Honeypot profiles and malevolent e-reputation attacks on Facebook
1. Honeypot profiles and
malevolent e-reputation attacks
on Facebook
Nasri Messarra, Anne Mione
Université de Montpellier 1 – MRM
International Network for Social Network Analysis (INSNA) Conference, Sunbelt XXXIV,
Tampa, Florida, USA, 2014
2. Background
Most SN studies in general & OSN studies in particular work on
already existing networks to find influencers (Trusov, Bodapati &
Bucklin), hubs, bridges, etc., and restructure them (Valente)
Social bots and fake profiles have one objective: maximizing the
number of friends and then, eventually studying the resulting network
(Boshmaf, Muslukhov). They are in majority based on sexual attraction
(Barracuda statistics) and are used as carriers of information – not an
influencers (empiric)
Recent studies show that size does not matter (Scarpi) and that one of
the keys to success is the initial seeding population (Liu-Thompkins)
We believe that Facebook is not really a non-directed network (in &
out degrees matter)
3. How activists attack on Facebook
1- Post directly on the
brand page
2- Post on their own social networks and
Facebook communities and wait for the viral
effect to reach the brand
Reported cases & littérature about Nestlé, Pampers, DKNY,
Marie-Claire, Capri Sun, Cooks Source, Bershka…
4. Weaknesses of both methods
1- Post directly on the
brand page
2- Post on their own social networks and
Facebook communities and wait for the viral
effect to reach the brand
Exposure of consumer posts on
the page was reduced since FB
has chosen to put all customer
comments in a small discrete
area
The user gets banned,
Communication is closed
Creating viral content is
demanding and complex.
Information may never break
through
5. Brand community
On Facebook
(Brand page)
Engaged
fans
Efficient schema for a blitz unstoppable
attack
Personal
timeline
What if a person can become friend
on his own timeline/profile with
engaged fans of a brand?
Diffusion will only need to be
organic, not viral (no need for WOM
for diffusion)
The brand will not have direct control
to ban the user
Engaged
fans
6. Honeypot profiles
Can we attract engaged fans of a brand
Around a high-prestige profile (in degree > out degree) with
common interests and call for reciprocity (the illusion of social
proof)
In order to create an efficient initial seeding population
That can be reached directly through organic communication?
7. Attraction by common interest
The cover experiment
Infiltration of a politician network: Obama, Hariri…, Infiltration of a movement:
Syrian revolution, Infiltration of a star: Ronaldo, Infiltration of a company’s
group – using a profile with a cover photo that reflect a common interest and
without any other content
Sending requests to 100 engaged users - 40% responded positively
8. Attraction through the value of friends
Creating network value by choosing
popular friends
The Spartacus experiment (male, very
small set of photos, standard mid-age
person with kids, etc.) - Different from
social bots: unattractive male, weird
name (cannot be confused with
someone else),
Started with 5 real friends & only
chooses the most popular mutual
friends
After 4 months from the beginning of
the experiment we started receiving
friend requests on a regular basis
(Prestige = 10% after 2 years)
Currently has more than 500 friends
9. Attraction by impersonation
Impersonating a public figure
Zero friend request sent
The impersonator just likes comments on Facebook talking about him and his actions
2 experiments, 5000 friends reached exponentially in a matter of weeks
10. The strategic power of honeypot profiles
Build their network
Which is an optimized seeding population
Of engaged fans of an adversary or friend.
Honeypot profiles
Cannot be banned from a page
Reach fans organically
Can have a very high prestige rate depending on the method used
Have the illusion of social proof, liking and reciprocity.
11. Discussion
We are aware that our experiments on honeypot profiles and malevolent influence
in online social networks raise ethical questions. Yet, we hope that our work will
make brands and consumer more aware of the manipulation they can face on
online social network and help them find ways to immune themselves again such
malevolent attacks.
We’re only seeing the tip of the OSN iceberg. Brands and consumers should be
ready for future and smarter evolutions.
The same experiments can be reproduced with real Facebook accounts and the
findings can be used as well to create engaged communities and improve a
company’s reputation or promote brands in an ethical way.
12. Thank You
Nasri Messarra: nasri@messarra.com
Anne Mione: anne.mione@univ-montp1.fr
Editor's Notes
We know how teens strive to avoid the low friends count humiliation and the effect of friend count on self-estime (Kim & Lee). Opt-in to become a friend with someone should have a value (empiric)
Trusov, M., Bodapati, A. V, & Bucklin, R. E. (2010). Determining Influential Users in Internet Social Networks. Journal of Marketing Research, XLVII(August), 643–658.
Tom Valente, Network Interventions
Kim, J., & Lee, J.-E. R. (2011). The Facebook paths to happiness: effects of the number of Facebook friends and self-presentation on subjective well-being. Cyberpsychology, Behavior and Social Networking, 14(6), 359–64. doi:10.1089/cyber.2010.0374
Scarpi, D. (2010). Does size matter? An examination of small and large web-based brand communities. Journal of Interactive Marketing. Retrieved from http://www.sciencedirect.com/science/article/pii/S1094996809000899
Boshmaf, Y., & Muslukhov, I. (2011). The socialbot network: when bots socialize for fame and money. Proceedings of the 27th …. Retrieved from http://dl.acm.org/citation.cfm?id=2076746
We’re eliminating both weakness
The messenger is the message (Tom Valente)
By the time the brand reports the attack to Facebook, the damage is done (time factor)
Boshmaf, Y., Muslukhov, I., Beznosov, K., & Ripeanu, M. (2011). The Socialbot Network : When Bots Socialize for Fame and Money. University of British Columbia.
Robert Cialdini 6 principles of influence
social bots
The classic method of making friends is creating a sexually attractive profile and using social bots to send massively requests to friends and mutual friends
Weaknesses of the method
“Friends” are targeted almost randomly, not “strategically”. The social bot is trying to maximize its social capital by increasing the number of friends.
“Friends” are attracted by the “looks”: the social bot has no “authority” value.