Working with algorithms: Practices of autonomy and heteromation [Algorithmic Practices, Session II]

Abstract

The stories we tell about algorithms in relation to labour are, frequently, binary: we are either the networked proletariat, blindly staggering into algorithmic servitude (i.e. the microwork of the crowd-sourced taxi system 'Uber' or Amazon's Mechanical Turk); or, we are the heirs of a new 'wealth of networks' (following Yochai Benkler), set to be relieved of monotonous work and freed to exercise creativity. These are, of course, familiar stories, which were told of the automation of mass production. At the heart of both stories is the attribution of a mythical autonomous agency to the unseen machine. As Lucy Suchman (2006) has argued, this is the enchantment of technologies through the obfuscation of the labour of production. However, neither the algorithm or the worker are autonomous. Rather, they are bound into a quasi-autonomous system (what Rob Kitchin calls a 'coded assemblage') that distributes work tasks to a labour force and vice versa. The rules of such systems may be opaque but they have authors (programmers, managers, service designers) and they are maintained, which both involve labour. Not only, then, are the outcomes of algorithmic labour practices distributed, but so are their creation and maintenance. This is perhaps a muddying of 'immaterial labour', but is nevertheless a powerful illustration of the new kinds of proletarianisation identified by the philsopher Bernard Stiegler. To understand this algorithmic 'agencement' this paper argues that what is required is a rethinking of agency, not in terms of automation but what Bonnie Nardi has called 'heteromation': technical systems that push critical tasks to end users as indispensable mediators. The aim of this paper, therefore, is to interrogate this distributed performance of knowledge and skill, the cost of which is arguably a loss of individual autonomy and skill, as the rise of the new 'proletariat' (following Stiegler) of technocultural 'composite workers'.

Full text

Working with algorithms:
"a cybernetic ecology where we
are free of our labours”?
Sam Kinsley
University of Exeter
RGS-IBG 2015

This talk
i. Stories about algorithms & progress
ii. Working with algorithms
– Automation
– Heteromation
iii. How can we study these changes to
‘work’

Algorithms
Recognising here that algorithms are:
Synecdoche
– refer to the complex
sociotechnical ensembles that perform
particular computational processes
Emblem
– of opaque processes of
automation

Technology and work
A story…
We are either:
the ‘algorithmic’ proletariat;
or, the beneficiaries of a new ‘wealth of
networks’.

All watched over…
I like to think
󲻌󲻌(it has to be!)
of a cybernetic ecology
where we are free of our labors
and joined back to nature,
returned to our mammal
brothers and sisters,
and all watched over
by machines of loving grace.
Brautigan, 1967

Long-standing geographical imagination

An economic story
Temptation to see automation as
reduction of labour
but… it’s not so simple.

An economic story
A return to ‘Post-Fordism’:
– ‘Offshoring’ & casualisation rather than
‘automation’
– Market mediated employment, anti-
unionism, individualisation
– Work intensification ~ neo-Taylorism
…but specifically in relation to ‘service’ labour

Different kinds of labour
In the same system BOTH:
Heavy vertical integration, Just-in-Time, ‘lean’
practices
AND (not OR)
Flexible, specialised, ‘offshore’, individualised,
market-mediated labour

Amazon
“
an algorithmically managed
infrastructure company
”
— James Bridle
A sociotechnical assemblage to manage
customers, buildings, other companies,
stock, workers etc. for profit

Automation in ‘customer fulfilment’
Arm-mounted terminals managed by warehouse
management systems (programmes)
– Instructs pickers/ loaders
– Surveils pickers/ loaders
– Constructs a measure of productivity

Automation in ‘customer fulfilment’
Amazon:
– Daily targets: load/pick
n
per hour
– Targets keep rising
– ‘Three strikes’
– Checks for unscheduled breaks
Policy/rules are coded: programmatic
policy directs (lower-level) management

Turkers: labour ‘as a service’
Amazon ‘Mechanical Turk’
Platform for microwork, or ‘Human
intelligence Tasks’

Turkers: labour ‘as a service’
Amazon ‘Mechanical Turk’
“You’ve heard of software-as-a-
service. Now this is human-as-
a-service” —Jeff Bezos, 2006
Establishes ‘workers as a form
of infrastructure, rendering
employees into reliable
sources of computational
processing’ (Irani &
Silberman, 2013)

Consumer as worker
Software
produces
new kinds of labour
The data-driven economy relies on the
labour of the consumer
– reviews & ratings
– captured habits
– identifying problems

Heteromation
✶ Automation ~ oriented towards
machinic uniformity
E.g. the production line & ‘computer says no’
✶ Heteromation ~ technical systems
that function through the acts of
heterogeneous actors (Ekbia & Nardi,
2014)
E.g. Mechanical Turk, Uber… Facebook &
Citizen Science?

Heteromation
“heteromated technology fills the gap created by
automation, but with a vengeance that unsettles
established mechanisms of reward…”
(Ekbia & Nardi, 2014)
‘Computational thinking’ (Wing, 2006)
applied to labour(?)
= the ‘precariat’ as (lumpen)proletariat?

Studying auto–/hetero– mation
This is as much about geographies of labour &
work as it is about code
It is not sufficient to study code alone
– Need to study the full sociotechnical
assemblage:
• how code & protocols become policy
• where/how code ‘acts’
• where/when code is not the principal actor
But it is difficult to identify these situations…
– Phenomena are often proprietary

Studying auto–/hetero– mation
Nevertheless, these phenomena are performative
– So: opportunities to intervene
Perhaps we ought:
• to avoid fetishising ‘algorithms’,
‘algorithmic–’ (pace Chun 2011)
• to be reflexive about the ‘enchantment’ of
technological projects
– Look to particular materialities and contexts
• to investigate the ‘affiliative’ powers of code
within the ‘agencement’ (pace Suchman, 2005)

Thank you.