Adventures in Anonymous Sourcing

Ryan Chittum picks up on a dog not barking in today's WSJ story about Goldman earnings:

The paper reported that Goldman Sachs's loss this quarter would be much worse than expected, news it attributed to "industry insiders."
That's funny attribution, but okay. But scan the rest of the story and you'll find that it appears nobody from Goldman was ever given an opportunity to comment.
Now, it's highly unlikely that these experienced reporters got a story on A1 in the WSJ without calling the company for comment.
What the lack of a Goldman attribution signals to us is that Goldman Sachs itself leaked this to the Journal as a way to feed hungry beat reporters and get bad news into its stock price before it reports earnings.
The Journal has a nearly iron-clad internal rule that says a story can't say a source declined to comment if that source is quoted elsewhere in the story.

What else indicates that the source is within Goldman? The story's placement on the front page of the paper: there's no way that the WSJ would give this story such prominence if the source didn't have first-hand knowledge of where Goldman's earnings are likely to come in.

There's something else very weird about the story, though, and that's its accompanying chart. Here's how the story begins, under the headline "Goldman Faces Loss of $2 Billion for Quarter":

Goldman Sachs Group Inc., known for avoiding many of the blowups that have battered its Wall Street rivals, now is likely to report a net loss of as much as $2 billion for its quarter ended Nov. 28, according to industry insiders.

And here's the chart:

It takes some very close reading of the chart's y-axis and small print to realize that the red bar at the end of the chart is not the $2 billion, $5-a-share loss that the headline is screaming about; instead, it's a loss of less than a buck a share, based on average analyst estimates from god-knows-when. If you've got a front-page-worthy scoop about the enormity of Goldman's fourth-quarter loss, why on earth discard that scoop and use analyst estimates instead for your chart?