The Way Forward - Discrimination and Privacy in the Information Society - page 339

population segments. Furthermore, the condition of these groups might potentially
be far worse than even those commonly discriminated against in the past and
today. These new groups might be dispersed throughout society. Thus, they will
lack the minimal political force to bring the issues of their misfortune to the
forefront of the legal discussion. Even worse, given the inherent obscurity of
the data mining practices (features this topic tries to somewhat mitigate in the
discussion set out in Chapter 17) those adversely impacted by these processes
might not even know this is happening!
Dealing with these new sets of concerns calls for substantially altering the
current understanding of anti-discrimination theory; for many years now the focus
of discrimination law has been on generation of bigotry towards "protected
groups" which were (and some still are) systematically discriminated against.
Even with the bigotry gone, much of the structural discrimination is set in place.
Therefore, much of the existing law is tuned towards discrimination which is
motivated by discriminatory intent - even if such intent cannot be proven. Or, it is
focused upon reintegrating insular groups into the general society. Yet the novel
forms of discrimination data mining might be setting forth do not feature these
elements. Therefore, they call upon academics and policymakers to rethink the
theory and practice of discrimination in this unique context.
As these last few paragraphs indicate, data mining practices might lead to new
and serious fears of growing discrimination. Therefore, should this understanding
not lead to an overall recommendation to limit or even ban these forms of
analyses? While this recommendation might have merit in some limited contexts,
we generally find it should be treated with caution. The issue of discrimination in
the context of data mining must be approached with an open mind. While the
potential detriments must be acknowledged, we must also consider a very different
option - public intuition is wrong, and data mining does not pose serious or
unique discrimination-based concerns.
Furthermore, we must consider whether the discrimination-based concerns have
resulted from an irrational, Luddite-like fear of these advanced models. 26 Or, it is
also possible that a much greater and sinister force is in play. The seeming
intuition that data mining leads to unacceptable discrimination is merely a
manipulation of the powerful trying to influence the weak. While automated
practices might finally lead to equal treatment, they might compromise the elite's
dominance and subject them to the same level as scrutiny as everyone else
(something they are not used to). Therefore, the elite might forcefully advocate
against these practices, pointing us back in the direction of human discretion
which has ruled in their favor time and again (and in that way hiding its self-
interest). For that reason, legal and other scholars must exercise extreme caution
when pointing to the discrimination-based flaws of data mining practices - as they
might be merely pawns in a much greater game. These last few arguments might
seem unnecessarily paranoid and probably are. Yet they still demonstrate the
importance of seeking out a sound analytical foundation to any regulatory step
taken to battle discrimination in the novel context of data mining.
26 See discussion in Taipale, K.A. (2004). For a very different perspective, see Solove, D.
(2011).