Ultralearning. Scott H. Young

Чтение книги онлайн.

СКАЧАТЬ won $77,000—the highest single-day winnings in Jeopardy! history at the time. Craig’s victory wasn’t a fluke. He broke records again, amassing nearly two hundred thousand dollars, the highest ever five-game winning streak. Such a feat would be remarkable on its own, but what was more incredible was how he did it. Reflecting on the moment, Craig says, “My first thought wasn’t ‘Wow, I just won seventy-seven thousand dollars.’ It was ‘Whoa, my site really worked.’”

      How do you study for a test that can ask any question? That was the essential problem Craig faced as he prepared to compete. Jeopardy! is famous for stumping home audiences with trivia questions that can ask about anything from Danish kings to Damocles. Thus the great champions of Jeopardy! tend to be brainy know-it-alls who have spent a lifetime amassing the huge library of factual knowledge needed to spit out answers on any topic. Studying for Jeopardy! might feel like an impossible task, as you would need to study almost every conceivable subject. Craig’s solution, however, was to rethink the process of acquiring knowledge itself. To do that, he built a website.

      “Everybody that wants to succeed at a game is going to practice the game,” Craig contends. “You can practice haphazardly, or you can practice efficiently.” To amass the wide-ranging trivia needed to break records, he decided to be ruthlessly analytical about how he acquired knowledge. A computer scientist by trade, he decided to start off by downloading the tens of thousands of questions and answers from every Jeopardy! game ever aired. He tested himself on those during his free time for months, and then, as it became clear that he was going to go on television, he switched to aggressively quizzing himself on the questions full-time. He then applied text-mining software to categorize the questions into different topics, such as art history, fashion, and science. He used data visualization to map out his strengths and weaknesses. The text-mining software separated the different topics, which he visualized as different circles. The position of any given circle on his graph showed how good he was in that topic—higher meant he knew more about that topic. The size of the circle indicated how frequent that topic was. Bigger circles were more common and thus better choices for further study. Beneath the variety and randomness in the show, he started to uncover hidden patterns. Some clues in the show are “Daily Doubles,” which allow a contestant to double his or her score, or lose it all. These extremely valuable clues may seem randomly placed, but having the entire Jeopardy! archives at his fingertips, Craig found that their position followed trends. One could hunt out the valuable Doubles by hopping between categories and focusing on high-point clues, breaking the conventional approach to the show of sticking within a single category until it was completed.

      Craig also found trends within the types of questions asked. Although Jeopardy! could conceivably ask questions on any topic, the format of the game is designed to entertain a home audience, not to challenge competitors. Following this reasoning, Craig found that he could get away with studying the best-known trivia within a category, rather than digging deep into any particular direction. If a subject was specialized, he knew the answers would be geared toward the best-known examples. By analyzing his own weakness on archival questions, he could see which topics he needed to study more to be competitive. For example, he found that he was weak on fashion and focused on studying that topic more deeply.

      Using analytics to figure out what to study was only the first step. From there, Craig employed spaced-repetition software to maximize his efficiency. Spaced-repetition software is an advanced flash card algorithm first developed by the Polish researcher Piotr Woźniak in the 1980s. Woźniak’s algorithm was designed to optimally time when you need to review material in order to remember it. Given a large database of facts, most people will forget what they learn first, needing to remind themselves of it again and again for it to stick. The algorithm fixes this problem by calculating the optimal time for reviewing each fact so you don’t waste energy overdrilling the same information, but also so you don’t forget what you’ve already learned. This tool allowed Craig to efficiently memorize the thousands of facts he would need for his later victory.

      Although the show airs only one episode per day, Jeopardy! tapes five episodes at a time. Craig was coming back to his hotel room after winning five games straight, and he couldn’t sleep. He said, “You can simulate the game, but you can’t simulate winning two hundred thousand dollars in five hours and setting the single-day record on a game show you’ve wanted to be on since you were twelve years old.” Combining unorthodox tactics and aggressive analysis, he had gamed the game show and won.

      Roger Craig wasn’t the only person I found who had seen his fortunes change as a result of aggressive self-education. I didn’t know it at the time, but in 2011, the same year my MIT Challenge would begin, Eric Barone was starting his own obsession. Unlike mine, however, his efforts would extend for nearly five years and require mastering many completely different skills.

      FROM MINIMUM WAGE TO MILLIONAIRE

      Eric Barone had just graduated from the University of Washington Tacoma with a degree in computer science when he thought, Now’s my chance. He had decided that he wanted to make his own video games and that now, before he got comfortable in a salaried programming job, was his opportunity to do something about it. He already had his inspiration. He wanted his game to pay homage to Harvest Moon, a charming Japanese series of games in which the player must build a successful farm: grow crops, raise animals, explore the countryside, and form relationships with other villagers. “I loved that game,” he said about his childhood experience with the title. “But it could have been so much better.” He knew that if he didn’t follow through with his own vision, that improved version would never be a reality.

      Developing a commercially successful video game isn’t easy. AAA game companies budget hundreds of millions of dollars and employ thousands of people on their top titles. The talent required is similarly broad. Game development requires programming, visual art, musical composition, story writing, game design, and dozens more skills, depending on the genre and style of game developed. The breadth of skills required makes game development much harder for smaller teams than other art forms such as music, writing, or visual arts. Even highly talented independent game developers generally have to collaborate with a few people to span all the skills required. Eric Barone, however, decided to work on his game entirely alone.

      Deciding to work alone came from a personal commitment to his vision and an indefatigable self-confidence that he could finish the game. “I like to have complete control over my own vision,” he explained, saying that it might have been “impossible to find people who were on the same page” regarding the design. However, that choice meant that he would need to become proficient in game programming, music composition, pixel art, sound design, and story writing. More than just a game design project, Barone’s odyssey would entail mastering each aspect of game design itself.

      Pixel art was Barone’s biggest weakness. This style of art harkens back to the earlier era of video games when rendering graphics was difficult to do on slow computers. Pixel art is not done with fluid lines or photorealistic textures. Instead, a compelling image must be created by placing pixels, the colored dots that make up computer graphics, one at a time—painstaking and difficult work. A pixel artist must convey movement, emotion, and life from a grid of colored squares. Barone liked to doodle and draw, but that didn’t prepare him for the difficulty. He had to learn this skill “completely from scratch.” Getting his art skills to a commercial level wasn’t easy. “I must have done most of the artwork three to five times over,” he said. “For the character portraits, I did those at least ten times.”

      Barone’s strategy was simple but effective. He practiced by working directly on the graphics he wanted to use in his game. He critiqued his own work and compared it to art he admired. “I tried to break it down scientifically,” he explained. “I would ask myself, ‘Why do I like this? Why don’t I like that?’” when looking at other artists’ work. He supplemented his own practice by reading about pixel art theory and finding tutorials that could fill gaps in his knowledge. When he encountered a difficulty in his art, СКАЧАТЬ