MARCH 10 2018 LISTENER 43
GETTY IMAGESA
m I alone in wondering why the list of
things that websites tell me “You may also
like ...” doesn’t include things I actually
would like? There’s a lot of money in get-ting this right. But rather than having a human
army tabulating who watched, read or bought
what, then collating the information and feeding it
back, computer algorithms are behind this.
At work are machine-learning programs that
not only cross-tabulate people’s likes, but also
use neural-style networks to build a broad set of
preferences and potential preferences: things you
mightn’t think you’ll like but will; things you don’t
even know exist because they’re so new.
It’s a phenomenon people are increasingly wor-
ried about, because it calls to mind The Terminator’s
Skynet and HAL’s wilful “I’m sorry, Dave, I’m afraid
I can’t do that” in 2001: A Space Odyssey. YouTube
users, for instance, are querying the logic behind
the selection of videos that automatically loads
when the current one finishes. Even if you die in
front of the screen, they’ll just keep cycling and,
ironically, adding this to the data from which the
algorithm draws.
Following a recent Guardian story about pro-
moted YouTube material in the lead-up to the
2016 US presidential election, Virginia Senator
Mark Warner has warned that the video host may
be a threat to democracy. The Guardian reported a
small increase in conspiratorial, disproportionately
anti-Clinton “fake news” landing in people’s feeds
before the November election.
That’s ironic, if true, given that “fake news” is the
term electoral victor Donald Trump pushed into
A force to be
reckoned with
Are people’s views being
affected by the websites
they’re directed to visit?
the mainstream to describe negative
stories about him, his campaign and,
now, his Administration.
There’s been a lot of lefty-liberal
hand-wringing over the rise of
internet-driven spaces where people
with even the weirdest views can find
kindred spirits, as well as “evidence”
to support their beliefs. People are
worried that these “echo chambers”
insulate the wacky
from the mainstream
and render anyone else
directed there by You-
Tube an uncritical idiot
consumer.
Hyperpartisan “news”
sites (not to mention
Macedonian teenag-
ers just making stuff up
for clicks) have become
all the rage, but they
pre-date the 2016 US
election. Just look at
New Zealand’s Whale
Oil “scandals” of 2014, back when
we were still so idealistic. These have
proliferated, however, like a virus.W
hat would your internet
history look like if you made
it available to me? Again,
researchers are using automation to
gather this data. In the US, you can
sign up to let survey firm YouGov
install a plug-in that will record the
websites you visit and how long you
spend on them and in what order.
Dartmouth College politicalscientist and professor Brendan
Nyhan and his colleagues have
made hay with YouGov data, after
working out how to crunch the huge
numbers it generates. For instance,
he’s particularly interested in news
consumption, but notes that people
use the internet for a lot of other
things. “There’s a lot of porn,” he
says, and there’s a lot of work to sift
and crunch it all.
So, what’s the evi-
dence for the scary echo
chamber? Or fear that
ilovemyfreedom.org and
the like are polluting
democracy and convinc-
ing voters that Trump (or,
in a minority of cases,
Clinton) is The Saviour?
Nyhan’s research is sort of
reassuring. Most people
draw about equally from
news sites across the
political spectrum. But
about 10% of those who read fake
news read only pro-Trump fake news.
Of course, these people are probably
Trumpian to the core, so they’re more
than likely reading news that sup-
ports what they already believe. What
it does mean is that they’re unlikely
to ever swing back to Clinton,
however.
“Do the people who use fake news
also watch more porn?” I ask Nyhan.
He pauses thoughtfully, before saying
he hasn’t looked to see. I bet he will
when he gets back to his office. lby Marc Wilson
PSYCHOLOGY
About 10% of the people
who read fake news read
only pro-Trump fake news.
Dartmouth College
political scientist
Brendan Nyhan.