737 Max Disaster: A Socioeconomic Trend
My extreme fear of flying drives me to pay minute attention to airplane disaster anywhere in the world. Pilot fatigue, sudden weather change, catastrophic mechanical failures are the major causes of commercial jet crashes. I’m not ignoring our own urge to take a flying machine down to ashes, as it transpired in 9/11 attack on World Trade Center, GermanWing Flight 9525 pilot on a suicidal mission in 2015 or the Malaysia Airline MH17 hit by missile while flying over rebel-ruled sky on Ukraine back in 2014, but those events are widely recognized as “human stupidity” or “act of terror”, no causal relation with the disaster itself. The disappearance of MH370 baffled the whole world on how a modern aircraft vanishes from the radar of modern civilization where every corner of Earth is under satellites surveillance. Joint effort from many nations could not uncover the mystery, even if some partial success were achieved recovering some parts of the fuselage from Indian Ocean. We still don’t know the fate of MH370 is due to some cascading events of failure or human stupidity. Thanks to the science and engineering, every airline disaster goes under the lens from multiple agencies across the globe. Finding the cause of disaster and making immediate attempt to avoid any potential disaster due to the same reason made flying far safer than driving a car on a national highway. The global nature of the commercial aircraft industry (primarily dominated by USA based Boeing Inc. and France based Airbus Industries) made such investigations truly global by sharing knowledge across forensic laboratories, scientists and experts without any national border. Every such investigation changed the course of flying on a positive direction, either by making significant design changes in the aircraft, or adding more buttons and screens in the cockpit, or retraining the pilots on how to deal with certain situations.
The news of Lion Air Flight 610 on October 29, 2018 drew my attention almost immediately, as the disappearance of MH-370 and AirAsia Flight QZ8501 in the same fateful year made the Malay Archipelago a region of special interest. The preliminary investigation revealed that the airplane struggled to gain height during the takeoff phase and dived into Java Sea within 12 minutes of leaving the port of Jakarta. The pilot, Captain Bhavya Suneja with 6000 hours of flying experience, was apparently fighting with the Automated Anti-Stall system of the brand-new product line of popular 737 series, 737 Max 8. The revelation was horrifying. If the preliminary finding happens to be true, the disaster is no match to any other disaster I knew before. The fight between man and machine for painful 12 minutes ended in a loss of 189 people on board. As the world was awaiting the full investigation report, the tragedy struck again. On March 10, 2019 the Ethiopian Airline Flight 302 crashed within six minutes of takeoff from Addis Ababa Bole International Airport. The striking similarity between two air crashes within six months put a huge question mark on the new generation flying machine. We will use modern science and technology to establish the facts, as we did for many past failures. However, the similar symptom and aftermath made these two recent crashes standing out as new trend we pretend to ignore. My macro-level analysis is a mere attempt to draw attention to the trend.
Software vs Life — The Tradeoff
As a software engineer by profession, I’m exposed to a bunch of problems in real world which we try to solve using software systems. Under current economic clout, we build software systems to eliminate repetitive workload from human, to gain efficiency which human can’t attain, to streamline some of the businesses for staying ahead in the competitive market, and above all, to scale up businesses across the globe. We also open new avenues with scientific innovation and technological breakthrough, but such innovation and breakthrough are not really driven by economy, rather by human nature of inquisitive observation and pushing the limit. As we build software systems, mistake is inevitable. We are living in an era where a nicely names ILOVEYOU worm sneaked to millions personal computers worldwide via a security hole in Windows operating system and caused damages in billions of dollars. An innocent typing mistake from one engineer caused a large number of service outage for hours, as it transpired in recent AWS S3 outage announcement. A seemingly small mistake can cause havoc, so we try to foresee impact of such mistakes (often termed as “blast radius”). To make such software systems work for the people, we formalize such failure analysis and attempt to quantify in terms of Availability, Reliability and Usability(ARU). Even then mistakes do take place. As a result, we lose money, user trust or even in worst scenario, entire business. But imagine such inadvertent and inevitable mistakes costing human life, would you still utter “we do make mistakes, let’s learn and move on” in the same breath? As replacing portion of cognitive load by software systems became modern trend, we introduce more complexity and fuzziness into those systems, essentially making the systems harder to test and less predictable on such failures. Since writing software for a modern-age flying machine demands more and more ARU, how much is enough? Or on a bigger context, what slice of human intelligence can be offloaded to software? Or even, do such market-driven intelligence shift worth more than human lives? There are more questions than answers.
Automation Beyond Wisdom
The madness of automation started with industrial revolution back in 18th century. The abundance of coal and then drilling of fossil fuel slowly took away the muscle power from human and domestic mammals, triggering mass production of goods and building urban jungles which would not have been possible otherwise. As we progressed and took the industrial automation to a new level, the introduction of silicon chip in the second half of 20th century started moving the automation from the brawn towards the brain. Even if it started with seemingly slow and as a great tool to assist human achieving complex computational tasks, the exponential growth of transistors in an integrated circuit (IC) over last few decades gave enormous power to computation, unleashing the opportunities to take decisions which would otherwise require human intervention. Any invention starts with some generous objectives, for example, the immense power of computation and compact storage in form of AI are being tried in unmanned fighter aircrafts to minimize human casualties in battle zone, in building autonomous rovers to hostile difficult terrains, or in making intelligent robots to achieve complex tasks in hazardous zones. But such successes in those controlled environments gave us false sense of confidence that we were closer to mimic human brain. We quickly commoditized such technologies and started applying in places where human enjoyed undisputed and absolute monopoly for tens of thousands of years, the wisdom. Can we automate human wisdom which were shaped by natural selection for millions of years? Can automation feel the danger and adopt the skill of survival on its own? Maybe it will one day, but we are nowhere near. Deploying anti-stall automation to a commercial aircraft is human stupidity fueled by their own wisdom, I believe.
Innovation & Economy — The Modern Trend
The invention and economy went hand in hand for thousands of years with every invention changing the course of economy. For example, the invention of wheel around 3500 BC led to a rapid development in transportation and now prevalent in modern civilization even after more than 5000 years of its first use. But in the last millennium has passed through the era of European colonization, not only on land with wheel-powered transportation, also by conquering the sea with ships. The faster mode of transport both on land and sea created mobility for army to conquer new lands, helped moving industrial goods to the remote corner of the world. Such rapid movement of human and industrial products destroyed local economy and culture, making way for larger economy connected through new land and sea routes, powered by faster transports and secured by technologically superior army of European. The bigger economy created bigger opportunities and steep competitions, a new kind of economy where slight edge over competitors provides huge margin, with the worst form where winner takes all. The people under the monstrous economy pushed to their limit to be more and more creative for staying ahead of others, if not for their own survival. While invention remains beyond the reach, innovation came to satisfy the ever-growing demand of the modern economy. Invention takes years of observation, study and ingenuity of human. However, it looked easy to invest in innovation to gain economic advantages and dominate the market. Technological innovations are aimed at creating demand, faster accomplishment, ease of use and cost saving. Those objectives made innovations integral part of every business organization in modern economy. The Egyptian shoemaker of last millennium BC would have acquired and passed the skill of making commoners’ sandal generation after generation, but modern shoemakers need to innovate several times during their lifetime. The slight change in shape and complex stitches of lather of different shades create new demand, industrial automation makes it faster to produce, designed for different size and shapes of toes make it easy to use and producing far away from the market at a factory with cheap labor keeps the cost low. The innovation drives economy with growth, in turn, receives more investment back. The cycle of innovation and economy quickly turns into vicious cycle when we pay the price with human lives. The larger and fuel-efficient CFM LEAP-1B engine is economic innovation aimed at saving cost. The innovation encouraged Boeing to start a new line of aircrafts, which created largest demand in the commercial aviation history subsequently. However, the cost of human lives took back seat in the trillion-dollar worth global aviation industry.
Where Is The Kill Switch
Little more than a year back, two world leaders were exchanging rhetoric while the world was watching with awe. The reason for the world to be worry of such exchange was that the rhetoric was about nuclear button. While one claimed that he had the button on his desk all the time, other claimed that his button was bigger and more powerful. They were literally talking about the Kill Switch which brings the apocalypse on Earth. Nowadays every country who weaponized nuclear energy possesses such Kill Switch, the size and power being moot points. The Kill Switch starts a chain reaction at atomic level, enormous massive amount of energy to burn life into ashes at a massive scale. While we are still debating and theorizing the creation of life on Earth, we achieved amazing feat in last century by drafting destruction theory and building tools for the same. Nevertheless, the Kill Switch in its concept not necessarily used to start chain reaction, we often use it to stop chain reaction too. I’m referring to the Kill Switch which we use in automated systems to stop machine stupidity and make way for human intelligence. Imagine you are on a jogging tour with your virtual trainer in Sierra Nevada mountains and suddenly got a feeling of breathlessness, what will you do? The big red button on the treadmill will bring you back to the real world immediately. While adaptive cruise control takes control of your car and gives you a sense of relief, you get the control back with a stroke on brake pedal. As software engineer, we practice such Kill Switch while we push significant changes to the system, such that, slightest hint of system stupidity can be undone with minimal effort. This form of Kill Switch often compensates automated system limitations with human intelligence and stepping stone for larger automation. It looks that the Kill Switch was sold in Boeing 737 Max as a feature, in a way, it was never been designed as Kill Switch. Selling AOA (Angle of Attack) Disagree Indicator and MCAS (Maneuvering Characteristic Augmentation System) Disable Switch as an optional safety feature is a worst safety standard and blind faith on automation.
Man vs Machine
Carnegie Mellon University fellows and IBM research took little more than 10 years to build a chess playing machine which could take on a grand master, they named it Deep Blue. The fight between Deep Blue and world champion Gary Kasparov in the last decade of 20th century symbolizes the man against the machine. Even if the machine employed a good old-fashioned Artificial Intelligence(AI), the scientists believed that playing chess against man was a good measurement of machine intelligence. So, when Deep Blue defeated one of the best chess players our world has ever produced, the fraternity were convinced that machine can mimic human intelligence, can surpass at certain fields too. Today the chess paying machines has grown to such extent that fight with human chess player is not attractive any more, thanks to faster CPU and Deep Neural Network based AI. The modern chess playing machine, AlphaZero claimed to have learnt chess on its own and reached superhuman level in 24-hour time. While we celebrate such a feat by machines which looks impossible for human to achieve, we ignore the fact that the game of chess is bound by a fixed set of rules with no exception. As the machine steps out of the 8x8 square board and ventures into the world full of uncertainties, the battle is still on. The tragic battle between man and machine lasted for few minutes, the seasoned pilots with more than 5000 hours of flight log were desperately trying to gain height opposing the wrongly activated MCAS system forcing a nose dive. Ironically both man and machine were fighting with same objective, to keep the aircraft from falling, but with completely opposing strategies. The battle ended in man losing against machine. The machine did not die, it was sent to correction home for a quick comeback.
Moving On
Modern flying machines built by leading manufacturers are the engineering marvels. By taking the luxury of flying above any natural or man-made structure on Earth, the giant birds celebrate many milestones achieved by fields of scientific study like physics, material science, aerodynamics, electronics, mechanical, electrical, electronics, thermodynamics and lately computer science. Over a century, the aviation industry became more and more reliable mode of transport, thanks to technological advancement and robust operating procedures laid out by International Civil Aviation Organization (ICAO). But for the first time, a trend emerged where market-driven factors like efficiency, automation became predominant over safety. The tragic end of Lion Air Flight 610 and Ethiopian Flight 302 forced the manufacturer to address the problems which would otherwise be obvious in highly reliable systems. For example, a state-of-the-art automation MCAS depending on a single AOA sensor created a single-point-of-failure (SPOF). Thus, the reliability of a complex system was brought down to the reliability of a sensor. But at the same time, today’s commercial jets are designed not to fail due to a single systemic failure. The packaging of MCAS kill switch as optional feature, lack of pilot training contributed to the disaster as well. The larger question we need to ask ourselves, will this trend continue? Are we going to redeem human lives more and more in favor of automation? Did we already wage an unstoppable war between man and machine without knowing the eventuality? Or we are slowly moving towards stratified humanity where one section will always pay for the betterment for another section? As Boeing Inc acknowledged the technical problem and working on the fix, we will possibly overlook the elephant in the room, the global economy.
Originally published at https://www.linkedin.com on April 28, 2019
