What makes games addictive? The better our answer to that question becomes, the more likely we are to crack that one mystifying puzzle: human motivation. Successful social networks tend to be similar to games in many ways. For example, the notion of experience is very important in role-playing games: a character advances over time and becomes more powerful. This is something I will come back to in future posts on the topic of motivation.
One idea which I already scribbled down before the launch of this blog is [[PictureFight]]. The goal here is to create an environment that produces unique, free (Creative Commons) photos with good metadata. While there are many photo sharing services that appeal to people who enjoy taking photos as a hobby, the PictureFight concept is meant to attract people more interested in playing games and participating in contests. Players submit free photos, which are then rated by a non-participating jury. Players can then fight each other, where the unique attributes of their photos (including metadata) are taken into account. See the wiki description for more details.
PictureFight could be substantially more complex than [[w:Songfight]], which is a simple poll-based approach where voters choose a song favorite for a given title. Simplicity is, of course, often an advantage, so the simpler Songfight approach might also be very workable for pictures.
This idea is underdeveloped and is therefore classified as a “stub”. Perhaps someone with experience with [[w:Magic: The Gathering|Magic]] and similar “battle card games” could comment and expand it.
[[PictureFight|View and edit this idea in the wiki]].
November 10, 2005 at 8:16 am
in a related vein, a friend told me about the “ESP” game.
basically we both log onto a game session and are shown a series of pictures and need to each provide a word to describe it. and if we get a matching word, then we score a point.
example:
– we are both shown a picture of a blue, nokia phone
– you say “handy”, i say “mobile”. no point =(
– we are both shown a picture of an elephant in the savannah
– we both say “elephant”. point!!
besides being a fun game to play — it’s an excellent way of building up useful metadata about the pictures.
can be applied to lots of related applications.
November 10, 2005 at 8:35 am
Yes, that’s interesting. It seems like many users are quite happy to use tags to describe blog posts, photos and pretty much anything. Since they are already providing the metadata, you can then simply try to find out the most common ways to describe certain things, and attempt to infer synonyms and related tags.
I wouldn’t be surprised if Flickr already used matching algorithms like that in clustering related tags.
November 10, 2005 at 8:39 am
True.
However, that’s only because “tags” are in vogue at the moment.
Once that passes, you need to have incentives for people to maintain good metadata?
November 10, 2005 at 9:09 am
As developer of FlickrLickr, I maintain a metadatabase of over 50,000 Flickr photos. I have reviewed thousands of them myself. I am certain that tagging is not simply a passing fad. And that’s in spite of the fact that most tagging interfaces still suck (no autocompletion, no synonyms, no localization). It shouldn’t be a surprise that low friction “soft” metadata is relatively easy to obtain from users.
That does not mean that it’s not worth incentivizing it through games, of course. But when it comes to incentives, I’d prioritize by decreasing importance:
provide content under free licenses
provide “hard” metadata (good content descriptions)
provide “soft” metadata (tags etc.)
While the soft metadata has problems, these can often be solved algorithmically. It’s the hard metadata that is often irredeemably bad on Flickr. Image descriptions are full of inaccuracies and irrelevant personal anecdotes or simply missing. And free content is still only a minority of images (even if you combine all CC licenses, which includes “non-commercial use only” and “no derivatives” in different combinations).
How do we get better “hard” metadata? In the PictureFight scenario, a judge could rate a photo with a detailed and accurate description more powerful than one without it. I’m not sure you can easily run matching algorithms, though.