Felix Stalder on Fri, 1 May 2020 14:36:33 +0200 (CEST)

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

<nettime> Tracking People and Modelling Society

[This is a text that I've written for the Rosa Luxemburg Foundation
in Berlin. It takes up some of the issues we have discussed here. It
focuses on tracking of people and the modeling of society and I try to
think about democratic potentials here, rather than the more obvious
authoritarian ones. Felix]

Tracking People and Modelling Society


At the moment, many people are sensing how the tectonic plates under
their feet are moving. It is hard to get one’s bearing on such
shifting grounds. Beginning with the virus itself, which is assumed to
have jumped from animals to humans (“zoonotic spillover”) sometime
last fall, there are simply too many actors in the complex dynamic
system of a planetary civilization whose paths have been altered
in hard-to-understand ways. This makes it impossible to plot the
cumulative effects of their interaction.

While a lot of things are fairly chaotic and improvised reactions
to fast-changing events, there is a certain structure to it, simply
because people and institutions draw on that material, political and
cultural resources which they have available. But which resources
to draw on, how to mobilize them and how to create new ones in the
process is the key question. While there is path-dependency and
continuity, even in the way actors can change paths, there is also a
moment of extraordinary openness. It is therefore important not only
to be vigilant against the authoritarian forces that are exploiting
this crisis, i.e. to defend democracy as it exists, but also to think
about ways of strengthening and expanding it right now.

One area where this is particularly urgent is the area of “big
data”, i.e. the collection and evaluation of large amounts of data
for the analysis of complex dynamics. In what follows, I would like
to focus on two areas of data practices where the changes currently
underway are particularly profound and enduring, and thus full of
potential: the way people are recorded informationally, and the type
of modelling we use to understand and influence society in real time.

Tracking People

The spread of the virus laid bare the shallowness of the notion
that we are all unique individuals, each master of our own destiny.
Rather than being able to rationally calculate our own paths,
unaffected by people with whom we do not choose to enter into explicit
relationships, the virus reveals the scale, scope, and intimacy of the
relationships we have with each other and many “other others” as a
basis of everyday life. Only now, the relatively privileged members of
society, locked into their own private spaces under a state of a range
of lockdown rules, are forced to confront themselves as truly atomized
individuals. And even they are experiencing first-hand how unnatural a
condition this is.

In order to follow and contain the spread of the virus, those
relations along which the virus can spread need to be cut, either by
interrupting relations across society as a whole by way of general
“social distancing” rules, or by trying to track the specific
paths by which the virus spreads through society. And since the
dominant way this virus spreads is human-human, it means tracing the
social relationships that come to define the singular person, which
extend far beyond those that are explicitly chosen.

There are two powerful, ready-made models for tracking people.
The first comes from the state. Since the mid-eighteenth century,
the state has been keenly interested in tracking its subjects for
purposes of taxation, conscription, security, and various kinds of
bio-political concerns including public health. With the expansion
of the state’s functions, not least through the welfare state, the
tracking of people across an ever-larger variety of contexts has
steadily increased. Increased mobility and social complexity shifted
this regime over time from static measures (such as border controls)
to dynamic measures (such as collecting communication meta-data). This
was part of a more general shift from a “disciplinary society”
centred around enclosing institutions such as the school, the
army, and the factory towards a “control society” focused on
tracking and manipulating cybernetic flows for some version of the
aforementioned purposes. Second, in the early 2000s, the most advanced
sectors of capitalism began to outstrip the state’s ability to track
people through the sphere of consumption. While the first platform
to keep track of individual consumers was created in the late 1950s
by the credit card industry, it has since spread across society and
consolidated into “surveillance capitalism”. Its main feature is
to re-organize ever-more dimensions of human activity to optimize
the tracking and manipulation of people, this time in the pursuit of
private profit. An approach that was spectacularly successful, if we
take corporate valuations or the source of extreme personal wealth as

While these two models have always had many points of interconnection,
as Edward Snowden revealed again in 2013, in the West corporations
and the state have not been fully integrated and often stood in a
contentious relationship with one another. A case in point are the
repeated battles over the use of encryption, which pits globalized
technology providers against national security apparatuses. Under the
pressure of the crisis, these two models are now beginning to merge
more openly and comprehensively. It is too early to say how and where
this will happen, but it seems very likely that the pressures and
desires to do so will intensify as part of the efforts to contain
the virus after the initial phase of the lockdown, the duration of
which nobody knows. The Orwellian implications of such a merger
are quite clear, and the crisis could easily be used to push such
concerns aside. In the West, the danger is greatest in the US, where
the security and corporate sector are already deeply intertwined and
privacy legislation is weak. In China, of course, these sectors have
never been separated.

But these are not the only available models. Civil society—in which
one might include the non-profit health care sector—has its own
way of thinking about the relationship between the individual and
society at large. Here, notions of care and mutuality, both on a
person-to-person but also on a person-to-collective level (say, in the
form of collective care institutions) are more central than control
or profit. We can see these notions implemented in privacy-protecting
tracing mechanisms, which focus exclusively on public health demands.
They rely on voluntary participation out of a sense of collective
duty. The recently proposed “Pan-European Privacy Preserving
Proximity Tracing Initiative”, for example, embodies this way of
thinking. It does not help to protect the users individually, because
it only warns of potential infection in the past. The goal is to stop
the spread of the disease by increasing the speed and accuracy of
self-isolation, a measure that in this scenario only serves to protect
others who can thus continue their lives without interruption.

The personal gain here is indirect, through the ability to continue
to live in a society that continues to function as a whole, rather
than direct, that is, through the improvement of personal health.
But Europe lacks the technological infrastructure to implement such
a solution on its own. All apps need to run on the mobile platforms
provided by Alphabet and Apple. Not only can they implement any
solution better because they can modify the operating system in
ways normal app developers cannot, but they can also quickly update
millions of phones at the same time. These two companies have now
announced cooperation in the development of a proximity-tracing
solution. Even if this solution is, at the beginning, similar to the
European proposal, it will be a massive challenge to ensure that it
will not be transformed—over time and under business or security
imperatives—from a care to control approach.

However, what this entire discussion shows is that there are two
possible trajectories along which such a changing relationship
between the singular and the collective can be expressed and
institutionalized. One employs the heavy hand of the state and/or
business in opaque and unaccountable ways, while the other builds on
a civic sense of personal dependence on known and unknown others, a
renewed sense of solidarity.

There can be little doubt that the comprehensive tracking of humans
and non-humans, of places and flows, will continue to increase. There
are very powerful security, commercial, and public-interest rationales
for it far beyond the containment of the virus. For example, if we
are to move to a circular economy that reduces pollution and waste,
we will need to develop new and improved measuring and accounting
procedures that include more tracking of resources and processes. If
we are to move towards decentralized energy provision and “smart
grids”, more tracking of energy supply and demand will be needed.
All of these developments have already started before the virus spread
around the globe, and once the planetary economy starts again (in
whatever shape), these will continue in an accelerated fashion.

The notion of an “autonomous individual” has been critiqued
at least since Foucault’s work on bio-politics and his notion
of the “death of man”, and it has completely vanished from
neo-behaviourist contemporary network science. But what will replace
it? The question of the relation between the singular and the
collective is no longer a purely theoretical one, but one that
technological infrastructures are actively shaping. In order to
reinvent democracy, it is not enough to try to minimize control and
profit-oriented tracking. It also requires actively developing and
implementing ways to make the relations between each other more
explicit under the impetus of care.

Modelling Society

If there is an image that came to represent the crisis across the
globe, then it is the one that overlays the dynamic curve of the
rising (and at some point falling) number of new infections and the
static and horizontal line of the health care system’s capacity to
care for patients.

From this, a simple political demand follows quasi-automatically:
“flatten the curve!” That is, do everything possible to prevent
the curve from crossing that line, because the world above the line,
for all but the most radical free-marketeers, is understood to turn
catastrophic and inhuman in an instant. A steady drip of stories and
videos from areas where the curve has indeed passed that threshold
powerfully reinforces this demand.

This model suggests a different vision of society from the one
neoliberalism projected over the last half-century. It has little
room for market self-adjustments, but rather assumes the visible
hand of intentional collective action (coordinated by the state or
other means). The goal is to purposefully coordinate towards outcomes
rather than just set conditions for individual action generating
“spontaneous order” (to use Hayek’s colourful term). Even more,
it becomes necessary to prevent the market from working properly,
that is, adjusting the prices of, say, medical goods to the sharply
increased demand. What the platform economy euphemistically hailed as
an innovative market model—“surge pricing”— is now seen, more
realistically, as “price gouging”.

What these models show is that society is not too complex to be
comprehended and hence best left to the invisible hand, but can be
understood as a whole, in near-real-time, and there are ways of
shaping society as a whole to create desired outcomes and prevent
undesirable ones. In dealing with the climate crisis, we will need
much more of such thinking.

The dominant way of such a reading of society is through data, as
the bureaucracies of the modern state—the first institutions to
face this problem—began to realize in the mid-eighteenth century.
The need for data is a powerful incentive for tracking and one of
the main reasons why the virus is simply accelerating and expanding
long-existing dynamics in this area. Tracking began long before the
virus and will not stop should the virus be contained. But data does
not speak for itself, particularly not if it is supposed to speak for
the future. Thus, data must not only be collected but also fed into
models, from which curves and predictions can be derived.

Neither data nor models are simply given. Rather, they are made and
thus reflect, unavoidably, conscious and unconscious biases and
the haphazard practicalities of their creation. For example, the
all-important rate of new infections that underlies the powerful model
for “flattening the curve” represents a chaotic and shifting
relationship of the number of actually infected people and the
health system’s ability to test for and verify the illness. The
model—that is, the extrapolation of the curve into the future—is
dependent on a large number of variables, none of which are precisely
known, but must be set according to certain assumptions. Depending on
how exactly these variables are set, the model produces very different
outcomes. Public resources like the “Epidemic Calculator” allow
everyone to tweak a dozen variables underlying this curve. This shows,
quite practically and intuitively, the complexity of the model, the
assumptions that need to be set despite a lack of precise knowledge
(such as “length of incubation period”), and also the possibility
to shape outcomes through social agency aimed at changing the values
of the model (such as “interventions to decrease the rate of
transmission”). The function of such models is not to predict the
future—on the contrary, it is to allow for scenario planning in
order to affect the future.

In many ways, this is nothing new at all. Rather, it comprises the
basic cybernetic model of society. Flows are continuously measured and
tweaked through all kinds of information and material interventions,
and the results of this tweaking are fed back into the model.
Such models are everywhere—they constitute basic management and
governmental tools. What is new is that our data-fied societies have
allowed the creation of such models for an ever-increasing range of
situations and the tightening of feedback-loops at various intervals,
from the sub-second timeframe of the financial markets to the
multi-year rhythms of ecosystem modelling. Social media have pushed
this way of understanding society into everyday life and personal
relationships. Not only by applying them to the intimate sphere
but also by providing users with tools to model their own social
communication in this way, and offering a range of variables that can
be tweaked to create better outcomes. These range from tips on how
to communicate more “effectively”—that is, optimize one’s
behaviour within the existing environment—to the ability to buy
advertisements and other signal boosters, effectively manipulating the
environment to one’s own advantage (at least, that is the promise).

While the basic modelling techniques are widespread, the
epidemiological model that currently dominates society differs in two
crucial ways from most other models driving political and economic
decision-making. First, coming from science, this model is both
transparent and contested. There is a broad professional discussion,
accessible to the public if needed, of these models, the validity of
the variables, how to determine their specific values in particular
case (for example, how to measure the “length of incubation
period” for this type of virus), and the way these variables relate
to each other. At the end of such a discussion, there are multiple
models or variable values that can be compared and judged against each
other. Second, there is a broad discussion about how the knowledge
gained through the model should be turned into social action. if
the model suggests urgent “interventions to decrease the rate of
transmission”, one can discuss which interventions are effective
and socially desirable. Should events be banned, self-quarantining
imposed, proactive masks be worn, tracker apps installed on
smartphones, etc.? Over time, even as the underlying model stays the
same, the answers to these questions will shift—reflecting both
scientific and political considerations. These discussions take place
in public, or at least they will need to be explained and justified to
the public, in order to create a minimum of voluntary participation.

The times where society could operate without data-driven,
feedback-adjusted, cybernetic models are long gone. They are key
instruments for a complex, dynamic society to understand itself.
Today, most of these models function as effective “black boxes”.
Their workings are intransparent and unaccountable, applied from above
without any input from those whose lives are shaped by them. Those who
are able to create and implement such models are able to implement
their own visions of society, their own set of assumptions about what
counts and what does not, and their own assessment of what constitutes
a desirable outcome. These “black-box” models are a key element of
the authoritarian, technocratic character of contemporary politics.

In contrast, the way that epidemiological models are created,
discussed, and the consequences thereof implemented do not necessarily
constitute a new “authoritarianism”, as many liberal commentators
have suggested, but rather can point towards an application of “big
data” that can strengthen rather than undermine democracy. The
explicitness of these models creates an awareness of system-level
dynamics aimed at stimulating debate about means and outcomes, rather
than keeping this knowledge secret and imposing measures from behind
the curtain. We will need much more of such democratic procedures for
doing “big data”.

A Framework for a Democratic Big Data Regime

While the authoritarian potential of the crisis is real, both in
respect to expanded forms of surveillance as well as the broader
political dynamics, there is also the potential to generate a more
democratic use of “big data”. This would require a strong legal
and institutional framework that implements and enforces rules,
ensuring four basic safeguards:

-- Data must be deleted after the immediate purpose of analysis has
been achieved and, if new data has been generated, the process of
collecting this data must end.

-- Analysis must to be restricted to predetermined ends of public
interest. This can be achieved by separating the entity framing the
research question(s), such as a public health institute, and the
entities undertaking the actual data analysis. This is necessary
to prevent mission creep and fishing expeditions based on the
availability of the data.

-- Data must be made available to multiple teams that are completely
independent from one another. This prevents the public from becoming
dependent on the analysis of data providers (say, social media
companies) and also allows for the cross-examination of different
methods of data analysis. Every model has its own biases, and their
justification can only be assessed in comparison to others.

-- Questions, methods, and results of the analysis must to be
published after the fact. This will allow public appraisal of the
efficacy and legitimacy of data gathering and various analyses,
prevents mistakes from being repeated and unnecessarily intrusive
methods from being reused. A clear analysis of the efficacy of big
data analysis could also help to stem the tide of “solutionism”,
where opaque and unproven approaches are sold with widely exaggerated
promises, often diverting investment from more mundane and yet more
proven approaches of public infrastructures.

Such a framework, that acknowledges both the potential of big data to
provide real-time social knowledge as well as the democratic character
of our societies, would not need to be restricted to the current
health crises, but would in fact be highly useful for other big data
questions that will inevitably arise in the future.

| |||||||||||||||||| http://felix.openflows.com |
| Open PGP | http://felix.openflows.com/pgp.txt |

Content-Type: message/rfc822
MIME-Version: 1.0

Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Subject: confirm 81a583635b5ec2a376fb66ab2ec82339896fefb6
Sender: nettime-l-request@mail.kein.org
From: nettime-l-request@mail.kein.org
Date: Fri, 01 May 2020 14:05:00 +0200
Message-ID: <mailman.2967.1588334700.54339.nettime-l@mail.kein.org>

If you reply to this message, keeping the Subject: header intact,
Mailman will discard the held message.  Do this if the message is
spam.  If you reply to this message and include an Approved: header
with the list password in it, the message will be approved for posting
to the list.  The Approved: header can also appear in the first line
of the body of the reply.

#  distributed via <nettime>: no commercial use without permission
#  <nettime>  is a moderated mailing list for net criticism,
#  collaborative text filtering and cultural politics of the nets
#  more info: http://mx.kein.org/mailman/listinfo/nettime-l
#  archive: http://www.nettime.org contact: nettime@kein.org
#  @nettime_bot tweets mail w/ sender unless #ANON is in Subject: