Wednesday, November 11, 2009

Mind the Gap

Slatest passes along a story, "Military Wasn't Told of Fort Hood Shooter's E-Mails," that appeared in today's Wall Street Journal as "Agencies See Gaps in Sharing."

The Slatest post is a great example of saying "OHMYGOD...actually might not be much of a story here," the WSJ article is a little less so. Both make out the story to be "we spent millions after 9/11 to improve information sharing but here's a clear case where information wasn't shared just like in olden days" (with the implication (though the articles admit this is not a sure thing) that things might have turned out differently if information had been shared).

Some sociology of information fundamentals at work here. The WSJ article actually describes what sounds like a pretty thorough process of assessing whether or not to pass along information. For what sound like good reasons, the decision was not to. Now, maybe they need to revisit the structure of their decision process (and this is not necessarily true as no one appears to have shown that the information would have made a difference), but that's different from the story being that agencies are not sharing information.

A senate official is quoted saying "[a]ll signs are indicating that something wasn't put together." But this might be misleading. The article uses a favorite phrase from 2001, "connecting the dots," and speaks of "intelligence gaps." I think both of these are uttered too glibly and unanalytically. These sorts of events bring out massive displays of "hindsight bias" -- the tendency to see things after the fact as a lot more predictable than they really were.

There is probably also a problem with the geometric metaphor of intelligence gaps. We might be able to distinguish topological gaps (information in one place does not reach another place) from topographical gaps (the empty spots in information that any particular knower has access to). When an event like Fort Hood occurs, we start to fantasize about a world in which the gaps pointed to by hindsight would have been bridged over. But to guarantee that we probably need to posit a world in which there are no gaps, but that's a world in which everyone knows everything and without a division of informational labor, the whole thing grinds to a halt.

We need to zero in on how humans share relevant information with those for whom the information is relevant. Competent nodes in an information network have good working models of the relevance systems of the nodes they are connected with. We don't want to eliminate intelligence gaps, we want to make the gaps (read links) more intelligent. And that probably comes most from interaction. And that's something that organizations and agencies are not naturally prone to. What the analysts should look at is what we've spent the millions of dollars on in our quest to fix the intelligence gaps rather than just implying that the effort has been wasted.

Saturday, November 7, 2009

Information Abhors a Vacuum?

Great essay today by Scott Simon on NPR's Weekend Edition. He called it "The Bombastic Fog Engulfs Fort Hood."

Long story short: very quickly after the events at Ft. Hood on Thursday afternoon there were items appearing in the media providing all manner of explanation of things that might or might not have anything to do with those events.

Simon's initial diagnosis is the structure of modern mass communication itself "...in these days where almost anyone can find some kind of audience." A certain kind of event, such as mass killings on the U.S. Army post, "encourages people to analyze and speculate in advance of a lot of actual facts."

He goes on to give a few other examples and then zeroes in on how journalists ran with the idea of pilot fatigue when some airline pilots missed their destination a few weeks back, making the jump from the speculation of experts in the absence of direct knowledge of the circumstances and details of an event to research on the science and politics of pilot fatigue. Any number of stories along this line were produced only to be proven irrelevant (in Simon's on-the-mark characterization) when, upon investigation, it turns out the pilots were playing with their laptop computers.

This reminds me of a few things. One is the propensity of some journalists (and some social scientists, too) to decide very early on "what the story is." That's what your editor wants to know as soon as you pitch an idea -- what's the story, what's the angle? We've all been contacted by a reporter looking for a quote that confirms a particular line they've decided to take in the story or a student who is looking for some research that supports a particular conclusion she wants to draw.

The second thing is that competition for eyeballs and ears forces people who talk and write for a living to talk and write whether or not they have anything to add to our collective knowledge.  Dead air is bad.

Those are professional errors, malpractice, if you will, even if of a mundane sort.

There's probably something else going on too -- something more at the "information order" level than the professional practice level. It is fundamentally difficult for a community to learn of an "untethered" fact (unconnected, that is, to a story that grounds it in the web of our taken-for-granted worldview (Weltanschauung)) without someone stepping up to tell a story that does ground it in the known. 

And so, the urge that insiders sometimes have to not announce something prematurely "because it will lead to speculation" is probably not nearly the pathology we often make it out to be. 

After listening to the essay I began to think about thought experiments in how to balance the incentives.  If, as Simon says (that phrase was going to come up in this essay sooner or later), it's because its so easy to "find some kind of an audience" (or at least a soapbox around which there could be an audience), then maybe we (members of the chattering classes -- both amateur and professional) should give some consideration to what we'd say if there were a word tax as well as a word rate.  If what you have to say turns out to be irrelevant, not only do you not get your $2 per word, you actually have to pay the rest of us for the bit of our information universe you filled up with worthless drivel.

Notification on TV

Once you start thinking about notification, you see it everywhere. Just in the last few days, it's figured centrally in episodes of PBS's "Masterpiece Mystery: Inspector Lewis" and AMC's "Madmen" (see also 9.20.2008 and 9.8.2008).

In episode 12, "The Grownups," Pete chats with Harry with the TV turned down. The audience can see Walter Cronkite talking about a news flash from Dallas but Pete and Harry are too engrossed in their conversation. We cringe knowing what they don't know but are about to find out. Other characters then crowd into Harry's office to watch the news. Don emerges from his boss's office to see the main work area basically empty but all the phones ringing. He's beside himself trying to figure out what's going on. Then he does. Later a few of the characters talk about the fact that they "just had to call" so and and so (this even though news of the assassination was one of the most quickly diffused messages in history up to that point.

And, of course, about half of the dramatic tension of the entire show is generated by all the secrets kept by characters from one another (with the audience tipped off and forced to watch painfully as characters they care about remain in the dark).

SPOILER ALERT. In the "Inspector Lewis" episode titled "The Quality of Mercy," Lewis' Sergeant discovers some information about Lewis' wife's death a few years earlier. He gets the info on a phone call while Lewis is sitting next to him but says "oh, nothing" when Lewis asks him what it was about. When he eventually tells Lewis later that day, Lewis is furious and takes it as a sign that their relationship is really quite flawed. Sergeant Hathaway explains that he withheld the information because of their relationship, but Lewis pretty much says "we don't even have one if you thought it was O.K. to wait to tell me."