1) Algorithms operate behind the scenes on websites and apps to personalize content like recommendations, search results, and social media feeds based on a user's past activity and shared data.
2) By focusing only on content we agree with, algorithms can lead to "filter bubbles" that isolate us in our views and limit our exposure to different perspectives.
3) Bots and trolls can take advantage of algorithms and cognitive biases to spread misinformation by making biased claims seem widely supported.
4) While our devices do collect significant data on our online activities, they do not actually listen through microphones as that would be inefficient; existing data is enough to effectively target users.
3. What is an Algorithm?
• An algorithm is the set of rules and
instructions that operate behind the scenes
in any electronic device or system that make
it work the way it does.
• The algorithm uses selected data to make
the value decisions its designer wants it to.
OLLI Fall, 2019
4. Algorithms in Action
• Netflix and Amazon’s recommendation
engines (You might like …)
• Your Google search results
• Your Facebook newsfeed (it is not the most
recent posts by your friends)
• Your Instagram feed
• Yes, even your Twitter feed
OLLI Fall, 2019
5. Algorithms Feed on Data
• In order to serve up the most personalized and
best possible results, an algorithm needs LOTS
of data.
• That data may be collected via cookies that
track your activity across the internet, or
existing data that is shared among companies
(Google, Apple, Facebook, Twitter, Amazon,
advertising networks, banks, supermarkets…)
OLLI Fall, 2019
6. Social Media Algorithms
• On Facebook, and more recently on Twitter
and Instagram, algorithms are used to
predict what you most want to see or are
likely to find interesting, based on what you
have liked and interacted with before.
• The specific formulas they use are top-
secret and the public does not know what
they favor or minimize.
OLLI Fall, 2019
7. Sounds Good to Me …
• Algorithms that personalize your experience
online make it easier for you to find what
you’re probably looking for. Customers
prefer and appreciate it!
• When you get positive results that make you
stay on a platform longer, the platform can
sell more ads targeted to you.
OLLI Fall, 2019
8. So?
• You won’t see totally irrelevant ads! Cool!
• You will see more of what already interests
you and that you agree with. Cool!
• There is no mechanism to tell you whether
what you are seeing is accurate, well-
sourced, or the full story. Uh oh.
OLLI Fall, 2019
9. What They Know About You
• Facebook and Apple and Google and
Amazon (and your bank and your apps)
collect and share data (for vast quantities of
money) about you to feed the algorithms.
• You phone isn’t listening to you – it doesn’t
have to (and it would be inefficient). It
simply finds patterns quickly across
everything you do in your day to day life.
OLLI Fall, 2019
11. What’s a Filter Bubble?
• The online corollary of an “echo chamber”
• Term coined by MoveOn.org president Eli
Pariser
• Filter Bubbles result from algorithms that
serve up information we are likely to agree
with and hide what might cause us cognitive
dissonance.
OLLI Fall, 2019
12. Weaponizing Cognitive Biases
• We also reinforce our filter bubbles by
unfriending people we disagree with, or
failing to interact with content we are not
interested in or which offends out
sensibilities
– In this way, our cognitive biases are dictating
the information the algorithm uses, further
isolating us
OLLI Fall, 2019
14. Facebook is a
Filter Bubble Engine
• Watch this ad for Facebook Groups
– FB Groups are usually closed communities
around a given topic or common trait such as
…
• FB groups for Middle-Aged women
• FB groups for people who grew up in Marin
• FB group for organizers of immigration policy
protests
OLLI Fall, 2019
17. Bias Feedback Loops
• An algorithm is only as good as the humans
who create them. None is fundamentally
objective
• Humans decide what data to feed an
algorithm, and those choices are influenced
by their own biases and limitations of
experience
OLLI Fall, 2019
20. What’s a Bot?
• An algorithm-based piece of software that
can, in some cases, mimic a human being
online, under a fake account
• Bots takes advantage of the algorithms and
their biases underlying existing platforms,
like Twitter and Facebook.
• Bots can be designed to amplify & spark
viral sharing of misinformation
OLLI Fall, 2019
22. Bots Make Biased Information
Seem Legitimate
• Your reliable and thoughtful friend shared
it!
• That friend only shared it because millions
of others shared it, giving it the sheen of
confirmation (we trust our friends).
Aristotle’s “Ethos”
• Millions of others only shared it because …
millions of others shared it.
OLLI Fall, 2019
24. Kate Starbird
“Perhaps the most dangerous misconception is that
disinformation targets only the unsavvy or
uneducated, that it works only on ‘others’.
Disinformation often specifically uses the rhetoric
and techniques of critical thinking to foster nihilistic
skepticism. … And we may have trouble seeing the
problem when content aligns with our political
identities.
…
OLLI Fall, 2019
Kate Starbird
“Perhaps the most dangerous misconception is that
disinformation targets only the unsavvy or
uneducated, that it works only on ‘others’.
Disinformation often specifically uses the rhetoric
and techniques of critical thinking to foster nihilistic
skepticism. … And we may have trouble seeing the
problem when content aligns with our political
identities.
…
OLLI Fall, 2019
25. Kate Starbird, cont’d
“Disinformation campaigns attack us where we are most
vulnerable, at the heart of our value systems, around
societal values such as freedom of speech and the goals of
social-media platforms such as ‘bringing people together’.
As individuals, we need to reflect more on how we interact
with information online, and the goals of social-media
platforms such as ‘bringing people together’. As
individuals, we need to reflect more on how we interact
with information online, and consider that efforts to
manipulate us may well be coming from within our own
communities.”
OLLI Fall, 2019
27. How to Detect a Bot
• Click on the photo and download it, then do
a reverse image search (via Google Image
Search or TineEye.com)
• See if you have friends in common (you
may, your friends may not have checked it
out)
• See if they post too much, or only or
primarily share others’ posts
OLLI Fall, 2019
28. Is My Phone/Alexa/Etc.
Listening to Me?
• Short answer – no.
• Longer answer — it doesn’t have to, and
listening would be less efficient. It already
knows more about you than you think.
• “Big Data” is an enormous industry. By
some measures, it is more valuable than
fossil fuels.
OLLI Fall, 2019
29. How Big Data/Data Mining
Works
• Everything you do – your searches, your
location data, your friends and their actions,
every time you swipe a credit card, tag a
friend, etc., data gets generated.
• That data can be compared to existing data
and “mined” to target you for ads or certain
types of content.
OLLI Fall, 2019
30. How to Better Control Your Data
• Facebook/Instagram: Go to Settings >
Your Facebook Information > Access My
Information
• Google/YouTube: Go to My Account >
Data & Personalization > Pause all settings
• Browser: Install Privacy Badger, AdBlock,
and/or uBlock Origin
OLLI Fall, 2019