David Yanofsky, the things editor at financial news site Quartz, told Talking Biz News about how it puts coding into stories so that they automatically update.
He writes:
We write little snippets of code that download data from known locations, calculate some key values, then save the results to a file. When a reader comes to the page, their browser loads the file and inserts the numbers into the corresponding spots. As cool as it seems, and as useful as it can be, it’s not the most complicated or technically difficult thing to do.
Not all of these are examples of the process above, but here are a few other examples of things we published where content dynamically updates based on calculations or information we know about a reader.
- Keeping the figures in this tracker of S&P 500 companies’ political donation plans up to date.
- Using IP address geolocation to tell a story about a bank near the reader
- Using the system clock to give a reader the time of a global event in their timezone.
As far as news automation goes, it’s a pretty basic system. We’re not algorithmically writing full sentences, paragraphs, or stories, based on the data, like say Bloomberg, The Los Angeles Times, or Narrative Science do. We’re just replacing a digital “TK” with a number.
That simplicity is a big part of the allure. First, and most importantly, it’s easy to explain to our readers what’s going on. Second, there’s almost no cost for us to operate it. Amazon Web Services has humorously high free-usage limits.