Networking Reference
In-Depth Information
many different institutions. Alice's actions impact Bob, and Bob's actions impact
Alice. Furthermore, Alice and Bob impact many other components of the network,
as they provide information and services that benefit others.
Often, our actions in the network are facilitated and constrained by our computa-
tional environment, the applications and the devices we use. These change how we
construct our daily reality. It is possible to rely on anonymous others, without any
social connection. We interact with many different computational agents and rely on
them for many different tasks. The interfaces we use help us create a mental model
of what their functions are and how reliable they are. We collaborate with others, in
very large numbers, to create highly sophisticated artifacts (like Wikipedia and open
source software). In all these examples, trust exists as a foundational element that
describes the interdependence between the different components of the network. A
modeling challenge is to understand the implications of these interdependencies.
How do the connectivity in the network and other design decisions impact how trust
evolves?
We have seen many different institutions related to trust, from cultural and
organizational institutions to those that are based on algorithmic constructs. In in-
formation creation, we analyzed Wikipedia as a unique collection of organizational
rules, participant activity and computational tools, all dependent on each other. The
information in Wikipedia is constantly under attack, as many attempt to alter the
content to serve specific interests. Understanding the vulnerabilities of a complex
system like Wikipedia requires a network-based analysis. The organizational and
computational elements impact the way the contributors can add and modify
content, and what the readers will see. The contributors' interests, objectivity and
level of involvement is a significant contributor to the reliability of content in user-
created sites like Wikipedia.
Similarly, computational reputation methods are built on user input, i.e., people's
evaluation of each other. The reliability of a reputation method is the main
determinant of how much it can be trusted for a specific action. To a large degree, the
output of the reputation method is dependent on the quality and quantity of the user
input. A site with a large user base with many honest reviews is likely to have more
reliable information about the reputation score of each individual. Furthermore, such
sites tend to foster a community feel, which enhances the perception of the site's
trustworthiness.
However, the actual reputation computation involves many other design de-
cisions: which inputs to aggregate; forget or disregard; how much fake input
to inject; how to aggregate the inputs and how to present the results. In some
cases, this is done in the interest of increasing the volume of interactions and the
associated user input, especially in cases where there is no initial basis for trust.
These manipulations alter to which degree the reputation value is dependent on
the underlying user input. However, the user input is also dependent on how an e-
business site and sellers are viewed, how their feedback is sought, and to which
degree they get attribution for their input. In short, reputation is a closed system
dependent on trust and supports actions that require trust.
Search MirCeyron ::




Custom Search