Tuesday, March 31, 2009

Assignment 7: Ashley Vernon

An online community that I am a part of is FanFiction.net. FanFiction is fictional stories about a show that you are a fan of. These are also known as fanfics. FanFiction.net is an archive and forum where fanfic readers and writers around the globe share their passion.

According to Resnick, "a reputation system collects, distributes, and aggregates feedback about participants' past behavior". Based on this definition, I don't think FanFiction.net has that great of a reputation system. Currently, the only information it displays about a member's past behavior is the number of fanfics they have written, and how long they have been a member of the site. It also displays their reviews of fanfics, but these are spread out all through the site, and difficult to find. I think their reputation system should be updated because it doesn’t help people decide who to trust, or encourages trustworthy behavior.

Resnick has defined 3 properties of reputation systems: (1) long-lived entities that inspire an expectation of future interaction (2) capture and distribution of feedback about current interactions that will be visible in the future (3) use of feedback to guide trust decisions. According to this, my reputation system for the community would consist of members receiving levels for their behavior, such as an enthusiast. I think that his would help people decide who to trust, as trust plays a role in this community. Members trust that the authors actually wrote the stories themselves, and are not plagiarizing someone else's work. Members also trust that other members are not going to “flame” their stories in a review. The desirable behaviors in this community are to establish continuing relationships with other members, write creative fanfics, and leave well rounded critiques in reviews of fanfics.

My reputation system would include a member’s profile including not only how long they’ve been a member and how many stories they’ve written, but also if they are a frequent contributor, and if they are making significant contributions. This scheme could be manipulated because users could just review everyone’s story and not really say anything significant. To prevent this, the author of the story should able to indicate how helpful each review of their story was. Other members should also be able to indicate if the review was good, possibly by clicking on a “thumbs up” or a “thumbs down”. I think this would deter flaming. I would reward members by highlighting top-quality contributors. For example, if one member always leaves very helpful reviews that are well thought out, they would be acknowledged.

2 comments:

  1. I like the idea of assessing quality rather than quantity. It is hard to judge whether frequent contributions are necessarily good ones. This would allow the community as a whole to rate a user, and I think it would be hard to cheat. There would have to be a pretty large coalition all working together to generate enough "thumbs up" for each other to make a real impact. Also, if a group of users consistently votes for its members, that would be pretty easy to detect.

    ReplyDelete
  2. I like the reputation system in which users are incentivized to participate. The issue of significance is a large one and one way of solving it is crowdsourcing it like you described. However, I think it may be cumbersome to have a ranking system of reviews which essentially has people are reviewing reviews but it may be the only way to do it. I would imagine that most of the content is at least acceptable with a few items being downvoted. If it were to happen this way, this would be an effective way to filter out bad users and bad content. It would be interesting to see if you could come up with a natural language algorithm that could distinguish whether the reviews or if there are too many subtleties such as distinguishing between flaming and creative criticism.

    ReplyDelete