Thursday, July 12, 2012

You Will Be Replaced


The words you are reading follow an ordered structure. In fact, the language itself follows an embedded hierarchy of ordered structures. You know who they are. It’s all the usual suspects: paragraphs composed of sentences, sentences composed of clauses, clauses composed of words. At their most atomic level, the words themselves are formed by characters.

If you delve further into this meta-reality, you will find that these characters have numerical values assigned by the Lords of Convention; within our shores, we have the American National Standards Institute, and beyond, the International Standards Organization. These ANSI and ISO defined values are stored as electronic impulses, to be retrieved and formatted by a vast array of programmable devices.

It’s no surprise that they’ve managed to program beyond the ETL phase. Transcending the base operations of Extract-Transform-Load, they’re programming the structure itself now. That’s right. There are companies which create software to generate writing. These aren’t pipe dreamers. These are serious people, engineering serious machines, and they have serious funding.

Maybe the most well-known of these would be Narrative Science in Chicago. It’s operated by several professors from places like Northwestern and Yale, with advanced degrees in subjects like Journalism and Computer Sciences. They have multiple backers, customer cashflow, and most importantly, a compelling dream.

What they do is this: their computers fit data into a template, apply a few angles based on rules, and churn out a sequence which is appropriate to the client. For instance, the play-by-play stats coming out of a game could be plugged into a template of winners and losers, rules applied to determine if it was a “rout”, and a story fashioned to report on those events. Besides sports, many other venues are ripe with data – corporate reports, financial analyses, and educational books, to name a few. Considering we live in the dawn of Big Data, the sky’s the limit.

An endeavor like this raises questions. For example, how good is the writing? If you wish, you can run a simple Google search, and judge for yourself the quality of their prose. To me the central question, for which there is no readily available Google answer, is this: Can you be replaced?

Of course you can. That’s why you need to deal with this and take control. Computer aided writing is no different than computer aided design. Computer aided anything is a good thing. The bottom line is that it’s going to happen. The question beyond the question, in turn, is this: How will you exploit new technology to augment your own writing experience?

This is simply the next step beyond a spell checker or an online thesaurus. You can write, “See Jane run,” or you can write, “Observe with your born-anew eyes the passing form of Jane.” They both say the same thing. A software program following basic formulae of grammar, and backed by a dictionary database could spew out hundreds, if not thousands of such variations, one of which would certainly be the above. Writing Is Rewriting. If the measure of quality is found in the quantity of expression, you won’t be able to beat a computer.

And this is just the beginning. If you think the program can’t improve itself on the fly with machine learning, then you, my friend, are sorely mistaken. All it takes is a clear-headed programmer. In this brave new world, nothing more is required. 

No comments:

Post a Comment