2. Motivation and objectives
• Future ICT systems as sophisticated assemblies of data-intensive, complex automation and
deep community involvement
• Defining social machines and their characteristic properties as necessary step towards a
principled understanding of the science and engineering of such systems
• Objectives of this work
– Identify and define the constructs to describe, study, and compare social machines
– Achieve a shared understanding of basic notions and terminology through involvement
from the broader community
• Useful tool for both researchers in social and computer sciences and for developers and
operators of existing and future social machines
2
3. General considerations
• Machine: ‘(1) an assemblage of
parts that transmit forces,
motion, and energy one to
another in a predetermined
manner; (2) an instrument (as
a lever) designed to transmit or
modify the application of power,
force, or motion’ [Merriam-
Webster]
• In relation to living beings: ‘one
that resembles a machine (as in
being methodical, tireless, or
consistently productive)’
[Merriam-Webster]
• Social machine
1. co-existence of and interaction
among algorithmic and social
components;
2. problem/task specification changes
as the system evolves;
3. operation of the system is governed
by a different set of rules;
4. different performance models and
approaches to measure them;
3[Courtesy of Dave de Roure]
4. The polyarchical relationship of social machines
• Platforms/technologies vs social machines created for specific
purposes. E.g., MediaWiki vs Wikipedia
• Broader vs narrower-scoped social machines. E.g., Twitter vs Obama’12
• Ecosystem of social machines. E.g., results from GalaxyZoo taken up in
Wikipedia articles
4
5. Social machines and related areas
• Computer science:
CSCW, social computing,
human computation
• Organizational
management/social
sciences: wisdom of the
crowds, collective
intelligence, open
innovation,
crowdsourcing
5
6. Social machines and related areas (2)
• Who defines the task/purpose of the system
– The system designer vs community
• What kind of tasks do humans undertake
– Creative vs computationally expensive
• Who is supporting whom
– Humans supporting algorithmic processes or machines
supporting human tasks
6
7. Methodology
• Repertory grid elicitation to derive an initial set of elements
(instances of social machines) and constructs (characteristics
of social machines) 10 grids, 56 elements, 117 constructs
• Consolidation and clustering of constructs 31 constructs, five
clusters
– General description
– Purpose and tasks
– Participants and roles
– Motivation and incentives
– Technology
7
8. Purpose and forms of contribution
• Contributions towards public vs private good
• Implicit vs explicit contributions
• Degree to which contributors decide what they can work on
• Degree to which contributors can change the
nature/purpose/development of the social machine
• How is the final result created/aggregation
8
9. Participation and interaction
• Who can contribute and what: roles, requester/worker,
game models, skills and learning curve
• Workflow management: task/resource assignment
(scarcity, requester-contributions cardinality),
parallelization, synchronization, aggregation
– Machine replacing/assisting humans vs humans
replacing machines
• Dynamics of participation model
9
10. Quality and performance
• Which contributions are validated
• Is there a ground truth and where does it come from: no one,
community, dedicated group, machine owner
• How is quality assessment performed: manually,
agreement/voting between participants, computed automatically
• Are criteria and quality control methods explicit/transparent
• Can contributors change the criteria or earn the right to perform
evaluations
10
11. Motivation and incentives
• Altruism
• Reciprocity
• Community
• Reputation
• Autonomy
• Entertainment/Fun
• Intellectual challenge
• Learning
• Competition
• Payment/Rewards
• Depend on
– Nature of the good
produced
– Goal
– Nature of the
contributions
– Existing social structure
11
12. Technology and engineering
• Requirements specification and evolution
• Security, trust
• Decentralization
• Data ownership and access
• Profile building
• Social networks
• Analytics on top of social network and actual data
12
[Courtesy of Dave Robertson]
13. Next steps
• Consolidate and use the classification
• Evaluation
– Task-independent using criteria from knowledge
engineering (completeness, correctness, readability,
redundancy etc)
– Task-dependent: Can the framework be used to describe
existing social machines?
13
15. Constructs: purpose of the system and contributions
• Purpose of the system, types of contributions, degree to
which these change
15
16. Constructs: people, roles, motivation
• Types of audience, autonomy and anonymity, roles and role
hierarchies
• Intrinsic vs. extrinsic motivation, rewards
16