Keynote presented at Locked out of Social Platforms: An iCS Symposium on Challenges to Studying Disinformation, IT University, Copenhagen, 27 Oct. 2018.
Pushed towards Dysfunction: How Social Media API Restrictions Distort Research Outcomes
1. @qutdmrc
iCS Symposium, Copenhagen, 27-28 Oct. 2018
Axel Bruns | @snurb_dot_info
Pushed towards Dysfunction
How Social Media API Restrictions Distort Research Outcomes
5. C-SPAN: “Cambridge Analytica and Data Privacy, Alexander Nix Testimony”
https://www.c-span.org/video/?446535-1/cambridge-analytica-ceo-alexander-nix-testifies-data-misuse
6. Sydney Morning Herald: “Facebook, Twitter, Google grilled over elections ads at Russia inquiry”
http://www.smh.com.au/world/facebook-twitter-google-grilled-over-elections-ads-at-russia-inquiry-20171031-gzcbw7.html
7. The Irish Sun: “What did Mark Zuckerberg tell Congress and did the Facebook CEO address the Cambridge Analytica data leak in his testimony?”
https://www.thesun.ie/news/2426061/what-did-mark-zuckerberg-tell-congress-and-did-the-facebook-ceo-address-the-cambridge-analytica-data-leak-in-his-testimony/
8. @qutdmrc
API, Disappearing
● Facebook API, v2.6:
“How often has this (fake?) news URL been shared?”
(https://developers.facebook.com/tools/explorer/)
● Facebook API, v2.9:
“What does Facebook think this URL is? What internal ID does it have?”
● Facebook API, v3.0:
“¯_(ツ)_/¯”
14. @qutdmrc
How APIs Shape Our Research
● “Easy Data, Hard Data” (Burgess & Bruns 2015):
● Easy data:
● Individual pages (Facebook)
● Hashtags and keywords (Twitter)
● Hard data:
● Multiple pages and their connections (Facebook)
● Profile activity and information flows (Facebook)
● Ordinary, unremarkable posts (Twitter)
● Everyday @mention and follower networks (Twitter)
● Complex, composite data across multiple platform affordances
API changes reduce our access to hard data, offering only easy data
16. @qutdmrc
Pushed towards Dysfunction
● Pushed towards public interactions:
● Overrepresentation of Twitter in the literature
● Focus on highly visible hashtags and pages, not personal interactions
● Isolated discursive islands in an ocean of social media activity
● Pushed towards extreme, self-selecting cases:
● Hyperpartisan hashtags, fringe community pages
● Patterns of dysfunction that misrepresent the platforms and their uses
● Pushed towards skewed, ill-considered emphases:
● Viral contagion – only 18% of all Australian tweets are hashtagged
● Echo chambers – social media are engines of context collapse
● Filter bubbles – extremists are especially heavy users of mainstream news
● Fake news – mostly consumed by a small percentage of hyperpartisans
19. @qutdmrc
The “King-Persily Paper”
● A new model:
● “the new paradigm for industry-academic partnerships in the King-Persily
paper” (Social Science One 2018)
● Weird notion of social science:
● “Social science insights are normally about population averages and broad
patterns, and for which facts about any one individual are unnecessary and not
of interest. Social scientists are usually interested in patterns about everyone,
not anyone in particular.” (King and Persily 2018: 8)
● Company interests outweigh scientific freedom:
● “The optimal way forward … is to find research questions that are of intellectual
interest to the scientific community and either provide valuable knowledge to
inform product, programmatic, and policy decisions, or are orthogonal to
company interests.” (King and Persily 2018: 12)
20. @qutdmrc
Social? Science? One?
● What is Social Science One?
● U.S.-centric
● Nebulous relationship with Facebook
● Limited involvement from media, communication, and related fields
● Intransparent application process controlled by self-appointed assessors:
● “the co-chairs at Social Science One … make final substantive decisions
about which proposals to support” (Social Science One 2018)
● Strange understanding of the scientific process:
● “Researchers are only permitted to perform the analyses, and estimate the
quantities, that are proposed and approved; new types of analyses and
quantities to estimate require additional applications.” (Social Science One
2018)
21. @qutdmrc
6 of 1300 submissions selected
Not Just Facebook
2 of 230 (?) submissions selected
(https://blog.twitter.com/engineering/en_us/a/2014/twitter-datagrants-selections.html
https://blog.twitter.com/official/en_us/topics/company/2018/twitter-health-metrics-
proposal-submission.html)
29. @qutdmrc
Four Paths
1. Give up. Walk away. Research something else.
● Leaves the platforms unscrutinised
● Leaves their users to fend for themselves
● Provides no input to regulators when they need it most
● The haters, abusers, trolls, bots, Russian hackers won’t walk away
30. @qutdmrc
Four Paths
2. Keep pushing for different data access regimes.
● We’ve tried, with limited success – what leverage do we have?
● How do we address legitimate concerns about user privacy?
● What alternative models of data access can we propose?
● How might platforms be compelled to implement them?
● What role can national legislators play?
● How do we get them on our side?
31. @qutdmrc
Four Paths
3. Bite the bullet and pay for commercial data access.
● Largely unaffordable for individual researchers and projects
● Difficult to find institutional support for ongoing access
● Commercial data access remains shaped around ‘easy data’
● API changes also undermining commercial data sector
● Likely requires consortia-based access models
● Scholarly data sharing remains a grey area
32. @qutdmrc
Four Paths
4. Explore scraping and other unsanctioned data access methods.
● Probably explicitly against platforms’ Terms of Service
● Which may not be legally sound and binding, however
● Could be sanctioned by ‘freedom of research’ legislation
● May require strong institutional support
● Could still be frustrated by platform countermeasures
● May not generate good-quality data
33. @qutdmrc
Four Paths, One Destination
● All of the above:
● Social media platforms are too important to be left unscrutinised
● Critical, independent, public-interest research is crucial
● Shaping of research limits by platform directives is unacceptable
● Terms of Service cannot trump public interest
● But also:
● Critical self-assessment of our uses of data
● Diligent adherence to established ethical standards
● Careful management of data and sharing practices
● We must demonstrate that we can be trusted with the data
We must press home our message whenever we get the chance
35. @qutdmrc
Cato the Elder, ca. 157 BCE
Ceterum censeo Carthaginem esse delendam.
Furthermore, I propose that Carthage is to be destroyed.
36. @qutdmrc
Social Media Researchers, ca. 2018 CE
Furthermore, we demand that social media platforms
provide data access to critical, independent, public-interest research.
37. @qutdmrc
iCS Symposium, Copenhagen, 27-28 Oct. 2018
Axel Bruns | @snurb_dot_info
@snurb_dot_info – http://snurb.info/
@socialmediaQUT – http://socialmedia.qut.edu.au/
@qutdmrc – https://www.qut.edu.au/research/dmrc
This research is supported by the ARC Future Fellowship project
“Understanding Intermedia Information Flows in the Australian
Online Public Sphere”, and the ARC LIEF project “TrISMA:
Tracking Infrastructure for Social Media Analysis.”