Gamification Ideas (General Brainstorming)

[bbernhard]

We already have quite a few awesome ideas regarding possible gamification topics (many thanks to @dobkeratops at this point), so I think it might make sense to pick one or two ideas and try to create a proof of concept to see if gamification could be something that convinces people to contribute.

Goals

The following are just my personal opinions and definitely open for discussion. I am more the "occasional minigame player", so any gamification best practices and ideas are really appreciated ;-)

• (relatively) easy to implement To be honest, I am not really sure if gamification works. I mean, it definitely helps, but I am not sure if it has such a big impact, that it's worth to invest a lot of time in it. So for the first game I would tend to implement something that's not that complicated to see if we are on the right track or we are wasting our times.

• adds immediate value I think we have quite a few gamification ideas, that are really awesome and for sure fun, but probably only add value over time. e.q: "showing small, cropped images to users to see if we can determine SIFT features from it". While I am a total fan of this idea and I definitely want to try that some day, it's something that might or might not work. So probably not the best idea for our first gamified application.

• instant gratification Not sure if this is a crucial point, but I am personally a big fan of instant gratification. So if I am labeling/validating something I would expect to get points/rewards instantly. However, as users are maybe operating on images where we don't have any reference data yet, it could be pretty hard to give them a reward when we don't know if they have labeled/annotated the image properly (I could imagine that some people would try to trick the system by fake labeling stuff just to get a highscore).

Ideas

I still have to go through the other tickets, to collect all the gamification ideas, but one idea that came to my mind today:

• tamagotchi like application: I could imagine a animated monkey that users need to keep a live by regularly annotating and validating a few images. I think with a tamagotchi like application it would also be possible to withdraw the earned points/coins/hearts... later in time, if we detect that the user validated the image wrongly on purpose.

Collected from already existing tickets

• odd one out (from #20): I really like that idea, it could be a simple and effective way to improve our dataset's quality. However, I currently see two problems:
  1. We do not have enough labels yet, so that the game would make fun
  2. When do we reward the user (with virtual coins, points..)?

The first point is probably pretty easy solvable by adding more labels and bulk-importing data from wikimedia commons. But I am not sure about the second point. As a user, I would expect to get a reward (virtual coins, points...) when I have selected the right answer. But as the user might be operating on images where we don't have any information in the database yet, we would need to give him his points in advance (and withdraw them later if we find out that he has chosen the wrong one). I am not sure that works from a gamification perspective or if users lose interest very quickly then?

[dobkeratops]

could we get graphical feedback of the database quality: perhaps we could make an animation of 'all the objects , sorting them by greatest similarity, i.e. you'll see a 'dog' morphing between many postures. As the quality of the data improves you could see these animations associated with any word 'get better', and current 'mislabelled' objects would show garbled,random sequences. with much less coherence. (bear in mind the ability to consider crops from labelled scenes in that)

(something engaging... could be a game , or just be visual feedback)

[bbernhard]

perhaps we could make an animation of 'all the objects , sorting them by greatest similarity, i.e. you'll see a 'dog' morphing between many postures. As the quality of the data improves you could see these animations associated with any word 'get better', and current 'mislabelled' objects would show garbled,random sequences. with much less coherence. (bear in mind the ability to consider crops from labelled scenes in that)

Interesting idea!

But I think that would be more useful for visual feedback, right?

[dobkeratops]

sure. it doesn't constitute a game , but it might make for a slightly more engaging activity/experience.

perhaps 'toy' rather than 'game'.