On 14 April 1994, two United States F-15s shot down two United States Blackhawk helicopters in the clear skies of northern Iraq while both flights were controlled by the same AWACS. How could this have happened? Scott Snook wrote a book about it. I read that book.

It's called Friendly Fire: The Accidental Shootdown of U.S. Black Hawks Over Northern Iraq. It is informative and compelling; my only mild criticism is of the writing itself, and even that might not exist but for comparison to the masterpiece that is The Challenger Launch Decision. Snook's book, like Vaughan's, starts with a simple summary of an outrageous disaster - in this case, a helicopter shootdown that killed 26 people. He then explores its individual contributing factors. As it turns out, the simple summary hid a bevy of problems, each theoretically troublesome but never previously catastrophic, that happened to combine in the worst way on one terrible day. The fighter pilots' training was insufficient for visually identifying friendly helicopters, the AWACS crew was poorly constituted and trained (and only nominally "controlling"), Army helicopters and Air Force planes didn't talk to each other for both institutional and technical reasons, and the IFF system flat-out broke.

Here we see a concrete difference between the two books. Vaughan was an ethnographer recruited to provide an outside view of an opaque institution, but Snook is a commissioned military officer seeing everything with an insider's perspective - and potential bias. (He is up front about this.) How his analysis differs from that of a counterfactual outside investigator we'll never know, but his account felt thorough and objective. He included quite a bit of testimony from the court-martial that followed the incident, although his habit of burying it in footnotes generated unnecessary narrative friction. (He is not a bad writer, but I think a great writer could have written a great book.)

The book finishes with Snook's analysis and proposed contribution to the accident and safety literature. His theoretical point of departure is Charles Perrow's "normal accidents" theory, which posits that accidents are unavoidable (or "normal") in systems that are both highly complex and tightly coupled (such as nuclear reactors). Snook's contribution is a time dimension. The planners of Operation Provide Comfort knew that the operation had a lot of moving parts that needed tight synchrony, and they created encyclopedias of rules and regulations designed to provide it. But over time, the men and women on the ground implementing the plan - understandably - made local adaptations to do their jobs. This gap between work-as-imagined and work-as-done should be immediately familiar to anyone who's had to reconcile managerial prescription with actual reality, and Snook is very clear that it's not inherently bad. It's frequently the only way for crews to accomplish their objectives. The tightly coupled system in theory became loosely coupled in practice, a phenomenon Snook calls "practical drift."

These various local adaptations were definitionally uncoordinated. (That's what "local" means, and is what made the coupling "loose.") But the overall plan still existed, and was still written down, and everyone thought that everyone else was still following it. And one day, the same arrow of time that had shepherded the adaptive loosening of Operation Provide Comfort's various discombobulated parts snapped them back into violent, transient conflict. It had not previously mattered that F-15 pilots couldn't visually identify Blackhawks, or that AWACS were effectively perfunctory, or that nobody knew what the hell the Army was doing - until 14 April 1994, when it very much did.

What can we learn from this? Snook admits that he is largely at a loss himself, because every chapter in the terrible story describes normal people doing normal things in normal environments. He falls back to this: The lessons we've already learned, we must actually apply. We should ignore our temptation to describe sensemaking as "error." We know how to build high performance teams, and our failure to actually do so carries consequences. Social redundancy has known risks. Organizational integration is ongoing work, and neglecting it brings peril.

This all resonates with me. I've been in meetings that waved off known organizational faults. I'm fortunate enough to have never worried about fatal consequences thereof, but the phenomenon is still familiar. This stuff matters.

I'm glad I read this book, even if the conclusion wasn't much more than "I guess the researchers know what they're talking about." I hadn't even known about this accident before reading it, and it was pretty cool to see the arguments of Perrow et al applied directly to a new incident. (Charles Perrow himself apparently reviewed the book, which is neat.) Maybe someday I'll have something of my own to add to the canon, although hopefully it won't be quite so deadly.