An online community that has an interesting use of trust in its system is Twitter. What makes it so interesting is how simple the system in terms of its basic functionality. That being said, its use has exploded, and in ways that are far from simple. Twitter has become the ultimate technology for the dissemination of information. Whether it’s for getting the word out about a political campaign, spreading breaking news, or if it’s used in a viral marketing campaign, Twitter is unmatched in its ability to get information to its users.
Trust becomes of great importance at several levels, the most obvious being the notion that a user is trusting of the other users who follow them on the service are not going to use the user’s information for harm. But to me, what is the most interesting is the system of trust that forms around the idea of who is an authority for a given subject on Twitter.
Users can essentially perform their own data mining in their searches for users to follow on any given subject. However, it’s clear that some users are more likely to provide a higher level of quality information than others. A user may claim to be a political science aficionado, but how can we really be confident of this?
A rather elegant and simple solution is to follow an approach similar to Kleinberg’s Hubs and Authorities model, but simply use a user’s followers and friends as a metric for importance. If Twitter or an external service through the Twitter API implements this system, it would serve as self-reinforcing mechanism that should identify who are the authorities on Twitter. Users who produce good content should gain more followers, and from this, the system rewards the user by boosting their authority. A higher authority should give the user access to more followers, and this process should continue until some equilibrium is reached. To an extent, this process happens naturally through name recognition and other marketing, but having an automated system would certainly help to identify these authorities.
Of course the scheme can be manipulated. An obvious way for this to happen is that users who are authorities in one subject area can use their popularity to claim authority in some other domain. However, I would suspect that users who are highly respected would not try to risk their reputation by manipulating the system.
Subscribe to:
Post Comments (Atom)
Posts
It is a great idea to use a reputation system for Twitter. This reputation system may not identify professionals or amateurs in some area of interest since people would follow users who cater to their interests and personalities, but still contribute to recognizing and increasing popularity of users who offer interesting information, which would encourage contribution to Twitter community. Also, I'm not sure if Twitter employs tags for status, but it would be nice if users could follow users' tags. For example, a user could follow another user on Entertainment news. Thus, the user's number of followers would depend on tags/ categories the user uses for each status. This would make it difficult for users to use their popularity to claim authority in some other domain.
ReplyDeleteWhen it comes down to online communities trust becomes the main relying factor when it comes to the acceptance of data. You commented on how people have the ability to stretch the truth about their identity. Some people may not be as adept at a topic as they say. Some people may not even know much at all. Because of this the idea of trust and the lack there of is a constant factor in a community where all the cues aren't present, and false information could be passed(like in twitter).
ReplyDeleteI agree Tom! Excellent idea, I am learning about Kleinberg's Hub & Authority model in INFO 2040 (Networks) right now. It seems like this would be a good project for "Twitter Labs" if such an analog of Google Labs existed. It would be very interesting if Twitter had an option to find other users this way. I would like to see a ranked hub model of Twitter, it would be very interesting to follow "THE" person on Twitter. The most common way of manipulating this system would be to remain overly active. Users who update often tend to gain followers relatively quickly. Someone on Twitter could be really really uncool, but as long as they update often, they may gain a nearly cult-like following, thus ranking them highly on the hub/authority scales.
ReplyDeleteI think that if implemented well, this could be a cool and useful feature in Twitter especially if the topic tagging feature could be integrated so users could follow specific topics and know which users have higher ratings. There is a good deal of data on Twitter that does not necessarily require accuracy of data but if there were a way to accurately track. This feature could effectively change the way Twitter is used from a stream of life updates and generally useless stuff to an information source. One concern that you mentioned is that users can be experts in one field but not experts in another and it would be difficult to give users rankings in every topic area.
ReplyDelete