16 - The (Big) Data-security assemblage: Knowledge and critique - Aradau &
Blanke (2015)
Binnen het debat over beveiligingspraktijken voor Big Data moet er volgens de auteurs meer
aandacht uitgaan naar kritische kennis uit de computer- en informatiewetenschappen. Deze
kennis kan gebruikt worden om veelvoorkomende beweringen van beveiligingsprofessionals
over Big Data aan te vechten.
• Hoe worden de besproken opvattingen over Big Data ontkracht met informatie uit de
computer- en informatiewetenschappen?
- intelligence agencies: big data organizations that employ data-driven methods to anticipate
future dangers
→ a collaboration between social and computer scientists can help go beyond the inscrutability
of algorithmic methods in security practices
- three moves that recast existing critical engagements with data-driven security:
1. from data/metadata distinctions to production of data as a complex epistemic entity
2. from a computational turn in surveillance to the division of labour between humans and
computers in socio-technical assemblages
3. from an underlying logic of algorithms to algorithmic practices and methods in security
analytics
(Meta)data and the remaking of security knowledge
- metadata: a set of data that describes and gives information about other data (content)
→ telephony metadata oozes with meaning, which makes distinction from content problematic
- in knowledge engineering, content is anything that can be expressed digitally
- data-information-knowledge (DIK): hierarchy starts with raw data and systematically builds
information and finally knowledge
- data and metadata both refer to practices of knowledge production, which simultaneously
draw boundaries between structured/unstructured data, information and knowledge
- theories of (meta)data production and the critique of the DIK hierarchy are important moves
that challenge the justifactory discourses of security professionals
Big data as artificial intelligence
- second argument in the controversies about digital surveillance had been formulated in terms
of controlled entries to and views onto data
- mass surveillance: collecting/analysing data about many people instead of individuals
- assumption: there is no surveillance when data is not seen by a human (but just a computer)
→ humans only come at the end of the data processing, so only see very little: better privacy
- new division of labour: humans and machines are brought together in the same infrastructures
to process the data
- topic modelling: unsupervised learning technique which auto-summarizes a collection of
documents into a number of common topics (human-computer assemblage)
- security analytics enrol computers, AI practices and data scientists
Blanke (2015)
Binnen het debat over beveiligingspraktijken voor Big Data moet er volgens de auteurs meer
aandacht uitgaan naar kritische kennis uit de computer- en informatiewetenschappen. Deze
kennis kan gebruikt worden om veelvoorkomende beweringen van beveiligingsprofessionals
over Big Data aan te vechten.
• Hoe worden de besproken opvattingen over Big Data ontkracht met informatie uit de
computer- en informatiewetenschappen?
- intelligence agencies: big data organizations that employ data-driven methods to anticipate
future dangers
→ a collaboration between social and computer scientists can help go beyond the inscrutability
of algorithmic methods in security practices
- three moves that recast existing critical engagements with data-driven security:
1. from data/metadata distinctions to production of data as a complex epistemic entity
2. from a computational turn in surveillance to the division of labour between humans and
computers in socio-technical assemblages
3. from an underlying logic of algorithms to algorithmic practices and methods in security
analytics
(Meta)data and the remaking of security knowledge
- metadata: a set of data that describes and gives information about other data (content)
→ telephony metadata oozes with meaning, which makes distinction from content problematic
- in knowledge engineering, content is anything that can be expressed digitally
- data-information-knowledge (DIK): hierarchy starts with raw data and systematically builds
information and finally knowledge
- data and metadata both refer to practices of knowledge production, which simultaneously
draw boundaries between structured/unstructured data, information and knowledge
- theories of (meta)data production and the critique of the DIK hierarchy are important moves
that challenge the justifactory discourses of security professionals
Big data as artificial intelligence
- second argument in the controversies about digital surveillance had been formulated in terms
of controlled entries to and views onto data
- mass surveillance: collecting/analysing data about many people instead of individuals
- assumption: there is no surveillance when data is not seen by a human (but just a computer)
→ humans only come at the end of the data processing, so only see very little: better privacy
- new division of labour: humans and machines are brought together in the same infrastructures
to process the data
- topic modelling: unsupervised learning technique which auto-summarizes a collection of
documents into a number of common topics (human-computer assemblage)
- security analytics enrol computers, AI practices and data scientists
