What happens when gamification is poorly done?

Question

Gamification originally sought to take advantage of the fact that intrinsic, embedded rewards are more effective motivators than external ones. It has also become exceedingly common, but the widespread understanding and implementation of gamification seems to be focusing more and more on external rewards: fill up this boring form and get 5 Quipes. Perform these menial tasks, and get 20 more Quipes - and nothing else happens.

I have observed in myself 'gamification fatigue' - when the work remains uninteresting and meaningless (see Ariely et al. [PDF] for some interesting reading on 'meaning' and motivation) but now has some flashy game-like mechanism tacked onto it, the desire to earn points sustains me for a few minutes, but then I get tired and stop paying attention to it entirely. However, I still get the task at hand done.

I am interested in knowing if this is always the case. We know from other research that 'an expected external incentive such as money or prizes [can decrease] a person's intrinsic motivation to perform a task'. Gamification may also present other side-effects, such as slow progress (e.g. Codecademy's tutorials are terribly paced for non-beginners looking to get into a new language, as they tie learning syntax and learning programming concepts very tightly together.)

Is there any research on how users react to poorly designed gamification? Can gamification cause results that are worse than non-gamified tasks? Under what circumstances does the motivation to perform a task once again turn from intrinsic to external - it's obviously not a simple matter of overexposure, but what other factors are involved?

---

Further reading found upon a quick Google: http://www.slideshare.net/playfulwingmen/gamification-extrinsic-vs-intrinsic-rewards-17681228

Answer 1

One situation of gameification that had poor results was Google Image Labeler. The premise was simple: in an effort to improve the quality of results from Google Image Search, two players were randomly paired and then shown an image. They had a time limit to create tags for the image and would score points if they had matching tags. The rationale was that by providing a time limit and requiring consensus, this would result in creating tags that matched the first impressions that someone searching Google Images might have.

Unfortunately, things did not go as planned. For one, the pressure to identify common tags as quickly as possible meant that simple words tended to be used. Players would be shown a picture of a sunset on a beach and the quickest way to reach consensus and score points would be to type "orange". Same with books; even if the book cover with the title was shown in the picture, players tended to choose the tag "book" in order to quickly score points. A later change made it so common words were worth fewer points than rarer words, but nonetheless it continued to be a problem because it was harder to gain consensus for rarer words.

In its original iteration, you could peek at your partner's guesses. A bunch of users started using obscure words like "congenita" (meaning "occurring at birth") for every single image that they saw. When other users noticed that their partners were using words like these, they faced a prisoner's dilemma with earning points and realized that the only way they could actually get points was to do the same thing as their partner. They also typed in these obscure words and were awarded points for creating a consensus of tags. Later changes only allowed you to see your partner's guesses after the round was over which alleviated the problem. They also eliminated a large number of problematic words such as "congenita", but this had the side effect of making it so that in the rare instance when those words did apply (e.g. tagging an image of a congenital disorder), they simply couldn't be used.

Unfortunately, this attempt at gamification has not been formally researched and most of the above was gleaned from firsthand accounts of using the game. Google never revealed exactly how effective the service was or the exact reasons for which they discontinued it in 2011. Nonetheless, it seems that at least in the early days the gamification of Google Image Labeler directly resulted in a lot of poor quality data for Google and caused results that were far worse than someone doing it solely for the warm fuzzy feeling might have done.

I suspect that users may have initially come in with a desire to help Google categorize images, but they experienced cognitive dissonance when they discovered that creating quality tags was at odds with getting points and going to the top of the leaderboards. The choices were either:

• Continue creating quality tags despite not getting any points (very unsatisfying and demotivating)
• Quit playing (taking the game out of the picture)
• Sacrifice quality in order to maximize points (thus maximizing the amount of rewards from the game)

Cognitive dissonance suggests that those who chose that last option might have created rationalizations to help justify it (e.g. "I have no chance to win by playing by the rules, so I'll do what everyone else is doing in order to compete.", "Google's not going to be harmed by one person giving bad tags.", "The game is actually more fun when I use obscure words.").