42 or 37?

Audio-book Review
By Chet Yarbrough

(Blog:awalkingdelight)
Website: chetyarbrough.blog

Written by: Brian Christian, Tom Griffiths

Narrated by: Brian Christian

“A Hitchhiker’s Guide to the Galaxy” explains that the ultimate answer to the meaning of life is 42; however, Brian Christian and Tom Griffiths tell us it is 37 in “Algorithms to Live By”.

Griffiths and Christian argue that if you want to have an optimum answer to a complex question, it will take 37% of an allotted amount of time to study the known and unknown details of a question to come up with an optimum answer.  Keep in mind, this is not a perfect answer but a probabilistic optimum answer; i.e. an answer based on what is known and unknown.

Christian and Griffiths outline what they argue is an explanation of human decision-making.  The implication of their conclusion suggests AI is unlikely to improve human cognition because it only adds information to complex human questions.

If you sit at a poker table for three hours, the first hour should be used to gather information about your competition.  You will never know everything you need to know to win a hand of poker.  But, you will improve your chances of winning by taking slightly more than 1/3^rd of your time gathering information about the way your competitors play.  This is a simplistic way of looking at Christian’ and Griffiths’ explanation of human decision-making.

The authors identify the discoverer of this algorithm as Merrill Flood, an American mathematician who, with Melvin Dresher, came up with the Prisoner’s dilemma, a model of cooperation and conflict.

Everyone loses in “The Prisoner’s Dilemma”. Christian and Griffiiths note the game can be changed by one variable.  The example given is the introduction of a Mafia leader that says anyone who rats on another will be murdered.  The introduction of this new variable changes the probability of either robber ratting on the other.

The Prisoner’s Dilemma” is the story of two robbers that are placed in separate cells, interrogated independently, and offered a shorter sentence if one rats on the other.  The prosecutor does not have enough evidence for conviction without one ratting on the other. If both robbers rat on each other, they will serve the same sentence.  If only one rats on the other, he/she gets a shorter sentence.  If neither robber rats on the other, the robbers are convicted on a lesser charge.

The 37% factor offers truth but fails to give much comfort to one seeking knowledge about life.  It reminds one of the funny idea suggested by “The Hitchhiker’s Guide to the Galaxy” with the number 42.  The authors suggest 37% is considerably better than knowing nothing but they imply the complexity of life makes outcomes entirely probabilistic.  One presumes–the more you know, the better your decisions will be. Christian and Griffith disagree with that presumption.  They suggest too much information skews the probability of truth.

With computers and the internet, one would think truth would be easier to find.  Christian and Griffith suggest computers only offer added complexity; not truth.  They argue computers are only tools for revealing complexity.

Christian and Griffith suggest 37% is the best one can do in getting to the truth.  The authors suggest there is a point of diminishing return with more information; i.e. too many accumulated facts distort the truth and take one farther away from a 37% probability. A recent example is statistical sampling concluding Hillary Clinton would be the next President of the United States.

There is much more in Christian and Griffiths exploration of algorithms, but it is disheartening to think human search for truth is constrained by a 37% information boundary. A logical extension of their argument is that artificial intelligence is as likely to mislead humanity as human intelligence. The authors argue–the nature of AI only increases information for answers to complex questions.  By adding too much information, more facts are known with less chance of knowing the truth.

This is an enlightening exploration of the world of algorithms and computer science.  On the one hand, it suggests human intuition is highly valuable; on the other, the authors explain it is unwise to rely on instinct alone.  Christian and Griffiths explain life decisions, even with complex computer driven algorithms are less; not more likely to be correct.