Data Mining as Search: Theoretical Insights and Policy Responses - Discrimination and Privacy in the Information Society - page 322

The “Mosaic Theory” argues that small bits of innocuous information, when
brought together, can provide a full mosaic of an individual's persona. Therefore
such practices of aggregation should be further scrutinized. It should be noted,
that very recently the US Supreme Court addressed this case on appeal (United
States v. Jones, 2012). It unanimously found the governmental search to be
unconstitutional, yet the majority relied on other grounds and left the acceptance
of the “mosaic theory” into the law for another day.
To conclude, this search-related theory of privacy can explain why data mining
must be limited, and when this must be done: in instances in which the tools used
by government prove extremely effective! The theory here presented is premised
on an interesting insight; data mining's analytical strength is the key to its
normative disadvantage. The public has learned to live and accept decision
making processes involving experts and field officers with their limited abilities.
These existing alternatives strike an acceptable balance between law enforcement
needs and civil liberty interests, even though they might compromise overall
effectiveness. Data mining presents a challenge which law must now answer to,
and a force which the law might find to be excessive if not properly checked.
However, this theory has clear limits - if the data mining process is not found to
be more powerful and insightful than other acceptable practices, this argument
loses its analytical force.
Limiting Searches to Limit “Fishing Expeditions ”
A third theory which can prove helpful in articulating privacy-related concerns
from the “search” perspective in the context of data mining analyses pertains to
their very broad scope. Usually, when considering invasive searches, laws and
courts find that they must be carried out narrowly, while limiting the gaze of
government as much as possible. Searches which fail to do so amount to a “fishing
expedition” on behalf of the state - the practice of looking through the files and
personal effects of individuals who raise no suspicion while striving to build a
case on the basis of information they might recover. Curbing “fishing expedition”
by governments is one of the central roles of judicial review (Solove, 2002). Thus,
this theory finds a normative flaw with very broad searches, which impact to non-
suspects.
Data mining initiatives famously call for actively examining and analyzing
datasets pertaining to a very broad realm of individuals, including those whom are
substantially removed from the matter at hand. The software does so while
striving to formulate patterns, trends, and clusters. Thus, data mining generates a
massive “fishing expedition” which resembles the most feared practices of
government - searching datasets in mass, while hoping to locate relevant evidence
(as opposed to initiating a search based on suspicion). On its face, this paradigm of
thought might be extremely helpful in grasping the concerns data mining
generates.
Yet again a theoretical obstacle blocks the application of this perspective in the
data mining context. If, under existing doctrine (and as explained above) the
government may review and analyze information which was lawfully collected in