CEO:
"There's no human author and no human editing. But the stories sound really good."Now, granted, taking the Joe Friday —
"Just the facts, ma'am" — approach works sometimes — and, to the eyes and ears of someone who never studied journalism or worked for a newspaper or a magazine or any other kind of newsgathering organization, the stories may seem adequate. They may even
"sound really good" — although Narrative Science acknowledges it is always looking for ways to make the software system
"a less bad writer.""The 1,000th story of a subject is materially better than the first," the company's CEO says. (Perhaps the software's writers should take News Reporting I.)
And, according to the Big Ten's director of new media, it's
"less expensive" than hiring beat reporters. So it meets the goal of saving some money.
Nevertheless, in spite of all these benefits, I don't think machines will take the place of general assignment reporters.
Oh, sure, it has its place. I used to work on the sports copy desk of a community newspaper. Space was limited, but the paper had made a commitment, long before I started working there, to print brief articles on Little League and T–ball games. This was something that was mostly done for the parents, who would presumably boost circulation a bit on the days after games were played.
On weekends, there might be a couple of dozen such games, and
someone had to go through the statistics of each game and highlight the most crucial ones in a paragraph or two — which also had to include the teams' records and when they were scheduled to play again.
If we had had such software in those days, I suppose we could have fed the information from the games to the computer. It would follow its fill–in–the–blanks article–writing procedure and generate stories on all the games in a fraction of the time a flesh–and–blood person would require.
But what would have been just fine for children's sports would not have been sufficient for the rest of the newspaper's sports coverage.
There was a large university in town, and a staff writer was assigned to cover the sports teams there. When the football team or the basketball team went on the road, the writer went along. Sometimes we had to wait for the writer to finish his story and call the office to dictate it (this was in the days before a writer could transmit a story electronically from a remote location).
But, compared to the
"data–intensive" articles that Narrative Science is capable of providing, the articles that were generated by the human tended to be superior. I wouldn't say our writers were even close to perfect, but they could interview players and coaches. They could explore angles that computer software could never come up with on its own.
And their writing skills usually were good enough to give each story a unique flavor — whereas, I suspect, Narrative Science's articles tend to seem formulaic rather quickly.
Bachman concludes that Narrative Science
"can make some writing by humans obsolete."That's probably true of
routine writing assignments.
But if you want quotes (and surveys show that most readers do) about big plays or upcoming opponents — or if you want articles that call for dogged reporting on things that coaches and athletic departments may be reluctant to discuss in much detail — the severity of a player's injury, for example, or whether a star athlete is going to face criminal charges for something — computer software won't get the job done for you.
"After tackling sports," Bachman writes, Narrative Science
"will move on to medical, financial, and survey data."Good luck getting that software to explain complex medical and financial subjects in reader–friendly terms.
