Chasing Rubber Tanks: Are Surface-Level Observations Reliable?

A blog post by GMTaC Affiliate Graduate Student Alan Zhang.

June 6, 1944 is a date remembered for Normandy, but an unprecedented military effort was at the same time on its way to Pas de Calais, a stage few miles north. It was to be the battle that decided the war, an offensive so bold that the General Patton himself led charge. Under his command were battalions of clowns, artists, and theater actors. The ghost armies pulled inflatable tanks through the battlefield, built dummy bunkers at front lines, and rolled vehicle tracks in their wake. Operation Fortitude South was one of five deception plans collectively known as Operation Bodyguard that distracted German attention away from Normandy, and it demonstrates well the relationship between observer and observed. When someone is watching, inevitably, someone else is acting. As with steps, the greater the stare, the harder the fall.

Allied forces carry an inflatable tank as part of "Operation Fortitude"
Allied forces carry an inflatable tank as part of “Operation Fortitude” (source)

Organizations have always exhibited a degree of gullibility regarding transparency. The open-hangar visibility of warehouses and manufacturing facilities were introduced as attempts by management to see their workers. Invariably, the behaviors on show were less than genuine. In Donald Roy’s classic ethnography of the machine shop in the 1950s, employees were documented to engage in coordinated quota restriction and collective goldbricking, leading supervisors to falsely believe that the output they saw was an accurate demonstration of effort. And yet, several decades later, companies have embraced aspects of the Toyota Way, a production method which prizes the “use of visual control so no problems are hidden.” More recently, the dramatic shift in office layouts to open-workplace designs reflects this persistent desire to foster transparency. In these settings too, scholars have reported opposite effects and noted systematic declines in social interaction and open dialogue.

An open-office floor plan
An open-office floor plan (source)

It does not matter where the observer perches, whether over the factory floor or above earth orbit. Setting up a constellation of satellites to image the earth surface is unlikely to change the game, despite current plans by companies like Planet to do just that. With eyes in the sky, false confidence may feel more justified than ever before. We do not notice what we do not notice. Every day people fail to notice the cell towers in their neighborhoods dressed as trees, pines, palms, water towers and church steeples. In the 60’s, hydroelectric substations across Toronto churned in neighborhoods without detection, dressed as “salt box” homes and “plain Jane” houses with the typical suburban look. For years, no one noticed, until a journalist finally did. If onlookers on the ground can be so deceived, would it be any different at higher altitudes?

A hydroelectric substation disguised as a suburban home in Toronto
A hydroelectric substation disguised as a suburban home in Toronto (source)

Assuming that remote sensing does not drive activity into subterranean depths, are surface-level observations reliable? If there are inaccuracies in what has been observed in earlier times, then caution should be taken before using that to predict future states. The constant, unceasing, voluminous deluge of visual imagery being collected by satellites today necessitates computer-assisted observations, but how good is computer vision if the training set is compromised? In fact, it seems that subterfuge will become easier when observation becomes nothing more than pattern recognition between past and present. Past deception reinforces present deception, and the illusion persists. What do we do then about the farmers in Italy who plant umbrellas that resemble olive trees to earn government subsidies? How will we distinguish between the patterns of truck activity that predict productivity and growth from those that drive around empty freights as decoy? In short, as organizations rely more heavily on algorithms to see and machines for vision, will transparency lead us into blindness, chasing rubber tanks and admiring the flora of cell-towers?