(The following are Rudolf’s notes made from attending the morning of day 1 of the Comprised Data Colloquium, 28 October 2013)
10am …what APIs do beyond collecting data, APIs open up research on social media. Software studies perspective, trying to study software as a cultural object with affect, imaginings, etc.
APIs are used rather than studied themselves. Protocol must be understood as a management style.
Developers reinforce values of self reliance, enterprising culture.
APIs are far from neutral tools, contain their own politics.
Michael Serres, Genesis
Mediators, quasi objects. Example is soccer ball, isn’t the game itself.
APIs are not intermediaries, mediatory objects.
The potential to produce constitutes the labour.
Sociomateriality, physicality, economic logics at work,
…social media as multivalence machine. Social media produce both vague and stable data. Example is Twitter removing favourites which functioned as bookmarks.
Platforms have a standardized set of activities, define a grammar of action. Predefined activities render social life into comparable and countable numbers. Specific enough to mean something, vague enough to work across multiple audiences.
Third party platforms recontextualized the favourites statistic. Aggregation and use as a popularity measure. Had effects on the main platform.
Multivalent data, Espeland & Sauder 2007, numbers can be recontextualized easily because they decontextualize easily.
Reactivity of rankings, the interface is designed to show that score is volatile and to encourage score increasing activities.
Klout provides financial value for users and communicative value for partners. The score, the number can be used in other circumstances taken into account for hiring people.
Multivalent numbers have different value for different actors involved, Marres 2012
New metrics through 3rd parties allow data to circulate to 4th and 5th parties.
Partible activities, Strathern 1991
Understand platforms as ecologies and ask who can realize what form of value.
…easy data, hard data. Social media as the big data moment for humanities and social science. Computational social science. Study of moods, Macy and Golder, 2011.
Computational turn in new humanities research, from tools to a new paradigm, computation itself becomes part of the study. Shift from close to distant reading, software studies and ANT approaches to new media platforms.
Accessing Twitter data is entangled with the politics and commercial aspects of the api. Conflict between research and commercialization of data.
Exploratory data driven research is difficult to sell to most funding bodies. Using propriety data for research, more of it is done internally for commercial purposes.
Mostly the talk covered what still needs to be researched and what data is required for it, which data is hardest to collect?
11:44am…computer based approach to analysis of social media. Multimodality includes all communicative components. How can we analyse multimodal data online. Describe social relationships qualitatively rather than quantitatively. From manifest content to latent content.
Atlas.ti for analysis
Picture analysis in social networks: anonymity problem and quantity problem
Text analysis: platform problems data access, multilingual, multi linear.
…data gathering and how it’s used in particular literature, engagement and mobilization. Over reliance on quantitative data driven approaches.
The terms engagement and mobilization lack clarity. One definition talks about it as a governance norm. Many studies don’t precisely define engagement or use a narrow scope, such as only looking at computer use and voter turn out. Others use more inclusive definition, example is politics such as using bumper stickers, joining a party, voting, attending a party meeting, etc.
There are skeptics like Putnam that claim there is a decline in social capital that used to be spent on other activities.
Others subscribe to normalization theory.
Finally there’s the optimists, the utopians.
The nuance and complexity is hidden behind quantitative studies. Most longitudinal and long term focused studies are focused on counting hashtags, followers, etc.
Over reliance on quantitative data produces studies that apply to narrow scopes and situations.
We see heightened levels of voter disengagement but at the same time we hear about social media platforms democratizing society further. We risk fetishizing the technology.
Weak democracy, deliberative democracy, monetarial(?) citizen
Innis, McLuhan, Dallas Smythe.
Technology is a myth. Science, capital, tools, technology, propaganda.
Need to acknowledge that these technologies are political, need to make sure human agency doesn’t fall by the wayside, the danger of positivism.
Not arguing for replacing data driven research but juxtaposing data driven with qualitative research.