Robot Journalist Writes A Better Story On College Baseball Perfect Game

Journalism

from the who-what-when-where-why dept

A few weeks back, there was a fun story about how the website GWsports.com, which (as you might imagine) covers sports at George Washington University, became something of the laughing stock of the press world by publishing a game recap on a game between GW and the University of Virginia. The game recap reads pretty straightforward, opening with the fact that the GW team lost, and then spending a few paragraphs covering the various efforts by different GW players. It’s only in the seventh paragraph, out of a grand total of eight, that it finally gets around to mentioning that the opposing pitcher, Will Roberts, threw a perfect game. If you’re not a baseball fan, a “perfect game” is extremely rare. As the article does (finally) note, this is only the eighth such perfect game in NCAA Division 1 history (since 1957) and the first since 2002. Normally, you would think that any press coverage would start with the perfect game bit.

The national press, including the Washington Post, picked up on this and GW folks tried to defend the writing by noting that they’re only in the business of promoting their own teams, not others:

?This is the George Washington website,? GWU sports information director Dave Lubeski tells Romenesko. ?We?re in the business to promote our athletes and our team. We?re not claiming to be journalists.? What some call ?the buried lead? was discussed after the story was posted, says Lubeski, and it was mentioned that the perfect game could have been noted in the sub-hed. But ?we?re not in the newspaper business,? notes the SID.

I had sent that story around to a few people, because I thought it was pretty funny, but didn’t have any obvious reason to post it here… until now. Apparently, in covering the story, Deadspin wondered if perhaps the original article hadn’t actually been written by a newfangled automated software program that’s been touted as being good enough to write sports stories.

Well, it turns out that the people who wrote that software, Narrative Science, were offended that their program might be thought of as having written an article so badly written, and fed their program the data from the game, and it popped out much better versions. They actually did two versions, a “neutral” point-of-view one, and one designed for GW fans. The “neutral” one brings up the perfect game in the first paragraph. The GW POV one does wait until the third paragraph, but still seems much better than the original…

So, perhaps the author of the original has passed the reverse Turing test, when you can’t tell if a human is actually a human or a robot…

Filed Under: baseball, george washington university, journalism, perfect game, reporting