September 2016
Recent data revealed that Americans drove just over three trillion miles in 2015, an increase of 3.5% over the prior year. Disturbingly, deaths due to highway accidents rose 7.2% over the same period1. Paralleling this news is heavy experimentation with autonomous/driverless cars that may change transportation as we know it. As we explore this subject, we learn that driverless cars—like many innovations, including those in finance—offer benefits as well as risks in the effort to protect humans from our bad behavior.
The Long and Winding Road
Starting in reverse, imagine a study conducted in 1920 wherein the pluses and minuses of the newfangled automobile were projected for the next century. A visionary transportation analyst back then would project the benefits of one day moving around at upwards of 70 mph so we could transport goods, visit friends, commute further to work and take family vacations galore—with no more silly horse mess to mess with.
She may also point out that we would ultimately pave an area equivalent to the size of Illinois. We would become dependent upon oil purchased from despots, entangling us in Middle East affairs. Finally, around 3.5 million Americans will die on our highways and many times that will be injured over the century ahead. With that foresight, would we have concluded that the automobile was a good idea?
Even when posed this way, the answer is still “yes.” The American spirit would have been willing to bet that progress was well worth it and future innovation could counteract the negatives; we just wouldn’t have needed to experience the downsides before working on the solutions (think better cars, tires, roads, seatbelts, airbags, and reduced impaired driving).
Get Back
So where do self-driving cars enter this equation? For one, we are reminded that looking back does not always adequately prepare us for the future—for transportation or financial markets. Part of the beauty of intelligent software is the ability to learn from the past and steadily improve functionality. What happens though, when software hasn’t yet learned to respond to cars with different decision-making algorithms or manual drivers who have adjusted their habits assuming cars around them are “smart?” The past cannot inform software engineers of everything that must be accounted for, and that can lead to disastrous outcomes for some unfortunate folks in the interim.
Financial markets traversed a similar path in the 2000s with the growth of “innovative” derivatives promising to more efficiently spread risks, thereby making our financial system appear safer. Even as Alan Greenspan touted these advances, he presciently warned in 2005, the “science (of derivatives) is based on the past behavior of markets, which is not an infallible guide to the future2.” As we all too painfully learned, this “efficient science” of derivatives ended up bringing the global financial system to the brink of collapse by 2008.
Don’t Let Me Down
Though we do not dispute that technology can enable vast improvements over the “old way” of doing things, and we suspect autonomous cars will eventually do the same, human judgment is required at times. Google, one company developing autonomous cars, recently had a car hit a bus after swerving to miss some sandbags. Though intuitive for an undistracted human, it may be a long time before software can judge, in all conceivable scenarios, whether to swerve and miss something innocuous at the risk of hitting another object where more lives would be at risk.
The “flash crash” of 2010 reminded financial markets that human judgment is still needed to avoid “swerving” at the wrong time. In that episode, a single large trade was entered, and as a result of a few “over-corrections” by computer trading algorithms, some large companies like Procter & Gamble traded as low as $0.01 while the Dow dropped over 600 points in five minutes3. As a greater percentage of assets are allocated to ETFs and traded instantaneously with computer algorithms, we expect episodes akin to the “flash crash” to repeat.
I Want to Hold Your Hand
Another impact of evolving technology is that it affects who is impacted by an activity. Self-driving cars can effectively address some fatal accidents caused by poor decisions (i.e. driving while impaired, texting, speeding, running red lights, etc.). However, people that end up dying in autonomous vehicles may very well be victims of bad software or “fluke” circumstance. Last June, Tesla’s experimental “Autopilot” feature killed its driver because “neither Autopilot, nor the driver noticed the white side of the tractor trailer against a brightly lit sky, so the brake was not applied4.” Over time, automobile-related fatalities may decrease overall with new technology, but a different set of victims may be impacted by some unfortunate miscalculations rather than their own driving decisions.
Global central bankers’ innovative monetary policy prescriptions in recent years, you might say, are aimed at “taking over the financial controls” with varying consequences for the public. They determined that, to the detriment of savers, those that had borrowed excessively needed to be rescued by low interest rates so debt could be refinanced. Banks were saved that arguably didn’t need to be rewarded for their bad decisions. Much good came from their actions too, so our point is not to debate the merits of their intervention but rather to remind that the composition of who was most affected was impacted. When the current equilibrium of low interest rates, slow but steady growth, and low inflation is interrupted by an unwinding of these measures, our worry is that many innocent people could experience some dramatic side effects from another crisis of some sort.
We Can Work it Out
At Hamilton Point we are watching the developments regarding self-driving cars with great interest. We take to heart the human insight offered by the fact that our society continues to “lean forward” with innovation in trying to solve large-scale problems. However, we’re also reminded that we wouldn’t have 35,000 people die each year in car crashes if we just drove slower and stayed off our phones. Even with innovation, it is our belief that driving, like investing, will remain both art and science, where sound judgment, complementing scientific progress, will remain integral components of the human story.
1) 2015 Motor Vehicle Crashes: Overview, National Highway Traffic Safety Administration, August, 2016.
2) Alan Greenspan Remarks to The Federal Reserve Bank of Chicago’s Annual Conference. May 5, 2005.
3) Lauricella, Tom, Kara Scannell and Jenny Strasburg, How a Trading Algorithm Went Awry, Wall Street Journal, October 2, 2010.
4) Tesla Press Release, “A Tragic Loss,” June 30, 2016.