In this episode, former FBI agent and newly discovered Chinese spy Colby Granger escapes from US custody. Charlie helps track him down by using a set covering deployment algorithm. During the pursuit, Colby contacts Don and claims to be a triple agent: he is trying to find a mole in the US government who is actually spying for the Chinese. Charlie develops a trust metric to help the team decide whether or not to believe Colby.
Charlie gives an excellent description of the set covering deployment problem in this episode. In order to avoid shipwrecks, we need to illuminate a rocky coastline with light houses. Of course, each light house is expensive to build, so we would like to cover the entire shoreline with the fewest number of light houses. This is what is known as an optimization problem. We want to optimize the shoreline lighting coverage-to-cost ratio.
One way to solve such a problem is by following a greedy algorithm. That is, we could put our first light house where it can illuminate the largest section of coastline, our second lighthouse will go where it can illuminate the largest section of coastline that isn't already illuminated by the first light house. We continue in this manner, placing each lighthouse so that it illuminates the largest amount of shoreline not illuminated by previous lighthouses, until the whole shoreline is illuminated. The theory behind this is that if we maximize the coverage-to-cost ratio at each step, we will maximize it for the whole process.
After working with Colby for two years, having him confess to being a Chinese spy, and then having him claim to actually be working for the United States, Don and the gang are a having trouble deciding what to believe. Charlie, grappling with his own inability decipher the truth, tries to construct a trust metric- a measure of how much Colby is trusted by the rest of the group.
The difficulties in creating such a measure are the same ones facing anybody trying to model any subjective quality: defining the essential characteristics of that quality and finding a way to measure them.
Let us apply this to a far less serious problem than determining if we can trust someone. We're going to try to determine who among Magic Johnson, Larry Bird, and Michael Jordan is the best basketball player. We need to think of some essentials for what makes a basketball player great. Of course, there are many measures of performance on the basketball court, so we will pick four important ones:
Now, we are going to consider all four of these statistics evenly. That is, we will consider each one to be just as important as all the others. We also need to set a baseline for each category. We have listed the maximum for each of our four categories attained by any player, ever, which will be our baselines:
Here are the career statistics for our three players:
| Scoring Avg. | Assist Avg. | Rebound Avg. | Championships Won | |
| Magic Johnson | 19.5 | 11.2 | 7.2 | 5 |
| Larry Bird | 24.3 | 6.3 | 10.0 | 3 |
| Michael Jordan | 30.1 | 5.3 | 6.2 | 6 |
Now, we're going to divide by the best possible in each category to get a number between 0 and 1 telling us how close each player is to being the best ever in a given category:
| Scoring Avg. | Assist Avg. | Rebound Avg. | Championships Won | |
| Magic Johnson | 0.65 | 1 | 0.31 | 0.45 |
| Larry Bird | .81 | .56 | .44 | .27 |
| Michael Jordan | 1 | .47 | .27 | .55 |
Taking the average of each row will give us our answer: