stong and weak ties

topic posted Wed, January 14, 2004 - 5:39 PM by  DUFFMAN
hi - i am new to this tribe.net thing but was glad to find a graph theory group! i am a research fellow at the research center for group dynamics at the institute for social research at the university of michigan. i am using graph theory in some of my work on cross-cultural differences in ego-centered friendship networks. but i have something of a different question. i have been reading a lot about strong and weak ties in the sociological literature. i am wondering if anyone has any opinions about differences in friendship network formation between a site like tribe.net and friendster.com? friendster has none of these tribes that may serve as a source of weak ties in which one may have the opportunity to meet a large number of people, some of whom might become strong ties. friendster, not having such groups and being based more upon strong ties, has less of a dynamic property to it - networks on friendster may be much more stable. has anyone thought about this?
posted by:
DUFFMAN
Michigan
  • Unsu...
     

    Re: stong and weak ties

    Wed, January 14, 2004 - 9:17 PM
    From what I can tell, there is a culture of maximizing friends (or at least personal network size) on Friendster that seems absent of Tribe, perhaps because of the weak ties available through the tribes. In fact I find Tribe more valuable as a threaded collection of newsgroups than as a social network in the Friendster sense.

    What do you mean by stability? Rapid convergence?
  • Re: stong and weak ties

    Thu, January 15, 2004 - 1:02 AM
    interesting points.

    i think with friendster the whole idea of a showing people how many others they are connected to created an element of competition and dicksizing. people started linking with others who werent even their friends just to increase the size of their local graph. i found it amusing that friendster had that little "are you SURE this person is your friend?" verification step. clearly that wasnt taken seriously. so not only do they have the problem of "mere" aquaintances linking to one another (perhaps a good example of a weak link), but also any bogus links to total strangers.

    so, in trying to stay on top, i think tribester also suffers from this a little bit, in that there are still some folks with 100+ connections - clearly these links represent something different than "this person is really my friend".

    i think its become somewhat ambiguous as to what an edge between two people actually means.

    i would propose two things, if anyone asked me, which no one has: first, weighted edges. it seems like the next logical step to provide a mechanism for people to express (perhaps by gradients), exactly how "serious" the edge is between oneself and someone else. secondly, distinct graphs representing different relationships. for example, both tribe.net and friendster describe adjacencies between individuals. there is only one bidirectional type of edge. alternatively, there could be a "friend" graph, a "colleague/contact" professional graph, an "aquaintance" graph, et cetera. the tribes SORTOF accomplish some of this, except they are really just sets, and that isnt the same.

    the ideas arent inclusive, but both step towards more accuracy in describing the social network actually being represented.
    • Re: stong and weak ties

      Thu, January 15, 2004 - 3:17 AM
      There are many threads in other tribes discussing the information contained in the friend links. An example here:
      www.tribe.net/tribe/servl...iewThread.vm
      Personally I think that giving more information for those explicite links is going to nowhere. It is just too incompatible with our RL notion of social ties (frienship). The incompatibilities come from two sources one is the expliciteness of the information, other is the complexity of any mathematical model that would be realistic enough.

      One model of implicite social networks is presented on my wiki:
      zby.aster.net.pl/kwiki/index.cgi
      • Re: stong and weak ties

        Fri, January 16, 2004 - 10:33 AM
        granted, complexity can be a pitfall, but why is more information incompatible with our RL relationships? our RL relationships ARE more complicated than a single edge relationship, why should a model be trivialized?

        also, the temptation to take "the next step" in graph complexity (like edge weighting) i think is worth considering in the context of this tribe.

        also, thanks for the link to the other tribe.
        • Re: stong and weak ties

          Fri, January 16, 2004 - 1:11 PM
          Our RL relationships allow for and actually require some ambiguity in the nature & strength of the ties. Too much explicitness in the links messes with that. Less is more.

          Tim
          • Unsu...
             

            Re: stong and weak ties

            Fri, January 16, 2004 - 1:50 PM
            I do see a place for heuristics to estimate the nature and strength of ties (though that's hard problem). But I agree that it's unreasonable to expect users to enter complex relationship data manually.
          • Re: stong and weak ties

            Sat, January 17, 2004 - 10:40 AM
            I agree with you Tim - In doing my project, I find that people are really uncomfortable in rating their friends. I was recently talking with Randy Nesse (he's an evolutionary guy - you've might of read him) about evolution and psychoanalysis - that in our social relationships we have to surpress certain aspects of the relationship because we would never have any friends if we kept too close a track of the degree to which people cooperate or cheat. Its an interesting point.
    • Re: stong and weak ties

      Sat, January 17, 2004 - 10:29 AM
      good point. in my own work i'm using weighted edges by having people rate the closeness of ties, frequency of interactions - these i view as contextual information about the relationship, and ratings of popularity and friendliness of each individual in the network, and finally I have people describe the nature of ties - are these people best friends, acquaintances, boy/girlfriends. I am finding one of the problem being reference points for likert scales, but this research is my first time dealing with network data.

      My interest in the topic is less mathematical than social - I am trying to explain how cultural information flows through networks, which may help explain differences in psychological and social processes and behavior.

      not having many tribe.net friends, I am not sure about how these relationships form relative to friendster, and it may be confounded by the much larger number of people on friendster. i think in some friendster networks, people join others who really are loosely connected - i mean, who hs 256 close friends? but it is interesting to me how others use it to keep in contact with their close friends who may be spread out across the country and world. it makes it so that people can chat in a kind of virtual living room or bar. this may be a response to the fact that our society is changing so that a greater number of people in their 20s and 30s have lived in multiple cities for extended periods of time than in the 1970s (NORC - GSS), and this is more true for people with higher education, who probably have more money to have computers and leisure time to noodle around friendster.

      but i wonder what is unique about these people with 100+ friends. I wonder if there is anything unique about them, and if the number of their friendster friends and actual, real life friends (real in that they physically interact with them) covaries...

      thanks so much for your thoughtful response to my question.
  • Re: stong and weak ties

    Thu, February 5, 2004 - 3:47 PM
    Here is my personal friend adding algorithm, if you will...

    On friendster I add friends willy nilly. I have 170+. Many of these people I have had little meaninful interaction with. Friendster has a one dimensional use, to display social networks.

    Tribe.net has message boards on each of the tribes. I find the linkage between a social network, personal information, and message boards to be extremely useful. When having a discussion with someone on a message board I often check out thier personal page and thier social network. The personal page gives me some context on what thier background is, what other related topics they may be knowledgable about, et cetera.

    If the person is linked to me through people I respect, then I weight thier opinion favorably. If they are distantly tied to me, I dont necissarily discount thier opinion, but I accept less of thier claims based on trust. Other people do this as well, allocate trust based on how much they trust the social trace. For this reason, I carefully groom my listed friends.

    I only 'friend' people who I know fairly well, or have some other method for verifying thier trustworthyness. As an example from my friends list, I do not know James Hughes and we have only exchanged a handful of e mails. However I 'friended' him because we have a half dozen friends in common, all of whom I respect.

    It is, however, important to have a large extended personal network. Access to personal and professional information is restricted if you are not connected in some way to someone. To compensate for a picky friends list I have 'friended' several people based on the size of thier friends list. One of my 'friends' is connected to about 300 people. I do not know this person, but through him I am connected within several degrees almost every person I have ever wanted to. This strategy is more important on friendster, where you cannot view the profiles of people not in your network.

    My friends and I, who talk a lot about the social physics of the system, refer to this friend list construction type as "hubbing." Keep your friend list tight, but keep a few spokes out to the masses so as not to restrict yourself to local optima.
    • Re: stong and weak ties

      Mon, February 9, 2004 - 3:25 AM
      great point Todd - thanks for the insight. It is amazing how many people out there are really thinking about social network issues. maybe I'll post what my research findings are and get all your opinions on it. Thanks again all of you. Sean

Recent topics in "graph theorists"

Topic Author Replies Last Post
Network Visualization software? Anthony 5 January 28, 2008
random graph dynamics watson 0 October 4, 2007
Capital Letters Karl 1 October 4, 2007
Introduction to the new TOE Unsubscribed 4 January 5, 2006