[iDC] "The first rule of data centers is...

[Mark Andrejevic]

'Don't talk about data centers'" -- this from a piece on data centers/cloud
computing in the NYTimes architecture magazine (
http://www.nytimes.com/2009/06/14/magazine/14search-t.html?sq=data%20centers&st=cse&scp=1&pagewanted=all).
Yes, it's a bit over-dramatic and conspiratorial, but I think it usefully
highlights the tendential omission of (infra-)structural issues in some of
the discussions of the status of online activity. The repeated comparison in
the article is suggestive: data centers are to the digital "revolution" what
factories are to the industrial one. These are structures populated not
by people, but by data about them ("We used to think that owning factories
was an important piece of a business’s value...Then we realized that owning
what the [social?] factory produces is more important.”). What I find
striking about the article is the portrait it paints of the emergence of
galloping privatization -- the commercial enclosure of information hithero
unavailable (or stored in other forms), non-existent, or too expensive to
systematically capture and sort. The energy statistics help to provide some
indication of the scale of this rapidly expanding digital enclosure ("From
2000 to 2005, the aggregate electricity use by data centers doubled. The
cloud, he calculates, consumes 1 to 2 percent of the world’s electricity.").


The servers, of course, provide convenient and useful services, including
the email application I'm using to compose and circulate this post. They
also represent the privatization of the products of our communicative,
social, and transactional data on an unprecedented scale. What economists
call the "non-rival" character of this ownership (at least in some
instances) creates a distinction in control over the product of our
information-generating activity vis a vis industrial forms of
production (one of the several reasons contributing to the growing salience
of intellectual property law in the digital era). Obviously the fact that
Facebook knows who my friends are doesn't mean that I don't.
However, Facebook also has access to aggregate forms of data (as well as,
increasingly, the means of sorting, searching, and manipulating the
aggregate). At issue here are what might be described not just at a
significant portion of the means of communicating, transacting,
and accessing information, but also the means of data processing, of making
sense of tremendous amounts of data, including conducting ongoing,
large-scale, controlled experiments and interpreting the results. These
means of data processing or sense-making (albeit in instrumental fashion)
are inaccessible to the producers of the raw data, and the products they
produce are equally inaccessible (I may know who my friends are, but I have
no notion of the significance my pattern of interaction takes on against the
background of millions of other patterns, the controlled experiments that
generated them, and the results of aggregating, sorting, and querying this
data). In this regard, products are generated based on the aggregate
activity of data producers that remains inaccessible to them. The question
of whether the data-producing activity is understood by participants as
work, play, consumption, etc., while an important one in many regards, may
not illuminate the complex process whereby private ownership/control of the
means of interaction relates to private ownership of the technology,
datasets, and algorithms for transforming the raw data into information
products inaccessible to producers, products that appear to them only
in forms that render their original contribution indistinguishable,
invisible, untraceable.

The enclosure movement and the eventual forms of real subsumption and
technological transformation that accompanied it (the shift from
expropriation of surplus product to that of surplus value, the rise of waged
labor and the eventual transformation from piece-work rates to hourly wages)
favored the emergence of factory space, not just to aggregate worker
activity, but also, significantly to monitor and eventually rationalize
it. In this regard, the server farms might better be described as server
"factories" -- spaces where the productive aspect of monitoring is separated
out from the range of monitored activities. The fact that these spaces are
private, commercial ones, is neither natural nor necessary, but is the
legacy of historical forms of enclosure -- the continuation of enclosure as
a form of separation (of users from their date, data producers from the
means of data processsing), what Massimo De Angelis has described as a
process of continuous enclosure.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.thing.net/pipermail/idc/attachments/20090625/5147ffc0/attachment-0001.htm