The Lottery - Why People Favor Worse Odds - Extra Credits

In October, perhaps the greatest experiment in player psychology to date occurred. It happened very quietly, without much notice being taken of it by the video game community, because it didn't happen in our realm. It happened in the world of lotto. As James writes this, the first great result of this experiment is about to play out. We may do a follow-up episode to this one after we see the results, but today I want to talk about the thinking and the player psychology that drove the experiment, rather than talk about its eventual validity. But, for any of you who want to know how this all turns out, since this episode will come out long after we reap our first major data from the experiment, simply look up the Powerball results in the first week of January 2016. So, what experiment am I talking about? I am referring to the great odds change in American lotto, specifically Powerball. You see, back in October of 2015, the Multi-State Lottery Commission, the folks who run what is essentially the largest lottery in the United States, changed the odds of winning from 1 in 175 million to an

astronomical 1 in 292 million. Not because they weren't making enough money at the current odds, but because, and here's where it gets interesting, they believed more people would play if they made the odds worse. Now, to understand the reasoning here, and why the data they have supported that theory, we have to talk a little bit about how Powerball works. Powerball is a simple lotto where you pick five numbers, and then a sixth bonus number. And if all six of your numbers match, you win the jackpot. But, here's the important part. If no one wins, the jackpot is carried over, and some of the money players spent on the game is rolled back into the next jackpot. So, as long as there's no winner, the jackpot continues to grow. By crashing the odds of someone actually hitting the jackpot, the lottery commission basically ensured that the jackpots would grow to higher and higher numbers. And here's the thing that really drove them. They saw that the higher the jackpot was, the more people would decide to play. And since higher jackpots lead to more players, jackpots

should follow an exponential curve where as the jackpots rise, more players come in, putting more money into the system, and creating even bigger jumps in the jackpot than previously seen. But this led to a conundrum because once you hit a certain number of players, you're statistically almost guaranteed to have at least one winner. And when somebody wins the jackpot, the jackpot resets to its baseline starting amount, meaning that in the past, just when Powerball were starting to really astronomical part of the exponential curve of player growth, somebody would win, and interest would die down, and the process would start over. So, their solution was simple: tank the odds. If you drop the odds of winning substantially, you push the statistical near certainty of a winner a few weeks down the line, thereby letting you really rake in the benefits of your exponential player curve. And therein lies the great question. In a system like this, at what point does lowering the odds actually discourage players? Is there such a point above zero? So long as there's at least one winner, so people can think that it might be them, does it matter

how irrational playing the game is? And that's a deeply important question for game design because very often we like to treat players, especially in incentive design, as rational actors. If I give you X experience for killing monster A and X plus five experience for killing monster B, and they're both the same difficulty, I expect the player to gravitate toward killing monster B. And if I give players, let's say 100 experience for beating monster C and 130 experience for beating monster D, but monster D is three times as difficult to beat as monster C, I expect players to favor the efficient route and fight monster C, which is why the Destiny loot cave happened. Although there will be a few outliers who want to take on monster D just for the challenge, which in video games becomes its own sort of reward, so we can discount those outliers. But what this lotto experiment is saying is that that might not actually be the correct calculus. So long as people think they have a chance at winning, they actually might opt to take a 100 to one shot on getting 50 experience rather than a guaranteed one experience win,

regardless of how interesting or banal the challenge might be. They just need to know about the 50 experience payout. And while you can argue that video games rarely come down to these sort of strict odds calculations, the underlying principle remains. It's the idea of the job versus the big score. This is especially prevalent in MMOs, but it's relevant in all incentive-based games. And the suggestion here is that we should be making more encounters that feel like a big score, even if they're way worse to grind as opposed to the slow but steady job-like mob and quest systems that many games use today. But there's one other factor to consider when thinking about what the Powerball experiment might mean for games. And no, it's not the extrinsic real-world financial reward of Powerball, but rather what it represents. When people hear about winning 400 or 500 or 800 million dollars, when they hear about that long elusive billion-dollar prize, they think, "If I win this one thing, I'll never have to win anything ever again." Which basically isn't an option

in a video game. We can't offer a prize that's so big, no matter how bad the odds, that if you won it, you'd win the game and you wouldn't have to do anything ever again. That defeats the whole point of a game. Which ultimately means that it might just work differently in video games than it does in something like the lottery. But there's an argument against that, too. Powerball's lowest jackpot is 40 million dollars. Now, most of us would think to ourselves, "If I had 40 million dollars, I could probably get through the rest of life okay." And yet, even though the lowest jackpot fits the criteria of the if I win this, I'll never have to worry about winning things again, it attracts far fewer players than the large jackpots do. As I've said, as prizes increase, there's an exponential growth in ticket sold. So even though materially to most of us there's not an enormous difference between being handed 100 million and 300 million dollars, more people chase that increase anyway. And the funny thing is, the jackpot gets split among however many players actually get the winning numbers. So, as the ticket purchase count climbs and the probable number of winners increases, often in the big payout jackpot

scenarios, any single winning individual doesn't actually see a significantly larger payout than in some of the more standard jackpots. So, because of the odds change made in October, the first week of January, the week that we're writing this, sees the largest lotto jackpot in American history at eight hundred million dollars. If the player turnout and tickets purchased continue to follow the exponential curve that the data up to this point shows, we may have to sit back and reconsider how we build incentive systems in games, especially in MMOs. Not that we'd like to. As a designer, James far prefers to think of players as being rational actors, at least as far as incentives are concerned. But, if people really are going to turn out in record numbers to play Powerball this time, even though they have substantially worse odds of winning than in the past, then, well, maybe it's worth at least experimenting with this concept in some future designs. See you all next week.