Wednesday, June 3, 2015

Stoplights





Ok, so I am in the last month of my career.  I have been asked what are the best and worst parts.  The best are the memories of incredible people and places.  The worst are the missed opportunities.  Providence gives us opportunities and it’s a sin to waste them. As the heir of the greatest generation my contemporaries and I were bequeathed almost unlimited opportunity in aviation.  We did a lot of good things, but we could have done better.  We could have more effectively established the partnership between threat identification and mitigation and standardized operating policy.   I regret the part I played in that.  Did we leave the industry with the more opportunities or are they just bigger challenges.  I am not sure there is a difference, but I am sure that we have not adequately prepared the next generation to understand and meet them.  We have taught them to obey rather than think.

Personally, as I reflect on my career, I have been trying to access what I have learned from four decades of flying airplanes for a living.   It all comes down to this.  We have only been marginally successful at exploiting technology in modern aviation. We have been mostly ineffective in this area because we now value data (information) over decisions.  Since technology is vastly better at information acquisition and management than humans, we are now working for the machines.  We are now the robots.  We don’t use technology to decide what to do, we do what the machines tell us to do.

This all became clear the other day when our family was in the car at an intersection when the car in from of us made a left turn with a red arrow displayed on the traffic light.  The turn was made after the oncoming traffic had cleared and was safe in every way except that there was a red light (arrow).  My family all thought the maneuver was very dangerous.  I asked them why and our discussion was very revealing. 

Just because the light is green is it safe?  Is it always unsafe when the light is red?  Can we trust that the light will always achieve the desired outcome?   Does the stoplight replace decision or just display information?  Can a display of information replace human analysis of the situation?  I believe we would all say no, however what do we practice?  Will the auto pilot always level the airplane at the right altitude?  How do we know it's the right altitude?  Technology can supply information, and lots of it, but only humans can provide the context.  Automation can perform complicated tasks accurately and efficiently, but in what combination.  Just yesterday local teenager was killed when his car rear-ended a semi truck while he was texting.  Is the answer to texting in the car collision avoidance technology?  Really?

Moreover, aren’t policy and procedure just the written version of stoplights.  Volumes are written that say do this don’t do that.  Go this way, don’t go that way.  Like the stoplight there is only binary instruction.  Red STOP and green GO is all we get and we must intuit the context.  Unfortunately, the “why” behind policy is usually much more difficult to derive than that of the traffic signal.  Often the “why” is simply unknown to the user.  Standardized verbiage at specific altitudes are a great example of this.  Compliance trumps understanding.

Machines and their instruction manuals are simply tools.  Tools are what have allowed man to achieve dominion over all other creatures on the planet.  Since the dawn of humankind, man has been using tools to improve their lives, albeit, not without significant collateral damage.  The proliferation of highly sophisticated tools, i.e. technology, is the greatest threat future generations of aviators will face.  This threat exists because of the fundamental flaw of machines, binary decision making.  It is all “1’s and 0’s”, a vast array of little digital stoplights.  Until we move from digital to the next higher level of technology these tools will all lack the uniquely human concept of “maybe”.  Judgment, or the resolution of “maybe” is where humans out perform machines.  A human, unless he cheats, cannot beat the computer at chess because it is computation.  There are a finite number of moves.  Knowing how to write the algorithm for the computer does not make the programmer a chess player either.  It just means he is good at writing code.  The computer is making the “decisions”.  However, when randomness in introduced into a game such as dice with backgammon or cards with poker the computer doesn’t do as well.  The computer doesn’t know what will happen, only what might happen.  The infinite number of “maybes” in the real world makes the roll of a dice seem almost boring. 

The concept of maybe is the inscrutable link between the one and the zero.   It is the fork in the road.  The choices we make are important and the factors in the real world are often unpredictable and random.  This is the difference between risk assessment and threat management.  Risk assessment is the statistical product of probability and severity.  Risk can tell you what the probability of an event is and how bad it can be.  Pretty cut and dried.  Threat management, however, is the process we use to hopefully adjust (lower) those factors (probability and severity).  The environment that is aviation is not predictable, therefore, we cannot totally abdicate the decision making process to technology.   We must use machines and their instruction manuals for what they are, tools. They are not and should not be used as a substitute for decision making.

The 787 Dreamliner that I am about to leave behind is the most sophisticated airliner ever built, but it is still just an obedient, but fallible machine that will only do what it has or has not been programmed to do.  Flying the 787 or any other modern airliner is not playing chess with a fixed number of choices.   We are operating in real time in a capricious and unforgivable environment that can only be tamed by the magnificence of the human brain.  We have turned the corner.  There are no human pilots that can fly the modern airplane as accurately and efficiently as it can fly itself.   How will we teach tomorrows aviators to effectively use these incredible machines?   Is the future of aviation really just the deployment of more and better stoplights?  I believe we could have done better.