What’s news? At a basic level, it’s information about something that’s different or surprising. So if the Pope is Catholic, that’s not news. But if he was Jewish…
Now that’s not a comprehensive definition by any means. But it is one way to look at how information is valued by people, and how we might be able to structure the way we cover/produce news to offer that value. So by this measure, at least, readers want to be informed when something comes up that’s different from what we expect, or when material things change. Say a company’s earnings are way out of whack with the consensus forecast; or when the school board votes in favor of – or against – a key resolution; or when a politician changes his long-standing position on a key issue.
A very smart colleague of mine on the business side at Thomson Reuters, Christian Ward, introduced me to a nice phrase the other day that perfectly describes this sort of news: Delta alert. It’s another way of saying that we want to be alerted when there’s an unexpected change (delta) in how we see the world.
How hard is this to do? How can we possibly we track all the possible changes that people expect, and build a system to set off alerts when things are different?
In the business journalism realm, it already happens fairly frequently. It’s not hard to maintain a database of consensus corporate earnings forecasts, for example, then to compare the actual results to them and fire off an alert if they’re materially different. In fact, you may not need human intervention to make that happen, assuming you have a way of getting the actual results into a machine-readable form. Similarly, when analysts change their ratings on companies from, say “buy” to “hold” or “sell,” there are programs out there that can take the salient points of that recommendation and turn it quickly into a story – “Research House X upgrades company A to buy from hold,” for example. No humans needed there either.
Part of the reason it’s – relatively – easy in those areas is that a fair amount of business information is structured, and hence not that hard to compare. Analysts have a finite number of types of recommendations; most companies report earnings in broadly similar formats; and so on.
It’s trickier for other kinds of coverage. But we can help it along if we adopted some of the ideas embedded in structured journalism – because if information was more structured, we’re much more able to compare it with other similar information. So a site like Politifact, for example, can in theory set up alerts each time they find a specific politician lying; or they could track when a politician with a pristine record on a particular topic suddenly pulls a fast one; and so on. Education reporters can build databases of school scores, let’s say, and have alerts sent out when new scores differ significantly from previous results. There doesn’t even need to be a huge amount of effort on the part of journalists, as long as the new information can be input into a structured database so it can be compared against other data.
It can’t work for everything, of course. But we can try to extract more value out of the kinds of things we’re already doing, and give readers more of what they want – or need – to know.