![](bg8.png)
concept of information invasion as a natural
phenomenon that needs to be recognised and
understood if problems such as those experienced by
Doe are to be properly resolved.
We have detailed an experiment in information
invasion that has presented some unexpected results.
Of particular interest is the finding in our
experiments that decreasing the rate of belief
propagation within a topology of enterprise systems
actually increases the rate of information invasion.
This is provided that a dominant belief existed
within the network before the new belief was
introduced. Information is able to invade an
information domain more efficiently. The
simulation counter to this where queue reading rates
were faster confirms this finding. In such a scenario,
the faster belief propagation rate hindered the
invasion of the new belief.
Inevitably, reality places some constraints upon
solutions to alleviate some of the discussed
problems. Consider the eternal conflict experiment
as outlined in section 4.2. If we stop proactivity (the
sending of information) altogether during an
execution of simulation two until eventually the
queues become empty, its execution becomes less
chaotic and more predictable. The same would be
true if we had just emptied the queues by throwing
away the messages. Of course, in reality this is both
impossible and undesirable. It is not always possible
to prevent other enterprises sending an enterprise
information, and we obviously have no control over
all information in transit, especially non-machine
oriented methods such as postal mail. Naturally this
may not be desirable in any case, since throwing
away new information can mean throwing away
business.
Exacerbating the phenomenon of information
invasion is the issue of the lack of responsibility and
accountability for information system accidents that
needs to be addressed (Nissenbaum, 1996; 2001).
Currently, in the large, enterprises are not obliged to
ensure their data is accurate before passing it on.
Nor are they accountable in many cases for
originating inaccurate information. Such
occurrences are simply considered unfortunate
mistakes. A lack of local authority can also mean
remedying problems caused by information invasion
are impossible. In Doe’s case, he was able to
convince the insurance company of his ‘alive’ status
yet they were unable to fix the problem despite this.
However, various means to address these current
attitudes and approaches to information present
themselves, and require consideration and
investigation.
Firstly, it is not difficult to imagine a common
process that is adopted by all those participating in
an information domain that is triggered when there
is a query regarding the accuracy of some
information upon discovering an inconsistency.
When this occurs, all participants, including the
authoritative source, engage in a discussion to agree
on the value of the disputed information. This
process could even be automated, utilising a
common negotiation framework (Jennings et al,
1998; 2000).
A second approach concerns the traceability of
information, in order to identify the original, perhaps
authoritative, source. One method of achieving this
traceability is to tag all passed-on information with
historical metadata. Each time an item of
information is received and accepted by an
enterprise, it appends it’s own identity. Any
amendments to the data are also added to the
metadata with time information. In effect, this
metadata encapsulates the information’s entire
lifecycle. By examining this metadata, observers not
only have access to the origin of potentially
inconsistent information, but are also able to
ascertain other properties of the information, such as
how regularly and frequently it changes. However,
this approach is not without obvious issues that need
to be addressed, such as security (altering metadata
to avoid or implicate responsibility), information
size (over time, the size of the metadata could
become unmanageable), and metadata inconsistency
(the potentially huge task of reconciling inconsistent
metadata). A more simple, but less powerful,
method of introducing traceability would be for
organisations to specifically maintain their database
audit information for this purpose. This information
is then provided subject to an appropriate and
validated request. This could also be an automated
process, perhaps provided by a secure web service.
An approach to address the issue of
responsibility consists of the notion of an
information responsibility contract, where
enterprises that are involved in sharing information
between themselves in a loosely-coupled manner
agree to be directly responsible for information they
pass on to their contract partners. Such a contract
would highlight areas of responsibility for different
types of information, depending on origin. This
would encourage individual partners to be more
proactive in ensuring their information is up-to-date
before they pass it on, provide accountability,
perhaps even liability, for mistakes, and help to
ameliorate the ‘denial of accountability’ attitude
(Nissenbaum, 1996) that can exist in such
collectives.
ICEIS 2004 - DATABASES AND INFORMATION SYSTEMS INTEGRATION
480