22 | New Scientist | 14 September 2019C
OME on Facebook, give us
the evidence. Months after
the company said it would
hand over data to help determine
if the social media site really can
affect election results, we are still
waiting. Now it looks as if the
whole project could collapse.
Facebook’s electoral influence
has come under much scrutiny
in recent years. Accusations
of fake news, allegations of state
meddling and the use of private
data for political purposes in the
Cambridge Analytica scandal have
knocked the company’s image.
It acknowledged as much in
April 2018 when it announced that
it would allow researchers to study
JOSIE FORDsocial media’s impact on elections
Comment
Timothy Revell is New Scientist’s
assistant news editor. Follow
him on Twitter @timothyrevellViews
The columnist
Graham Lawton on
when to stand up for
sacred values p23Letters
Health apps need
to be a regulated
public good p26Aperture
An epic model
that could save the
Mississippi delta p28Culture
Can capitalism be
transformed into
a force for good? p30Culture columnist
Chelsea Whyte
enjoys crowdsourced
medicine on TV p32using anonymised user data. “The
last two years have taught us that
the same Facebook tools that
help politicians connect with their
constituents ... can also be misused
to manipulate and deceive,”
the firm wrote in a blog post.
The question is: how much?
Just because people see a political
post or advert on Facebook doesn’t
mean it affects how they think or
vote – even if they share, comment
on or “like” it. Studies before have
shown correlations. Facebook’s
offer seemed to promise the data
for a full post-match analysis.
In April this year, the Social
Science Research Council (SSRC),
a US non-profit organisation
administering the initiative,announced the first research
projects. They included one
looking at the impact of people
sharing fake news, another
measuring the extent of
disinformation campaigns
by the Internet Research Agency
in Russia and a third assessing
the spread of polarised content.
The projects span elections
in countries from Europe,
Asia, North America and
South America.
Yet, months on, researchers are
still waiting for much of the data.
Now, as BuzzFeed News revealed
last month, the SSRC has said that
unless the data is handed over by
30 September, the funders backing
the project will withdraw.Nobody said this was going
to be easy. There are privacy laws
to navigate – Facebook cites the
recent EU data protection
regulations, GDPR, as one such
problem. Issues around how
anonymous anonymised
information actually is means that
if Facebook gives away too much
data, it could easily be linked to
individuals, but if it doesn’t give
enough the data won’t be useful.
The company has had its fingers
burned before, too. It used to be
easy for researchers to get their
hands on Facebook data. Privacy
breaches and the Cambridge
Analytica affair were the result.
Those stumbling blocks
shouldn’t derail the project.
More than 2 billion people use
Facebook each month and the
alleged influence on elections
cuts to the heart of democratic
values. Facebook has just over
a month to put its best brains on
the case. It should share as much
data as possible, being transparent
in how it does so, while still
maintaining people’s privacy.
If there are genuinely
unsolvable issues with sharing
the data, the public deserves to
know: in its public statements
so far, Facebook barely admits
there is anything wrong.
The firm’s motto was once to
“move fast and break things”.
Now that things are broken,
will it move fast to fix them? ❚Facebook, hand it over
We are still waiting for promised data to help us determine
how social media inf luences elections, says Timothy Revell