Stotlebots.

This is not Lenten, but since it's already mostly written I thought I'd go ahead and post it. Excerpt from a paper I just turned in:

Consider the case of extremely advanced androids whose behavior is entirely determined by circuitry, yet whose behavior appears outwardly to be very much like our own. These androids have built-in subroutines that direct them to perform necessary and complicated maintenance, such as collecting energy and disposing of waste. These subroutines are analogous to our own bodily desires. They also have advanced defense routines, so that when they are in danger they alter their behavior in order to survive the situation. The defense routines can even be activated when nearby androids are in danger. These routines correspond to our own spiritedness and our biological fight-or-flight responses. Finally, the androids have a complicated unit that handles their long-term task management. The android has a database of long-term goals. Progress toward these goals proceeds according to a complicated rubric; no single goal takes precedence at all times. When the android has sufficient energy to do so and is not in danger, the long-term deciding-unit selects the top goal at the moment. Then it uses a randomizer based on something very subtle, like variations in atmospheric static, to generate a list of possible actions to achieve that goal. It runs an evolutionary algorithm through several generations, optimizing at each stage and generating several new options that are similar to the best option of the previous stage. What is important here is that the android can override some of its basic maintenance and defense routines, and even goals which are subsidiary at the moment, if its long-term planner decides that such an option would be the best way to achieve the long-term goal that is currently prioritized.

The behavior of such an android is completely determined by its physical capabilities, its programming, and the output of the randomizer that its long-term planner depends on. But, if the randomizer is good enough and the long-term planner can generate diverse options, the behavior of the android will be impossible to predict. We know that its behavior is determined by physical states—but knowing the input to the randomizer doesn’t tell us anything about the action that the robot ends up performing. That is to say, we know that the android’s behavior is ultimately caused by its physical state, but in describing its unexpected behavior it only makes sense to trace the cause back to the long-term planning unit. If we say the robot is responsible for an action, we can mean in a technical sense that the action came from the long-term planning unit.

This gave me an idea for a cartoon series called 'Stotlebots. The first series will be about the battle between the Aretebots, led by Sokraton Prime and his lieutenant Plator, and the Sophisticons, led by Gorgiastron. They battle it out in scratchy 80s anime, and later in sharper 90s animation. Then, after a while, we'll switch to CGI for the next series, in which the Essentiabots (including Aquinox and Dunobot Scotus) and and the Nominacons (Megoccamtron's team) crash-land on an abandoned planet (which turns out to be Earth, far in the past) and duke it out for a few seasons. Then the CGI gets smoother, and the Essentiabots finally go back to their homeworld, where everything changes.

In short: my new fake life mission is to reproduce the work of Alasdair MacIntyre via the genre of giant-fighting-robot Saturday-morning cartoons. The massive revenue I will get from merchandising, especially with awesome action figures, is just a side bonus.