| reputation Readings:
Online dating sites have become extremely popular - market researchers estimate that 40 million people visited these sites in the month of August alone. In terms of reputation and networks, they are very interesting. First, the people on the sites are connecte to each other, but as a series of dyadic connections, not a network. Second, as they interact with each other, they build up a personal store of infromation about each other that woudl be of great interest to others on the site, but there is no mechanism for them to share this information and more importantly, no motivation for them to do so. To think about how these sites (and other similar structures, such as collaboration finders for employees in large corporations, etc.) might evolve, first sketch a model of this (non)network and describe how it differs from a typical network. Then, think about where the information about people in the network resides - and where it would be valued. What sorts of systems might be used to effect the flow of this information. How would this change the dynamics of the site? Reputation Systems In comparing online systems where reputation is attached to a subject and gossip based networks where reputation passes through social networks we are really addressing an argument of time versus quality of information. Hess and Hagen identify the skills necessary to build and defend reputation as someone who can "communicate more information more accurately and in shorter periods of time". Resnick and Zeckhauser concur that "the Internet substitutes a much better distribution of what information there is". Motivations shift from negotiating resources to presenting and reading signals. Reputation attached is generally controllable by the subject and is thus comparable to physical dimensions of reputation that are easily proven and so "cannot be impugned by gossip" readily. These attached reputations can also be personal presentations which are often likely to be biased or selectively presented. It is also observed that cliques or coalitions are less likely to form online. Similarities can be drawn between on and off line reputation and its effects. For example, it is clear that perceived threat of reputation loss is a strong factor in peoples choice of trusting e-bay just as perceived off-line threat and cost of actual challenge strengthens coalition ties. Despite the differences in online and gossip based reputation forming, Resnick and Zeckhauser have demonstrated that for e-bay users, as with 'real world' interaction, "reputation profiles were predictive of future performance". Online Gossip In trying to form an effective online gossip system establishing reliability of information and motivation to gossip relies on several corroborative sources which requires small groups. Trust must also be established before information and character attacks will be shared for fear of retaliation from the subject should an untrustworthy group member be selected. Revisiting Resnick and Zeckhauser's reputation factors in the context of personal rather than retailer reputation we can discern several key reputation factors. Local interaction is necessary in the form of small groups as described above. To combat the difficulty of no repeat encounters, establishing a cost for entering a group would encourage smaller groups and thus more frequent repeat interactions and the passing of reputation assessment from peers. Reputation through association could be established by providing a method for support from others or backing by strong gossip group members. Providing a means of transferring reputation from one context to another would support wider reputation networks. To support the presentation of reputation over time is dependent on providing accountability of users. Anonymous or pseudo-anonymous users are unable to accrue long term reputation. Significant expenditure is a strong assessment signal and can be seen online in the cases of "selfless" acts for the good of a community. These must be made visible and more permanent to play a role as an effective form of signalling. Online Dating With such a strong and singular goal in the particular case of online dating, information collected from unsuccessful interaction (i.e. not resulting in a relationship) becomes useless to the collector but invaluable to others. There is currently little mechanism or motivation to share. The need for trading and negotiation of assessments of those with whom online daters interact is great. The information plays a key role to those attempting to assess future success in social interaction from the limited information communicated online. Not only is a description that someone presents of themselves likely to be a careful selection of qualities and possibly false, there is also a definite difference between, on the one hand, what someone finds interesting when assessing another and what someone feels is interesting and the things they feel can and cannot be presented in a particular context. Typical vs Dyadic In typical networks, connections between those you are connected to plays an important part in network forming and group cohesion as well as maintenance of trust. In the dyadic evaluation interactions, this mutual connection to a non-present third is absent and this missing link is significant in effecting the methods that online daters must use to retrieve information relevant to them. This has the effect of shifting the emphasis from "What do you think of him" analysis with others, to a direct question and answer approach in which considerably more effort must be expended to assess information alone and where direct confrontation is more likely. It would be interesting to look at homosexual online dating forums where the divide between those that can offer information and those you seek is closed. Is a typical network established in these cases, or is the dating interaction still predominantly dyadic? Sketching Online Dating  My sketch is for heterosexual dating, and emphasises the
lack of
communication on one side of the 'gender divide'. Each person is
represented by a circle and sits in a compartment that is viewable from
the other side but not by its neighbours. Each person maintains a
filter between themselves and the world (the outer image 'given') and a
collection (trash can) of past impressions and interactions. These are
no longer of interest to the individual, but would be a rich source of
information to others. I also distinguish
between profiles viewed but
not acted upon, and those which have developed into longer and richer
interaction (broader lines).Changing a Site In order to effect the online flow of information collected in (non)networks, such as dating, about people, motivation to exchange must be established. Affilative relationships provide benefits by sharing the work load of collection, analysis and dissemination of information. This provides strong motivation to establish online systems to encourage the formation of coalitions in dating forums. The question is then how, considering that these groups must remain reasonably small and intimate to succeed. Any heterosexual dating network is also split into single genders which results in low chance of a discussion being challenged by its subject. Taking an example of females seeking men, one suggestion is a group into which prospective male dates can enter only if invited. This provides an initial stage of filtering (sharing of resources) as incentive for women to join and a 'hub of activity' to attract male interest. A discussion board cold be provided for the group to provide feedback on men and would be viewable by all females. This is then a semi-public (viewed by all women, added to by privileged group members) and semi-private (not easily viewable by the men it discusses) space. Group membership could be regulated by high cost entry process, such as recommending a certain number of prospective men that fit the groups tastes. Cross examination also appears to be a good tool in character assessment and is considered more reliable than the single source information often presented online. As Hess and Hagen describe: "The probability that two people
benefit by telling the same lie was smaller than the probability that
one person did so."
Owens, Shute and Slee (2000) give the following account of reasons for gossiping (in their case in incidents of victimisation): "creating excitement, getting
attention, using it to gain inclusion in a particular group or maintain
a position in a group, alleviating boredom, using it to avoid being
victimized oneself, and retaliation."
It is then these motivations that online systems should try to encourage. To establish effective channels to support the above, I think a blurred region between private and public must be provided. Although increasing gossip and other social network reputation systems may aid assessment and interaction in online dating it also carries the burden of victimisation, attack and hierarchy that many look to the web to provide. This must then be a feature available but not all inclusive to succeed. Finally, due to the fact that an online dating presence often continues only until successful, negative comments generated from unsuccessful interaction would be likely to dominate feedback. Also present might be recommendations based on profiles viewed but not acted on and interactions that "just didn't suit me". It is unlikely that someone would share a "good catch", although this information could be delayed (like share prices). There may also be a stigma attached to something as singular as forming a relationship that would underlie information presented that "if they weren't good enough for him/her, why should they be good enough for me". |