experiment might have on society. For a study on
voting, this might be an electoral risk. For a study on
fake news, it might be decreasing trust within society
in real news. For a study on manipulating people’s
emotions, it might be the emotional wellbeing of the
The second recommendation is that ethics approval may be needed before receiving or handing
over data from an experiment run in the past. You
may, for instance, need to obtain informed consent
to hand over data to a third party even though there
may be no additional cost to the individual to pass
on their data once it has been collected. Similarly,
you may not be able to receive data from someone
else without seeking informed consent for this new
use of the data. The only circumstances that informed consent may not be needed when handing
over or receiving data are when an appropriate IRB
gives a waiver. If such a recommendation had been
in place, the review board at Cornell would not have
been able to conclude that approval was unnecessary
for the analysis of the already collected data in the
Facebook news feed experiment.
The third recommendation is that subjects of any
experiment may need to be informed directly after
the study about the results and their participation.
This is especially important when informed consent
has been waived and the participants were not aware
that they had been the subject of an experiment. It
might not be adequate simply to publish the results
in the scientific literature or even the wider media.
Subjects might have to be contacted directly.
Had such recommendations been in force in
2010, the IRB at the University of California, San
Diego might still have granted a waiver to consent
for the Facebook experiment on voter participation.
However, the board might have required a demo-graphical balance to minimize the electoral impact,
and might have required Facebook to e-mail every
participant after the election with a summary of the
outcome. Had these recommendations been in force
in 2012, the IRB at Cornell might still have granted
a waiver to consent for the Facebook experiment on
news feed manipulation. However, they might have
required informed consent from the participants, as
well as a followup after the experiment to check on
their emotional wellbeing. Had these recommendations been in force in 2014, and had Aleksandr
Kogan applied for ethical approval to collect psychometric data in an academic study, he might have been
prohibited from sharing this data with Cambridge
These recommendations are not the final word on
how ethical approval needs to take account of new
technologies like social media. It may, for instance,
also be desirable if companies running experiments
on social media go through a similar ethics approval
process as public institutions.
The author acknowledges support from the Euro-
pean Research Council (AMPLify Advanced Grant
670077) and the Asian Office of Aerospace Research
and Development (Grant FA2386-15-1-4016). The
author thanks James Fowler for providing copies of
the documentation submitted to the IRB at the Uni-
versity of California, San Diego for approval.
1. Michael Buckley, Facebook’s vice president for global business communications, was quoted in Mother Jones as saying
“We’ve always implemented these tests [to increase voter
participation] in a neutral manner. And we’ve been learning from our experience and are 100 percent committed
to even greater transparency whenever we encourage civic
participation in the future” (Sifry 2014).
Bond, R. M.; Fariss, C. J.; Jones, J. J.; Kramer, A. D. I.; Marlow,
C.; Settle, J. E.; and Fowler, J. H. 2012. A 61-Million Person
Experiment in Social Influence and Political Mobilization.
Nature 489(7415): 295–8. doi.org/10.1038/nature11421.
Dritschilo, G. 2010. Recount Upholds Results, Raises New
Question. Rutland Herald November 30. www.rutlandherald.
Duggan, M.; Ellison, N. B.; Lampe, C.; Lenhard, A.; and
Madden, M. 2014. Demographics of Key Social Networking Platforms, Social Media Update 2014. Washington, DC:
Pew Research Center. www.pewinternet.org/2015/01/09/
Kramer, A. D. I.; Guillory, J. E.; and Hancock, J. T. Experimental Evidence of Massive-Scale Emotional Contagion
Through Social Networks. 2014. Proceedings of the National
Academy of Sciences of the United States of America 111( 24):
Sifry, M. L. 2014. Facebook Wants You to Vote on Tuesday.
Here’s How It Messed With Your Feed in 2012. Mother
Jones October 31. www.motherjones.com/politics/2014/10/
Verma, I. 2014. Editorial Expression of Concern: Experimental Evidence of Massive-Scale Emotional Contagion
Through Social Networks. Proceedings of the National Academy of Sciences of the United States of America 111( 29): 10779.
Vermont Public Radio. 2011. Lawmakers Uphold Buxton’s
One-Vote Victory. Vermont Public Radio (VPR News) January 28. archive.vpr.org/vpr-news/lawmakers-uphold-buxtons-one-vote-victory/.
Toby Walsh is a guest professor at TU Berlin, and the Sci-entia Professor of Artificial Intelligence at the University of
New South Wales and Data61. He has a BA from the University of Cambridge and an MSc and a PhD from the University of Edinburgh. He is a Fellow of the Australia Academy of
Science and of the Association for the Advancement of Artificial Intelligence, a Humboldt Award winner and recipient
of the New South Wales Premier’s Prize for Excellence in
Engineering or Information Communication Technology.