Could Algorithms Replace Writers?

Written by and published

Writers like me rejoiced when Google launched its Panda update. Content mills, whose low-quality articles hogged the top spots in the SERPs for certain keyword searches, created an environment where writing didn’t get the respect it deserves, both from inexperienced and indifferent writers looking to make a quick buck, and from publishers who want to spend as little as possible for content.

The way writers make their living was devalued by posers and cheapskates. Why would a company pay a premium for high-quality content when the good-enough content was obviously ranking? Panda changed that, and conscientious, skilled writers who take pride in their work and command commensurate fees breathed a sigh of relief.

Now Narrative Science comes along, and writers are once again wondering about their place in the world.

Meet the content robots

If you’ve missed any of the recent buzz surrounding the startup, Narrative Science has developed an algorithm that can mimic human writing. Through a collection of data, statistics, and other components, the algorithm is able to produce a fleshed-out piece of content in narrative form.

“Psh!” you say. “As if a computer algorithm could sound anything like an actual person!”

Yeah, that was my first thought, too — until I read something produced by Narrative Science:

The majority of analysts (75%) rate Yum as a buy. This compares favorably to the analyst ratings of its nearest 10 competitors, which average 45.7% buys. That rating hasn’t budged in three months as the average analyst rating of the stock has remained constant.

Okay, so that’s not exactly Shakespeare, but it gets the point across, right? The mildly disturbing part is that it’s an excerpt from an article on, which now lists Narrative Science as a contributor to its blogs.

An algorithm with a byline on a respected news site — many human writers never reach that point.

Because Narrative Science compiles text based on structured data and statistics, the industry feeling the most territorial encroachment is journalism — specifically, sports.

In fact, two of the people on the company’s leadership team also worked on Stats Monkey, a similar algorithm aimed at sports information and reporting. Another sports reporting program, GameChanger, actually uses Narrative Science to produce its game recaps.

Tools still need human input

By itself, Narrative Science is not a sufficient replacement for a human writer. Then again, there is that Forbes byline. Still, the program relies on humans to enter data and certain parameters before it can produce content. But after that interaction, the algorithm takes over, and much of the end result is automated.

In this respect, it’s similar to other algorithm-based tools. For example, users might liken Raven software to a digital marketing Swiss Army knife because of its broad applications and numerous functions. Yet it still requires some level of human interaction in order to do what it’s supposed to do.

But compare Google in its infancy to what it is today. The search engine has gone through thousands of updates and is now one of the most powerful algorithmic tools in the world. It’s also one of the most apt examples of automation ever. Can you imagine having to sift through the Internet manually to find information? It may still require us to enter data, but it’s getting better and better at anticipating what we want, sometimes before even we know what we’re looking for.

What about passion?

Which bring us to the question: Can Narrative Science really replace writers? And if so, can other algorithmic tools replace humans?

As a writer, I have to say no. An algorithm may be able to put data into a logical, pleasing order, but it cannot unfailingly imply context or convey emotion. I have a hard time imagining algorithm-generated content making the reader cry or laugh.

But other than reaching the reader on an emotional level, what about the search engines? Rand Fishkin wrote a post in which he shared a theory about Google using authenticity and passion as ranking signals.

Some of what he said in that post has been borne out in the Penguin update which, as we know, addressed content overoptimization, which is another way of saying Google likes authenticity. The jury’s still out on the passion part.

So while a tool like the one created by Narrative Science can convey ideas — and maybe even that authenticity — it requires quite a stretch of the imagination to see it expressing passion. To me, it’s akin to machine translation — it may have the mechanics down, but context is often a difficult concept to grasp, even for humans.

More publishers catching on

But then the science nerd — and science fiction aficionado — in me has to wonder about the implications as this type of algorithm continues to be developed and updated, and whether it may eventually evolve into the very thing that is now meant to use it in its current iteration. And yes, I said evolve. Come on, as if thoughts of Skynet hadn’t already crossed your mind.

Barely two years old, Narrative Science is already on its way to bigger and better things. The startup just raised $3 million in funding in less than a month, and more publishers have begun using it.

While large organizations like Forbes may find it useful, I have a hard time imagining it catching on with individual entrepreneurs, marketers, and bloggers. At least, not yet.

What do you think? Are tools like Narrative Science just that — tools to be used rather than to replace? Are writers in danger of losing their livelihoods? Or are people getting worked up over nothing?