The present promotion of employing big data emphasizes its potential for making beneficial decisions.   In this context, big data includes the wide range of associated technologies for analytic predictions or prescriptions presented using rich visualizations.  There are many examples of prior successes where the rapid exploitation of very large and diverse data sets have produced beneficial outcomes for businesses in terms of increased market share, higher profits revenues, or improved efficiency for reducing costs.  Success stories are largely self-confirming.  The evidence of the success justifies the investment and practices that supported the original decision.

Receiving less attention are cases where the decision fails to produce beneficial results or even produces damaging results.  In recent posts, I’ve been using examples unrelated to big data where we have have trouble obtaining accountability for decisions that turn out bad.   Certainly, there is an incentive to not draw attention to a failure.   Also, failures take more time to acknowledge than successes.   We can announce a success as soon as we observe positive results, but we will postpone declaring a failure on observations of negative results because we think the project needs more time or more refinement to succeed.    There is a risk of prematurely declaring failure for a project that may be destined to succeed ultimately.   In contrast, there is an incentive to prematurely declare success to attract more funding or attention.

In my last post, I presented another reason we don’t hear as much about failures.  In that post post I emphasized society’s needs for accountability when decisions result in unfavorable consequences.   To maintain order, we require authorities (people given public trust) to provide a persuasive accounting to defend the wisdom of their decisions in order to satisfy those with grievances and to win active support of a super-majority in favor of the decision that happened to have a bad outcome.

The problem with failures is that we confuse the demand for accountability with the demand for justice.   The two are mutually exclusive.   Our justice system permits a person to remain silent in order to not self-incriminate but it will use any person’s statement against that person for the purposes of incrimination.    Accountability demands that the decision maker speak up in defense of his decision.   Justice demands the defendant to remain silent in order to force others to gather evidence to overcome doubt.     Accountability demands complete cooperation from the decision maker while justice demands the opposite.

Accountability is very important for maintaining super-majority consent of government or of widespread good-will toward a particular product, company, or brand.   A decision maker demonstrates accountability by effectively answering his critics: those who feel aggrieved by his prior decisions.

In recent decades, there has been a rapid increase in exposing decision makers to more risk of criminal or civil prosecution.  There is very little room for decision makers to provide accountability without jeopardizing themselves for criminal or civil charges.    The result is less accountability.    When confronting unfavorable consequences of decisions, we must find indirect evidence because we have no access to the direct accounting from the decision maker himself.   The decision maker is advised to remain silent in order to avoid incrimination.

Accountability should come with some immunity from incrimination.   Accountability can be more important than justice as evidenced by recent events of destabilizing public protests.   Accountability is required to quiet the protests or to build a strong super-majority coalition in support of a decision that happened to have unfortunate consequences.    With that accountability, we can maintain or strengthen the peace.   The benefit of peace and consent balances cost of not obtaining justice.  Sometimes, accountability is more important than justice.

For example, a recurring problem is the collisions of vehicles with pedestrians resulting in pedestrian death.    The operation of the vehicle resulted in a homicide.    Instead of demanding justice for the fact that the homicide would not have occurred if the vehicle were not moving, we obtain accountability from the operator to find out the that the operator’s decisions and actions were acceptable within the rules and demonstrate reasonable diligence.    Although the death would not have occurred if there was no vehicle involved, we do not demand justice from the operator.   In some sense, we grant an immunity in the form of saying that if the vehicle operator follows the rules, he can not be guilty of homicide.   We find closure with the honest and thorough cooperation of the operator to describe the actions and decisions that led to the accident.   That closure is sufficient to placate the ones who suffer the loss of a loved one and to reinforce widespread support for the current policies of operating vehicles in public places shared by both vehicles and pedestrians.

We endure vehicular homicides of pedestrians of about 12 deaths per day without any destabilizing protests.   We maintain this peace in large part due to the cooperation of the vehicle operators to provide accountability of their actions.   That accountability is possible because we grant an exception from justice if the operator followed the rules.    In modern times, we take this kind of example for granted.  However, it is not hard to find historical examples or even examples from other cultures where justice would not be excused so readily.    There was a homicide.   The cause of death involved a collision with a vehicle under control of another person.   A concept of justice could demand prosecution even if the operator followed the rules.

Justice can expect retribution either to make the aggrieved whole again or to impose an equal injury on the one who caused the first injury.  This concept of justice is destabilizing.  It is impossible to balance a mistake with no ill intention with a deliberate action with an explicit intention of harm.

We need a way to quiet the demand for justice in order to have access to accountability.   Justice answers an individual need.  Accountability answers a social need.   The bargain for the individual to withdraw a demand for justice is the benefits obtained from readily obtained accountability.   Accountability makes possible peaceable social arrangements that enjoy widespread support and consent.    Accountability provides valued information about mistakes so we can learn how to avoid making the same mistakes in the future.   The unfairness of an immediate event may be balanced with a more fair future that this learning can provide.

Early in automobile history, we observed that a number of collisions with children involved cases where the children were concentrating entirely on chasing a ball.   The accounting of the details of the accident taught us to be alert to moving balls in addition to moving children.   We brake for the moving ball because we know of the possibility of a child hidden from our view may be chasing that ball.   Although such ball-chasing events still explain some collisions I suspect the number of such collisions are smaller than if we ignored the significance of a ball moving near a street.

I find it fascinating how thoroughly we accept accountability for vehicular accidents instead of justice.   Every time I hear of damages, especially involving injury or death, I am amazed at how readily we determine that it is an accident that does not require criminal charges.   Even civil lawsuits are rare as a result of quick settlements with insurance policies.   For the specific cases of damages caused by personal vehicles, we are content with accountability without justice.

Operating a vehicle is not much different from operating anything else.   The operator makes decisions based on his training to interpret the information available to him and his ability to control the system.   The decisions sometimes can have bad consequences that result in injury (including death) of others.   However, in most other systems, we have far less tolerance for pure accountability when decisions go bad.   We also demand justice.

A company executive may make a highly skilled decision on the best available information.   From a perspective of accountability, this decision may be far more justified than the actions of the operator of a vehicle that lead to injuring or killing a pedestrian.   In spite of that, if the decision results in unfortunate consequences we readily prosecute the executive with the jeopardy of incarceration for the rest of his career.

The authoritative operation of just about anything other than operating personal vehicles, involves a very strong threat of criminal prosecution for the sake of justice if something goes wrong.   That threat has the effect of silencing the authority and thus denying the direct accountability that we really need for the larger projects.    We benefit in many ways to a candid explanation of the decision making process so we can understand why the decision was a good one that happened to turn out bad, or so we can understand how to avoid making similar mistakes in the future.   The best way to get this information is directly from the decision maker who is not silenced by the threat of extracting justice.

We do need a way to evaluate decisions.   At the start of the post, I mentioned the abundant discussions of successful decisions.  There are many incentives to quickly and frequently advertise successful or beneficial decisions.   At the same time there are many disincentives to draw attention to the unsuccessful decisions.   When deciding something like trusting a new technology to replace an older one, similar to the promotion of big data technologies to replace deliberative decision making, we need information about the failures.  Even for the successes, we need to know about the counter-factual missed opportunities.  In both the failure and success stories, it is possible that a different decision based on a different approach could have resulted in better outcomes.

In my last post, I asserted that evidence-based decision making is insufficient for accountability.   When asked to defend a decision, the decision maker needs to address the non-evidence of doubt.   The concept of due diligence involves the exercise of doubt.   When we seek accountability, we ask whether the decision maker sufficiently identified all areas of doubt and that he adequately balanced those doubts against the evidence.   Decisions involve judgement that can not be automated.   The demand for accountability is an evaluation of the quality of judgement.    Accountability can teach us how to make better judgement in the future or it can allow us to select better candidates for fill the role of being decision makers.

The consent for cooperation by a coalition of competing groups depends on this trust in the judgement of the decision makers.   The decision makers are intelligent and act with good intentions, but they also appropriately balance doubts with evidence.   Accountability provides the proof that the decision maker did not act recklessly.

Being held accountable for a decision, a decision-maker will pay attention to all of the opportunities and balance the known information with the doubts.   Investing this consideration of all options prior to the making of a decision improves the defense of the decision later.

The attack or defense of a decision involves an evaluation of the missed opportunity.   In an earlier post, I discussed the challenges of evaluating the missed opportunity.  We can not know what would have happened if we had made different choices.   We only have facts or evidence for the consequences of the actual decisions.

Returning to an automotive example.   Several years ago, I stopped my car at a yellow light that was about to turn red.   Apparently the car behind me figured there was still time to get through and he ended up hitting my car.   The damage was very slight and there was no claims made.   This trivial fender-bender type of encounter could have been avoided if I had continued through the intersection without stopping.   The only facts available are what actually happened.   Something far worse could have happened if I had not braked.   Chances of that happening may have been small but those chances would change to certainty if I had proceeded and caused a worse accident.   My decision to stop clearly resulted in becoming rear-ended, but was it the best decision.   This is an interesting example because even though the other driver may be at fault, the aftermath was a result of my decision to stop.   I can see good defenses for the decision (obeying traffic laws, recognizing risks of collision, etc) but there could also be good criticisms for the decision (failing to recognize the intention of the driver behind me, failing to appreciate the local custom for this particular intersection).

The advice for even fender-bender collisions is to exchange nothing more than basic contact and insurance information.  In this case, the other driver did not even stick around for that much.   Even if we exchanged information, that information would offer nothing about what actually happened.   The advice is to remain silent about the most important details about what happened.    We can not account for the accident.    In the previous paragraph, I described a possible learning experience that this particular neighborhood has a well accepted tradition of entering the intersection on the initial moments of a red light.   It is just how everyone drives there.    This is similar to the custom of driving about 30% faster than the posted speed limit because this is the prevailing speed, it doesn’t cause any harm, and police don’t enforce violations below this amount.    Obeying a law may be a bad decision.

Again, personal automobile examples are relatively trivial.   Even when collisions result in major injury or death, we frequently attribute the collisions to accidents instead of accountable decision making.   For most other decision making, we are far less inclined to accept the explanation of an accident.   We need to know more about what went into the decision.  We want confirmation that the decision maker is competent to make difficult judgement calls of selecting the right decision.   We want to know that the decision maker had access to all relevant evidence and doubts to make the decision.    This investigation gives us the information we need to accept the consequences because we accept the wisdom of the decision.

In these recent posts, I have been contrasting my concepts of justice from accountability as opposing or mutually exclusive concepts.   For a particular decision that results in harm, we can either expect justice or accountability.   Justice answers an individual need for closure while accountability answers a social need for closure.    Justice silences the decision maker from offering his candid explanation for a decision while accountability requires precisely this explanation.   We can’t have both.

When we seek accountability we seek information about recognizing the missed opportunity.   In the future we hope to be better prepared to select the better opportunity or be more confident that the choice we made the last time remains the better choice.   The difference between accountability and justice is a difference of scale.   Justice is about individual level punishment to compensate for an injury.   Accountability has a broader objective of maximizing the welfare of society as whole.   For accountability, we want to learn from the decision maker his considerations and judgement that led to dismissing the missed opportunities.