Thinking Digitally

Is technology an aid to our thinking processes, or is it a distraction? My own conclusion is “yes.”

By connecting us with information, we have the opportunity to make more informed decisions. By taking on the routine tasks such as adding numbers or performing complex calculations, we are freed to think about the implications of such results rather than carrying out the drudgery of arithmetic. Remember doing long division by hand with three digit numbers? Beyond calculation, technology opens the door to finding information. It is hard to think about doing research without Google. Steven Johnson makes a great case for the role of technology in innovation in his wonderful video. The benefits of technology are obvious.

Unfortunately, the other side, where technology interferes with our thinking or keeps us from thinking, is too often overlooked. Most of the examples I offer can be simply summarized: We accept and believe the results from technology, while in many subtle and not so subtle ways, we should ask questions instead. The problem is, many have not been trained to ask the questions.

If you told a person (well, most people) that two years ago you heard the Grand Canyon was 2 billion years old, you would conclude, assuming the source was accurate, that today the Grand Canyon is still about 2 billion years old. A computer would conclude the Grand Canyon today is 2,000,000,002 years old. Similarly, when told that a circle has a diameter of about three inches, and then asked the circumference of the circle, a reasonably thoughtful person would say it is about nine inches. Again, a computer would determine it to be 9.424777960769379 inches, or something like that, depending on its internal representation of numbers! In the digital age we give the time as 9:27, while in the analog era we would say it is almost 9:30. Technology presumes a level of accuracy in the data that is often not warranted.

The problem is amplified when an error is made in entering data. A person walks into a fast-food restaurant and orders a meal. The total comes to $5.26, and the person hands the clerk a $10 bill. The clerk returns $14.74 in change. Obviously the clerk entered the payment as $20 rather than $10, but has no recognition that the change is greater than the payment. Human error is a reality whether in an analog or digital age. The problem in the digital age is an over-reliance on the validity of the results coming from the technology.

In a potentially more serious example of this overreliance, a woman drove her rental car into a lake following the directions on her GPS. “I don’t know why they wouldn’t question driving into a puddle that doesn’t seem to end,” [Bellevue fireman] Lt. Eric Keenan said. Basically this is the same as the fast-food story with more challenging results, though the women in the car were able to walk away.

For many problems, it is more than one number being in error or improperly approximated. Rather, the challenge comes in responding to information created by calculations using accurate and inaccurate data. A tragic result came several years ago when pilots had to deal with erroneous calculation results from the onboard computer system:

“Air France Flight 447 was en route from Rio de Janiero to Paris on May 31, 2009, for an overnight trip, when it vanished. There was no warning before the flight crashed into the Atlantic Ocean in the early morning hours of June 1, 2009 — nearly four hours after take-off. The 2009 crash, which killed all 228 passengers and crew on board, is considered one of the worst ― and most mysterious ― aviation disasters in modern history,” according to an ABC News report.

More recent analysis of what happened can be found in the ABC Nightline Report by Matt Hosford, Lauren Effron, and Nikki Battiste, July 5, 2012:

“‘[The pilots] seemed to have trouble looking past the automation they were accustomed to and not really able to continue with the old raw information that pilots used to depend on,’ he said. ‘Clearly the report shows that there was a lot of difficult communication on the flight deck, a lot of incomplete thoughts, a lot of confusion.’

“According to the report, a speed sensor on board the plane, called a pitot tube, stopped functioning after becoming clogged with ice at high-altitude while the plane was flying through a thunderstorm. This caused the auto-pilot to disengage and shift the controls back to the pilots. While flying in heavy turbulence, the pilots failed to properly diagnose the severity of the problem because the pitot tube, a critical piece of equipment to the aircraft, was sending inaccurate data to the cockpit, the report said. The pilots put the plane into a devastating stall and it fell rapidly from the sky, before pancake-ing into the ocean.

“’Despite these persistent symptoms, the crew never understood that they were stalling and consequently never applied a recovery maneuver,’ the report said.’

Most business cases are not nearly so dramatic or tragic as this one, but they have huge economic consequences as well. Again, the problems occur when approximate numbers, some of them incorrect for a variety of reasons, are combined to provide insight and efficiency in the solution of complex problems. I offer two examples in the manufacturing world.

Lean production, with “just in time” delivery of product to the manufacturing floor allows for great cost reduction and speed of assembly. In the case of an automobile manufacturer, for example, rather than receiving a shipment of tires in the warehouse, the company receives only the tires needed in the next hour delivered directly to the assembly line. The result is lower costs (no warehouses, no inventory management, no delivery costs from the warehouse to the assembly line) for the auto company. A digital model, including the times of delivery of the tires to the factory, correlated with the factory production schedule, will show that everything works fine. Often heavy traffic delays the delivery truck bringing tires to the assembly area, backing up production. Other times, the production line has delays and more tires are delivered than are needed. The times for delivery are not as precise as they are estimated.

In practice, suppliers often solve this problem by creating warehouses of their own near the assembly plant. Several years ago, I visited a Wal-Mart distribution center in Arkansas. They bring in the products needed for each of the regional stores at one end of the distribution center, while at the other end these products, automatically sorted on switching conveyor belts, are loaded on trucks for the daily delivery to these regional stores. To deal with traffic and variable demand, however, the suppliers each have to buffer their goods with a warehouse of their own. Driving to the Wal-Mart distribution center, we saw warehouses for all major suppliers and lines of trucks waiting to deliver to the distribution center. There is lots of variability in this problem, including the quantities needed in the daily orders from the stores, the timing of delivery to the distribution center, as well as delays in the manufacturing process of the goods from the suppliers. The numbers may look great from the technology models, but the reality requires many workarounds, often very costly for the suppliers.

The same problem, on steroids, comes from outsourcing the manufacture of complex parts of a design. In addition to the variability in demand, delivery, and manufacturing, we can add two more issues: time for the design of the parts and the fit (i.e., size, electrical connections, and the sometimes subtle interaction between the part and the rest of the product) in the final product. Designs can be delayed months or years in the case of a new engine for an automobile or an airplane, for example. When the supplier has a complex supply chain itself, the challenges are compounded.

Technology can provide us with dramatically improved productivity and efficiency, but if it is not used wisely, it can also be at the root of dramatic errors, cost overruns, and delays. Accounting for approximations, thinking through alternatives, and generally managing risk is a vital part of any process. Unfortunately in this era where speed is so essential, that thinking is usually put off until the problems happen and then it is much more costly to fix. To evaluate risk, the kind of thinking required includes estimating, approximating, and anticipating. I believe one of the places I learned these skills is trying to do long division on three-digit numbers, a skill completely ignored when the numbers are punched into the calculator.

Our lack of skills combined with the pressures of speed and efficiency make up a lethal combination leading to many dramatic, costly, and sometimes deadly failures.

Both education and business need a reset for the digital age!

Al Erisman is executive editor of Ethix, which he co-founded in 1998.
He spent 32 years at The Boeing Company, the last 11 as director of technology.
He was selected as a senior technical fellow of The Boeing Company in 1990,
and received his Ph.D. in applied mathematics from Iowa State University.