Here's a letter to the editor in The Wall Street Journal -- in the year 2025:
Recent major accidents in driverless high-speed trucking have pummeled driver confidence. The Aug. 1 We Haul Trucking Co. software glitch that caused trucks to crash while making hairpin turns at 250 mph is merely the latest in a string of accidents. Serious errors have plagued FastTruckers' handling of their trucks, the National Association of Super Driverless Automated Highways' driving of BookFace's trucks, and driving throughout Spain.
Clearly, the industry has reached a tipping point. Given the complexity involved, no amount of testing will guarantee that today's driverless systems operate flawlessly. Errors are inevitable, and it's time the industry treated them that way. We must recognize system risk -- the potential for losses due to system errors -- as a foreseeable but manageable threat, on par with ice and pothole risks and deserving of similar analysis and protections.
As an industry, we act surprised every time a traffic-stopping accident occurs. We shouldn't. Driverless systems are as hard to build as aerospace systems. This challenge is compounded by the increasing complexity of the network of participants and the regulations governing them. Intense competitive and regulatory pressures propel everyone involved to develop faster and faster software without enough time for NASA-caliber testing that spans decades.
National Highway Traffic Safety Administration Chairwoman Maria Shapearoh has indicated that demands to abandon high-speed trucking are unrealistic. That is welcome news. The DOT instead appears intent on pursuing steps to reduce errors and limit their impact. Discussions have long focused largely on testing, lane-change controls, and accelerator governors -- worthy measures, but insufficient. More recently there have been reports of a DOT proposal for regulations requiring executives to certify their software. But no amount of testing or effort will ever completely ensure error-free software.
What is needed? Sweeping trucking change. Everyone involved must recognize driverless system risk as part of modern driverless trucking and take steps to manage it.
This will require a new breed of risk software that can quantify the risk of potential errors, ensure sufficient insurance reserves, identify conditions that warrant trucking halts, and act decisively to halt trucking when conditions warrant it.
Recent glitches have shown that it can take just minutes to end thousands of lives and bring trucking firms to the verge of collapse. We Haul Trucking Co. simply has the misfortune of being the latest. There will be more such fiascoes until the industry elevates its attention to system risk. Driverless trucks are here to stay, and so are errors.
-- Sam Walmoe is a former driverless system programmer and the CEO of Insanely Fast Driving Software.
The above letter to the editor would never be written, as the nation would be outraged at the loss of lives and would never be willing to accept "system risk" from driverless vehicles.
Nonetheless, driverless vehicles will likely be a part of our lives within the next 50 years. These vehicles will be controlled by complex algorithms and sensors -- much like trading.
We've learned through the years that driving at high speeds is dangerous. But we have not yet learned that lesson in trading. While trading crashes are not as dire as car crashes, we could apply something we learned in driver safety to market traffic: Speed limits help prevent traffic accidents and could help prevent trading accidents.
Others have suggested we need an "air traffic control" for trading or foolproof financial technology. What we really need is a national speed limit for trading.
Trades are now executed in less than 200 millionths of a second, and exchanges compete to cut order times by microseconds. The SEC wants to spend billions of dollars to track every trade so it can figure out what went wrong after the fact. Instead of reacting to trading disasters -- e.g., Knight Capital's
Electronic trading is here to stay, and so are errors. High-speed trading doesn't have to stay.
