Computer-Generated Game Recaps: Future?

In my first job out of college, working for what may have been the first online start-up dedicated entirely to sports, we wrote homemade game recaps.

For every college basketball game of the night, my co-writer and I would write 75-150 words on what happened -- using the box score, the AP game recap and any highlights on ESPN.

It was the 1996 equivalent of Ronald Reagan calling a baseball game from the news ticker, pretending it's live.

Of course, we wanted to make it different than a typical newspaper game story. We also were writing for displaced fans using AOL to follow their teams -- or other college hoops games.

They were quick-hit and clever, by necessity: The most germane stat or play. Putting the win into context of how the rest of the season is going. Just having fun with it. (We also entered all the game stats into a home-made database... by hand. Yikes.)

It was an amazing training for doing the Quickie or blogging or simply writing with some level of optimal efficiency.

It also foreshadowed a future where game-recap information was entirely commoditized, that it was all about adding value to what most fans already knew about the score or game itself.

Here's a fascinating post by David Carr examining the product of Northwestern's project to get a computer database to write a competent game story.

The upshot: The computers' story is pretty good. By far the most impressive thing is the way the computer cites OPS as a relevant statistic (in the 4th graf of the story, no less).

Less good, as Carr points out, is that it neglected to put the pieces together and talk about how the game represented the series-clinching win.

But that's what an editor is there for.

What I'm trying to figure out is the practical applications: Local papers are always going to want to add value to readers by having their beat reporter interpret the game result.

(Sports game recaps had already gone away from dry play-by-play -- which most fans either got from watching the game themselves or checking the AP recap online -- and moving to more analysis.)

And non-local papers (or national Web sites) already have a perfectly low-cost (and human-created) game-recap system, as provided by the AP or other syndicator.

(Alternatively, you could pay a kid $18,000 a year -- like I was paid back in 1996 -- to gleefully write original homemade recap after recap, night after night.)

So where's the market? It probably doesn't exist.

But as an intellectual exercise of wondering whether a computer could write a passable game story, the answer is: Yes.

(Oh, and a reminder that beat reporters who don't have a minimal facility with the latest conventional wisdom in quantitative sport analysis are doing their readers a disservice.)

I ask this every time I post about that first job: Does ANYONE out there remember using Real Fans Sports Network on AOL, back in '96/'97? Anyone?

-- D.S.