Richard Berk, Professor of Statistics at the University of Pennsylvania, has developed an algorithm which will help Philadelphia’s Adult Parole and Probation Department make decisions on which candidates for parole will be the most likely to repeat offend.*  The department supervises nearly 50,000 individuals.  Since their prison population vastly exceeds capacity it is urgent they release as many as can be safely returned to the community.  It is that “safely” part that Berk’s algorithm is intended to help identify.  Using the records of 100,000 inmates, and going back as far as 1962, Berk’s algorithm forecasts the likelihood of repeat offense with a great degree of accuracy.  He has found that the age at which a first offense is committed has much greater significance that the severity of the offense – a man who was convicted of armed assault at 13 is likely a greater threat to the community than a man who committed murder at 22, but had no prior incidence of violence.

            The Philadelphia Algorithm has been accused of being seriously flawed.  Some see Berk’s refusal to use race as a factor in his calculations as a grave mistake.  Others see his use of ZIP codes as a factor another way of profiling certain neighborhoods, and sneaking race in through the mathematical back door.  It, and other similar mathematical predictors are widely used now, and will be more so in the future.  This is the old CBS crime-drama “NUMB3RS” come to a police station near you (or the new CBS crime-drama “Person of Interest”).  Can all human behavior be reduced to an algorithm?  If so, would it be acceptable for government to base a decision about who to parole based solely (or even largely) on mathematics?  If we are all so predictable, where is there any room to hold a person responsible for choice?  Are we not all slaves to the external factors accounted for in the algorithm?

            The Philadelphia Algorithm may be exponentially more accurate than its predecessors at predicting criminal behavior – but it is not perfectly accurate.  It is only pretty accurate.  It could only be perfect if it factored in every possible variable.  That would take omniscience.  Omniscience would render such an algorithm obsolete.  If one is omniscient, one would know what is about to happen, and who is about to make it happen.

            One does.  That one is God.  He is omniscient (Psalm 139, Matthew 10.29-31).  His omniscience does not take away our own responsibility for deciding and acting.  We are still responsible because, although God knows ahead of time what we are going to think and do, He does not force us to think or do anything.  We are responsible because God’s law is not based upon variable factors, but upon the changelessness of God Himself.  A lie is a lie because God is truth.  It becomes no less a lie because of environmental factors.

            It seems sensible to use precise data to guide the administration of law and order in a society.  It seems sensible to remember that humans, in all their unpredictable prejudices and presumptions, will be gathering, sorting, and applying that data.  Thus, it is also sensible to remember that our human ability to achieve justice will never be better than nearly accurate.

            God’s administration of justice can be calculated perfectly.  Isn’t it grand that although God in his omniscience could bring harsh mathematical justice to bear upon us, He has offered us another equation.

If anyone is in Christ he is a new creature.  Old things are passed away. Look – all things have become new. (II Corinthians 5.17).

            Grace wipes away past offenses, taking no thought of recidivism – other than to provide payment for further offenses (I John 1.5-7). Although God’s omniscience is not used to prevent human sin – His omnificence is used to address it.  This is something no algorithm, regardless of how perfectly calculated, will ever enable us to do.

*Nadya Labi,  “Misfortune Teller” in The Atlantic, February 2012, pp18-19.