Center

Center had not always been a middle-finder. They had grown up, by their own account, in a colony where everyone had an opinion about what the typical day looked like and almost no one bothered to define typical first. As a small bee, Center had once spent an entire summer afternoon listening to four older bees argue about the typical pollen yield. One bee insisted on the highest number anyone had ever seen, which was a record set by a great-aunt during a memorable bloom. Another insisted on the lowest. The third averaged the colony's last week and produced a number nobody had actually gotten. The fourth pointed out that more than half of the colony's yields fell within a narrow window that none of the other three numbers came close to. Center, listening, realized that all four bees were correct and all four were also wrong, depending on which question you were really asking. The arguers were not arguing about pollen at all. They were arguing about what to mean by typical. That afternoon, Center decided that being a middle-finder was a real job, one that needed doing, and that nobody in the colony was doing it. They asked the colony elders for permission to study the matter. The elders, who had been losing the same argument for years, agreed.

By the time the workshop's young statisticians met Center, they had been at the academy for nearly three weeks and had begun to notice that real questions were almost never as simple as they sounded. Center hovered just inside the doorway and announced themselves with the kind of formal politeness that some characters use when they are trying not to take up more space than they need. "A perfect problem," they buzzed, peering at the table where the kids had been arguing. The kids had spent the previous evening playing a treasure game, and Leo had found a mega-chest worth one hundred and fifty Glimmer-Gems. The other three players had piles of seven, eight, and ten. The argument was about what a typical player got. Leo, citing the average, claimed thirty-something. Maya, citing her own pile, claimed eight. Sam had stopped trying to decide and was eating a cookie. The kids had been told, around kit two, that Lessons-layer characters arrived when a problem had bent itself into the right shape. Apparently this problem had. Center introduced themselves, settled into a low hover above the table, and clicked their device twice. The device purred, which the kids would later learn meant that the data had passed Center's first inspection: real, finite, and worth thinking about. "You're asking what's typical?" Center said. "That is three questions in a costume. Let us pull off the costume."

The kids wrote their four numbers on cards and laid the cards on the floor. Center began with the mean, because most people did. The mean, Center explained, was the arithmetic average: you summed every value, then divided by the count. Sum was one hundred and seventy-five. Count was four. The mean was forty-three point seven five. Leo crowed. Maya crossed her arms with the precision of someone who had been rehearsing the gesture. Center waited, because Center always waited at this moment, and let the kids notice the wrongness on their own. The wrongness, when it arrived, arrived as a feeling. The average of forty-three did not describe any of them. It was, Center said, the right calculation answering a question that wasn't quite the question. The one big number — Leo's mega-chest — was an outlier, and the mean was famously sensitive to outliers. A single extreme value could drag the average so far from the rest of the data that the average no longer described the data at all. The median, Center continued, did not have this weakness. They had the kids sort their numbers smallest-to-largest — seven, eight, ten, one hundred fifty — and look at the middle. With four numbers, the middle was the average of the two middle ones: nine. Maya, Sam, and the narrator nodded; nine felt right. Nine described their actual piles. The median, Center said, was the honest middle when the data was skewed. The mode was the third tool, the most frequent value. None of their four numbers repeated, so this set had no mode at all. Center wrote a new set of cards — twelve, fourteen, nine, fourteen, eleven — and asked which value showed up most. Fourteen, the kids said. Fourteen, Center confirmed. The mode was the answer when you wanted the most common choice: the popular T-shirt size, the bestselling ice-cream flavor. You would not, Center pointed out, want the average flavor of ice cream. The image was so disagreeable that the kids laughed.