Wednesday, March 26, 2008

Spring is here, and we're a little Boyd crazy

George P. Burdell recently called my attention to this article, about how John Boyd's theories of maneuver conflict could be employed in the management of a library. "Wow! Who knew the Dewey Decimal System could be so fun," I exclaimed.

George P. Burdell (a man known by many for his agility and maneuverability) and I have been reading about and participating in discussions about the ideas of the late Col John Boyd since the publication of Boyd: The Fighter Pilot Who Changed the Art of War by Robert Coram in 2004. We're relative newbies to this stuff, so we're still learning.

"You know, George, as I was reading this article, I recalled that you said that the author didn't quite get the OODA Loop, even though he discusses its use quite well."

George closed his eyes and nodded. "Yes, you should check out his resources. he actually points back to an article by Fred Thompson on our friend's, Chet Richards, website as the source for his understanding. The discussion is good, but the depiction of the OODA loop is what bothers me.

As I read the article by Fred Thompson, I had to admit, with George P. Burdell's guidance, that he didn't seem quite to capture graphically the essence of the Boyd Cycle either.

"Maybe Chet needs to provide some gentle counseling to them."

I continued. "So, as I've been thinking about the OODA loop and reading Chet's Neither Shall the Sword, it occurs to me that possibly the best way to describe the OODA loop is not a process loop but as a feedback loop, a type of cybernetic control function."

"In a very simplistic way, I can think of my thermostat-heating system as a kind of OODA loop. The slowest possible configuration I can imagine might involve my running up and down the stairs from my bedroom to my basement to turn my heater off and on in response to temperature measurements I take. In this case, exhaustion and the elements are likely to win since they are always on and relentless. However, if I create an electric feedback loop (as opposed to my running kind) from a thermometer in my bedroom to the switch on my heater in the basement, my heater will respond to the variance in measured temperature to desired temperature faster than before and respond accordingly. In other words, there isn't really a process cycle taking place in the thermostat-heater system in which some agent steps through various commands, that is
  1. Measure temp
  2. Compare to desired temp
  3. Turn on heater for x minutes
  4. Return to 1
The electric thermostat-heater system is actually always on (This is certainly true for analog systems in which measurement is continuous. Digital systems do step through a measure-test process, but so fast that it is essentially continuous. Regardless, the benefit is derived from how fast information is transferred.) The benefit of this configuration is not really related to how fast it goes through process steps, but how fast information (measurements, observations) can be reconciled with the environment and acted upon. Since the transfer of electric signals are faster than heat diffusion, my system 'wins' in that it effectively manages the variance from a desired set point of temperature."

George seemed to go into a trance at this point, and he began to channel the disembodied spirit of Chet Richards, which is odd not so much because Chet is still very much active right here in Atlanta, but because George totally disavows such voodoo-like stuff. "This is a very important observation - it's a dynamic process (for lack of a better word) - like a whirlpool.  It's always 'on.'  Boyd was strongly influenced by the ideas of the late Ilya Prigogine, who used the phrase 'dissipative structures,' of which a whirlpool is a classic example. Absent the continual flow of energy provided by moving water, it can't exist at all. It's either 'on' or it isn't there. Similar thing in biology.  As I understand it, even at rest, nerve cells are firing at a basal rate of around 60 Hz.  To send a signal, that rate shoots up.  So orientation is continuously sending 'signals' to / controlling action via the implicit guidance and control link."

"Your feedback loop analogy is interesting.  To work, the system must not only transmit signals faster than the room heats or cools (decide / act), but the thermostat must register changes much faster than the temperature in the room is actually changing (observe / orient).  This is, of course, the classic control problem:  Make it too sensitive and the pilot will skitter around all over the sky.  If it isn't sensitive enough, or if there are significant lags between input and effect, the pilot can induce uncontrollable oscillations."

I continued with my own thoughts, staring curiously at George, who was now levitating in a rigid prone position above the floor. "In more complex and dramatic human contests, the winner is the one who can process measured information (which metrology tells us is always less than absolutely certain) about the environment, interpret the observation, and reconcile the interpretation with reality in such a way that allows transition from one state to another faster than the opponent who is also doing the same thing. Again, it's not that the winner went through process steps faster, but that the winner was able to measure and interpret information faster to allow for faster, more unpredictable transitions from one state space to another such that its measured state space by an opponent is increasingly variant from reality. In other words, it's not speed along a given dimension that makes me a more effective contender, but the ability to accelerate between dimensions unpredictably."

"Hold on a minute! I think we can describe this mathematically. If we think of a state space as a vector of dimensions of concern of an agent X, representing it as {X}, we can represent the measure (observation) of X's state space by another agent as {X*}. The absolute variance could be given by Vx= |{X-X*}| . X's opponent Y has a state space {Y}, and X's measure of Y's state space is {Y*}. Then X is most likely to win over Y if Vx'' > Vy''. In words, X is most likely to win over Y if the second order time rate of change of the variance between the actual state of X and its measured state space (as observed by Y) is greater than the second order time rate of change of the variance between the actual state of Y and its measured state space (as observed by X). In conflict, you're more likely to win over me if you are able to switch position faster than I can accurately measure and respond to."

Suddenly, the phone rang. George floated over to the phone and answered it. "It's for you. It's Col Chet Richards." Now I really felt like I was in the Twilight Zone. George rotated around his waist-line axis into a head-down position now, arms outstretched, feet pointing to the sky, suspended, unattached to any of his acrobatic wires, four feet above the floor.

"That's a very perceptive observation, Rob, and I think you've penetrated to the heart of the concept - 'broken the code,' as Boyd would say.  Using airplanes, which was the original inspiration, define the state vector to be its altitude, airspeed, and direction.  'Maneuverability' is the ability to change the state vector relative to time, that is to climb/descend, accelerate/decelerate, or turn in any combination.  'Agility' is the ability to change maneuver state, which is the second derivative of the state vector function.  Boyd said that the most agile aircraft wins." Chet's voice was nonchalant.

"And that's true, IF (big 'if') the pilot of the more agile aircraft knows how to use this agility advantage to end the engagement on favorable terms (i.e., shoot the other guy down).  That's really the problem Boyd addresses in his Discourse on Winning and Losing."

"So it's as you said, change your position - or 'state' however defined - more rapidly than the other side can comprehend.  And keep doing this until you create some type of advantage that you can exploit to end the situation on favorable terms.  And you can be sure that you will reach such a position because by operating inside the opponent's OODA loop, you're degrading the ability of your opponent to function as an organic whole. Very shortly, he's going to start fracturing into 'many non-cooperative centers of gravity' that begin to pull and push in different directions (e.g., split into cliques that bicker among themselves).  You'll sense that he's having ever more problems responding effectively - in the business world, products are late and seem outdated when they do arrive (Vista, anyone?), proposals don't correspond to what the client wants, huge amounts of time and money are written off time and again (GM / Fiat) and so on."

"It's knowing what to do...which sounds so simple, doesn't it? AND being able to do it, more rapidly than your opponent can understand.  So that by the time he does kind of understand, you're already doing something else.  It won't take too long before he starts coming physically, mentally, and morally unglued."

"One point - it's not 'more likely to win.'  If one can keep this flow going, one will win.  Hence the title of my book -- Certain To Win: The Strategy Of John Boyd, Applied To Business."

At that point, Chet unceremoniously hung up the phone. George maneuvered his way to the floor, rotating as he did into a supine position.

"Whoa. What just happened?" George mumbled, still groggy from the trans-corporeal channeling.

"I think it's spring, and you've gone a little Boyd crazy."

No comments: