The third intention in Patterns 132 reads: Select initiative (or response) that is least expected. The standard explanation is that the least expected response will produce surprise which we can then exploit. Seems obvious, but if you think about how most organizations pick their actions, it’s by some formula or just what they’re comfortable doing (“If sales are down, lower prices.”)
In war or the martial arts, surprise often produces disorientation and a moment of confusion and hesitation. This leads people to assume that time is of the essence, that we need to operate at a faster tempo than opponents to keep them off balance. This can be a powerful tactic, as Boyd explains in an “illuminating example” in Strategic Game, pp. 39-44.
But operating at a faster tempo isn’t strictly necessary, especially in forms of conflict other than war. But the “unexpected” effect can still work when, for example, you can let the opponent’s imagination do your dirty work for you. A great example of this is the chess match between then-reigning word champion Garry Kasparov and the IBM computer Deep Blue in 1997. The computer made an unexpected move:
“It was an incredibly refined move, of defending while ahead to cut out any hint of countermoves,” grandmaster Yasser Seirawan told Wired in 2001, “and it sent Garry into a tizzy.”
Turns out that the unexpected move was the result of a bug in the software, but the effect on Kasparov was decisive:
The irony is that the move had messed with Kasparov’s mind, and there was no one to fix this bug. (emphasis added)
“Kasparov had concluded that the counterintuitive play must be a sign of superior intelligence,” Campbell told Silver. “He had never considered that it was simply a bug.”
If we are able to operate at a faster tempo, we can decrease the time opponents have to figure things out, to operate inside their OODA loops, and pump up ambiguity, but the effect works even when this isn’t possible.
Note that the unexpected move was the result of an error in the software, which was corrected between games. Errors have produced similar effects on the battlefield — Boyd would often cite the Union attack on the center of the Confederate lines at Missionary Ridge that launched Sherman on the road to Atlanta.
Did a Computer Bug Help Deep Blue Beat Kasparov? Klint Finley, Wired, September 28, 2012
Slightly East of New 
Surprise can also select something that has the least countermeasures … especially in areas where there tend to be focused point defenses. A decade ago there was large pilot deployment in the US of a electronic payment chipcard (common in Europe). The people involved had myopic focus on the chip characteristics, especially countermeasures to lost/stolen card. However, it was possible to use same skimming techniques (used with magstripe cards) and it was almost as trivial to create a counterfeit chipcard (as magstripe card) from skimmed data (skimming attacks were over a decade old by this time). The counterfeit cards got the label “Yes Card” .. since they answered yes to all three questions (by terminals) 1) was the correct PIN entered, 2) should it be offline transaction and 3) was transaction within credit limit. Since “Yes Card” always answered yes to the PIN question, it wasn’t even necessary to skim the correct PIN.
When this was described to the people getting ready for the pilot deployment, their response was that valid chipcards would be configured to always do an online transaction. The issue is that the countermeasure for counterfeit card has been to invalidate the account number (which is checked in an online transaction). The problem is that the configuration for a valid card has no effect on the configuration for a counterfeit card and (countermeasure for counterfeit card) invalidating the account no. has no effect on a counterfeit “Yes Card” that never goes online. The description made by Secret Service at ATM Integrity Taskforce meetings resulted in response “they managed to spend billions of dollars to prove chips are less secure than magstripe”. After the magnitude of the problem finally began to dawn, the pilot seem to evaporate without a trace … and created quite a bit of resistance to repeating such a pilot until the technology had been much thoroughly vetted.
I am reminded of something algebraic coding theorists like to say (riffing off a theorem in Claude Shannon’s epochal Statistical Theory of Communication) – “Any code of which we cannot think is good.”