Sunday, October 22, 2017

Simulated Annealing and the Art of Detection



“My way of learning  is to heave a wild & unpredictable monkey-wrench  into the machinery.”
-- shamus Sam Spade, in The Maltese Falcon (1929)


Dashiel Hammett’s P.I.’s find themselves in large, chaotic situations, in which the methodical self-discipline of, say, Dragnet, won’t get you anywhere.   Joe Friday is scrupulous, professional, polite;  Hammett’s guys  are none of those things.
Nor do they follow the cool, cold-blooded approach of what we might dub “Newton’s method “of successive triangulations and approximations.  Instead it’s: stir up the pot. Stir and observe, stir and observe;  add bullets to taste, p.r.n.


cf. Newton’s method of repeatedly refining an initial guess.

~


The police want to grill a man they suspect of having abused and murdered his step-daughter, but have no tangible evidence.

“I don’t know.  What are we going to ask him?”
“If he reads the Evening News. … Christ, Tommy, we can ask him what his favourite colour is.  I just want him in here, under pressure, so we can see what happens.”
-- Val McDermid, A Place of Execution (1999), p. 205

~

And thus it is with Murphy, Private Eye.  He is no master of empirics nor of ratiocination, like Holmes;  nor of intuition into souls, like Father Brown;  but he’s got heart, and he can take a punch, and he’s got a gun;  plus what’s a fella to do?  He’s got a Mission.

So Murphy has what starts out as yet another hopeless case:  His so-called “client” is a ditsy dame with no simoleons to pay him;  plus she gets bumped off practically at the outset;  but he soldiers on.  Because that’s what soldiers do, when they’re on a Mission.
And so it looks hopeless but, he stirs the pot a little, tries this and that, and the next thing you know …

To follow this sobering  yet inspiriting  adventure, simply click here.

For more from Murphy -- philosopher and P.I. -- check out his blog:

=>   http://murphybros.blogspot.com/



~

In sum:   Jiggle the system, let it settle into a local equilibrium;  then jiggle again.  Repeat (with decreasing strength of perturbations as the ideal is neared).


Related to this is an epistemological principle, valid for scientists and detectives alike:   Don’t jump to conclusions.   Or, in a quantum-computing metaphor, don’t provoke a premature collapse of the wave-function.  Keep the various possibilities  juggling in the air, until one or more can be definitively ruled out.
 

No comments:

Post a Comment