How Engagement Rewards Backfire: The Overjustification Effect and the Peak-End Rule

It took a few tries to find the right framing for this one. The original idea came when I noticed there were a few games that I enjoyed for a while but which left me feeling very negative and I was trying to pin down what they had in common. While they tended to be games that didn’t end, that clearly wasn’t the problem - my friend Iceman provides an excellent case study here, because he played a lot of both World of Warcraft and Team Fortress 2, neither of which end. In WoW, he felt pressured to play more than he actually wanted to in order to help out his guild, so when he finally quit he did so about as hard as I did and he never played an MMO again. But with TF2, he’d frequently put it down for a while and then come back just as strong as before.

My first attempt at classifying the problem was what I termed “paraludic incentives” (or “gameplay-adjacent incentives”). The idea was that these were mechanics that were connected to the core gameplay, but off to the side in a way that reached outside of the game to provide external pressure (via real-world time limits, social obligations, or whatever). These created situations where the reason you were playing wasn’t the gameplay itself, but something it connected to - if someone asked you “Why are you playing that game right now?” your answer wouldn’t be “Because I want to,” but would be something like “To get the daily reward,” “To make sure my town doesn’t decay,” or “Because my guild has a scheduled raid.”

This turned out to be really confusing. I had trouble explaining it to people and even managed to confuse myself with questions like whether achievements or internet scavenger hunts counted. I realized that the real issue was what player behavior these mechanics were trying to encourage, and once I saw the parallels to growth hacking and started thinking instead of systems that directly reward player engagement everything fell much more smoothly into place.

Also, my first draft of the post (which briefly went live on the website because I forgot to properly set the draft flag, eep) spent a lot more words cataloging various popular kinds of engagement rewards. I do think there’s definitely room for posts that dissect and catalog mechanics in depth, but it’s important to recognize when that’s not really the point of a post. I don’t actually have much to say about the different kinds of engagement rewards - the point I want to make is that they have some non-obvious problems. So I breezed through some examples in a single paragraph just to show that I was talking about much more than just log-in bonuses and moved on.

This did mean that a few things that I’d been planning to say were suddenly out of scope. (Or rather, that I suddenly recognized they were out of scope.) One is pointing out that some types are just differently-framed versions of other types, but that the framing can make a big difference (see Framing and World of Warcraft’s Rest System for a look at a specific example of that). And despite the huge possible range of ways to frame these incentives, developers mostly seem to have settled on a handful - which makes sense, because if you’re trying to motivate specific behavior, you want the cues to be as clear as possible, and familiar systems are more clear than unfamiliar ones.

One thing I went back and forth on a few times was whether to discuss the different effects of rewards that require minimal engagement versus those that require more. After all, the more work you have to do for the reward, the more of a chore it is if you’re ever doing it when you don’t want to. My biggest example to illustrate this was Jetpack Joyride which I used to really like but now have almost zero interest in. A major cause is the daily reward system that eventually got added: the strong arm machine. Once a day you can use it to get a reward, and every fifth reward is a costume piece that otherwise costs real money. The trouble is, using it requires actually playing and collecting tokens, and the rate at which the tokens show up decreases as you approach the better rewards. The first day, you might get the tokens right away; the fifth day you might have to play for several minutes to collect them all. That gets old fast if you’re just trying to collect the reward because you have a spare moment and don’t actually want to play just then - I got way more annoyed about this than I would have about just launching the game once a day.

But proper analysis of this would have necessarily gotten into several variables that ultimately would have been about how to make engagement rewards more or less annoying, and that’s not really my point. So ultimately I left it out. There’s a temptation to make posts longer when they feel short and when I feel like I have more smart things to say, but I try to remember that the best posts are a complete examination of a single idea. Digressions are for Extra Content. :)

There were also two more sources that I originally planned to quote. One was Alfie Kohn, the author whose book Punished by Rewards changed how I thought about behavioral psychology and which I also decided not to quote in a previous post. (He’ll make it in some time, I hope!) The other was The Overjustification Effect and Game Achievements, which is definitely related reading that deals with multiple concepts I was using. But in both case, I just couldn’t find a quote that added much of anything to the point I was trying to make - I got more useful details elsewhere.

Anyway, that’s it for this blog post. Hope you enjoyed!