Will Narrative Science’s Algorithms Replace Editors of Professional Content?

John Barker
Written by John Barker
on January 22, 2014

I’ve posted in the past about algorithms assisting professional editors in summarizing, topically classifying and applying metadata to tax, legal and regulatory content. That was focused on algorithms as assistants.

A September 10, 2013 post in the Wall Street Journal’s Venture Capital Dispatch blog states that algorithms are already doing much more than merely providing assistance. The blog post talks about how a company called Narrative Science is able to automatically generate news stories about basketball games. But Narrative Science’s Quill product can write about much more than just sports. It can write – without human assistance – financial and real estate news stories. However, the blog post notes that Narrative Science’s Quill product was initially trained by journalists. Apparently from the moment you purchase Quill, you must fine-tune it for approximately 90 days before putting it into production. While Quill can truly write stories, it is being positioned more as a decision-support tool that helps companies make more sense of their data. In fact, you can read some stories authored by Quill at Forbes.

Naturally, I wonder how Quill might be used to help editors of professional content generate more meaningful insight. Can Quill summarize the holding of a case or extract the key conclusions from a clinical trial? Can Quill inform a lawyer when a regulatory action has de facto carved out an exception to the application of a statute? Can Quill keep annotations of cases, laws, regulations and administrative rulings updated? Perhaps publishers could use Narrative Science to cover a broader range of content and enable editors to focus on other tasks.

Law firms, accounting firms and hospitals increasingly act as publishers themselves. Law firms and accounting firms produce newsletters for their clients. Patients of doctors increasingly expect consumer-friendly explanations of illnesses and associated treatment regimes. Professional services firms, in order to not “reinvent the wheel” when giving advice, invest in knowledge management technologies. If I understand Narrative Science’s Quill product correctly, it might be an interesting tool for these purposes.

I certainly do not think that Quill replaces editors of professional content. But I am convinced that it is worth testing on professional content. Perhaps it could become a sort of editorial assistant enabling publishers and their professional customers to generate new insights at a scale never before economically possible. In theory, time that used to be spent on writing summaries could be invested in writing software that increases professionals’ productivity. Perhaps Quill could generate insights from blogs and social media as well as free public domain content that can be meaningfully combined with content produced by professional publishers as well as professional services firms and hospitals.

Have any of you had experience with Narrative Science or Quill?



There have been made comments on this article

  1. Kris Hammond on at

    John – thanks for your interest in Narrative Science and the work we are doing. I thought I’d offer some explanation regarding the questions posed in your post.

    The core capability of Quill, Narrative Science’s authoring platform, is the generation of language from data. The central idea behind Quill is that the data we now have in our hands (or in the “hands” of the computer) not only describes our world but also can be mined for insights as to what is happening, and how we might respond to it.

    The important point here is that Quill starts with data. It needs to have access to the important data points in order to identify and report on the interesting insights that they can provide.

    As to whether Quill can act as an editor (crafting summaries of larger documents, and creating new documents out of existing ones), it can’t, but this is not because the task is too hard or even forever in the realm of human expertise. Rather, it’s just not what Quill was designed to do. Quill discovers meaning and insight in numbers and symbols. This requires that it understands the world and is able to recognize the patterns that matter to the world as it reviews the data. It sees meaning in numbers, because people have a very hard time doing this on their own.

    The editorial task that you describe requires seeing meaning in the same way that people already do. That is, read pieces of text, pull out the primary points and pull these together into summaries. It requires reading, a skill that Quill simply doesn’t have.

    Much like IBM’s Watson that cannot understand speech but can think and reason at blinding speeds, Quill cannot read, although it can find genuine meaning in masses of data that would drown most people. It is the automation of what it means to be a data scientist. It can deal with numbers, figure out what they mean and then communicate that meaning to us. It is smart and in being smart, makes us smarter.

    Hopefully this helps clarify for you and your readers!

  2. Saskia on at

    There are already plenty auto summary tools. If anything, big data means more rather than less editors. I do not understand this hostility towards editors. They’ll always be needed. Much unlike publishers.

  3. John Barker on at

    Saskia, I view automated capabilities as helpful for editors, not hostile. For example, Microsoft Word makes authors more productive, even though it automated some authoring tasks. Automation makes it possible for editors to contribute more value. Analytics can help editors and authors provide insights over large amounts of content that otherwise might not be possible.

  4. John Barker on at

    Kris, thanks very much for the clarifications about Narrative Science. I hope that Narrative Science will continue to blog/post about how its capabilities are moving up the editorial value chain. It is impressive technology.

Exploring trends, content, technology, and new ideas in the global information industry. New posts every Monday, Friday, and whenever the innovation bug inspires us. Visit www.wolterskluwer.com to learn all about us.
Recent comments