Reamde

Now, in all the entire (real) world’s airports put together, the number of people who got into concourses by walking the wrong way through exit doors amounted to maybe one or two per year: not enough to hold the attention, or assure the vigilance, of even the most rabid T’Rain player. So the APPIS system now sweetened the pot by automatically generating fictitious, virtual wrong-way goblins and sending them up that tunnel at the rate of one every couple of minutes, every day, forever. Some balancing had to happen—the value of the rewards had to be tweaked relative to the frequency of wrong-way goblins—but with a minimal amount of adjustment they were able to set the system up in such a way that 100 percent of all the wrong-way goblins were apprehended. The total number of wrong-way goblins that had to be generated per year was about two hundred thousand—which was no problem, since generating them was free. The trick, of course, was that a tiny minority of those one-way goblins were not, in fact, computer-generated figments. They were representations of actual human forms that had been picked up by airport security cameras as they walked the wrong way into airport concourses. In reality, of course, this happened so rarely that testing the system was well-nigh impossible, and so they ran drills, several times a day, in which uniformed, badged TSA employees would present themselves at the exit and show credentials to the bored guard and then walk upstream into the concourse. In exactly 100 percent of all such cases, some T’Rain player, somewhere in the world (almost always a gold farmer in China) would instantly raise the Horn of Vigilance to his virtual lips and blow a mighty blast and rush out to confront the corresponding one-way goblin: an event that, through some artful cross-wiring between Corporation 9592’s servers and the airport security systems, would cause red lights to flash and horns to sound and doors to automatically lock at the airport in question.

Corvallis and most of the other techies hated this idea because of its sheer bogosity, which was screamingly obvious to any person of technical acumen who thought about it for more than a few seconds. If their pattern-recognition software could identify the moving travelers and vectorize their body positions well enough to translate their movements into T’Rain, then it could just as easily notice, automatically, with no human intervention, when one of those figures was walking the wrong way and sound the alarm. There was no need at all to have human players in the loop. They should just spin out the pattern-recognition part of it as a separate business.

Richard understood and acknowledged all of this—and did not care. “Did you, or did you not, tell me that this was all marketing? What part of your own statement did you not understand?” The purpose of the exercise was not really to build a rational, efficient airport security system. It was, rather (to use yet another of those portentous phrases cribbed from the math world), an existence proof. Once it was up and running, they could point to it and to its 100 percent success rate as vindicating the premise of APPIS, which was that real-world problems—especially problems that were difficult to solve because of hard-wired deficiencies of the human neurological system, such as the tendency to become bored when given a terrible job—could be tackled by metaphrasing them into Medieval Armed Combat scenarios, and then (here brandishing two searingly hip terms from high tech) putting them out on the cloud so that they could be crowdsourced.