Weaponizing Cyberpsychology & Subverting Cybervetting: For Fun, Profit & Subterfuge
Video in TIB AVPortal:
Weaponizing Cyberpsychology & Subverting Cybervetting: For Fun, Profit & Subterfuge
Formal Metadata
Title 
Weaponizing Cyberpsychology & Subverting Cybervetting: For Fun, Profit & Subterfuge

Title of Series  
Author 

License 
CC Attribution 3.0 Unported:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. 
Identifiers 

Publisher 

Release Date 
2013

Language 
English

Content Metadata
Subject Area  
Abstract 
Almost everything we do in life leaves a personality footprint and what we do on social networking sites like Facebook is no exception. During this talk we will examine: * What it is possible to determine about someone's personality from their facebook activity * What to look for when you are trying to identify the most pwnable person in a group * Whether facebook activity can indicate a high probability of having or developing depression * How you could weaponize 'sockpuppets' by giving them certain personality traits * Cybervetting and your rights (or lack of rights) to privacy * Steps you can take to manage or even alter your 'NetRep' (online reputation) We conducted a research project called 'The Big 5 Experiment' with the objective of determining whether there were any significant correlations between a user's facebook activity and their answers to a personality questionnaire called 'The Big Five Inventory'. The Big Five Inventory was created by Prof Oliver John, to measure personality dimensions known as the Big Five. Considering the ubiquity of personality tests such as the MyersBriggs for employee selection and the growing number of companies adding cybervetting to their selection processes, it can only be a matter of time before we see the two activities merge and at what cost to society? You should leave the talk with an insight into how the Big 5 Experiment results could be used in attack and defense strategies. Should you wish to conduct your own research, related or not, you should also learn from what proved a rather fascinating experience in carrying out the experiment.

Related Material
Video is accompanying material for the following resource
00:00
Game theory
Family
00:48
Cybersex
Point (geometry)
Category of being
Facebook
Selforganization
Bit
Quicksort
Musical ensemble
Information privacy
01:55
Facebook
Projective plane
Bit
Theory
Twitter
02:47
Slide rule
Facebook
Statistics
Information
Decision theory
Quicksort
03:32
Digital photography
Universe (mathematics)
Computer science
04:08
Term (mathematics)
Computer science
Perturbation theory
04:56
Multiplication sign
Bit
Quicksort
Tunis
Wave packet
05:38
Type theory
Characteristic polynomial
Quicksort
Extension (kinesiology)
06:42
Type theory
Context awareness
Estimation
Matrix (mathematics)
Selforganization
07:38
Type theory
Graph (mathematics)
Observational study
Agreeableness
Right angle
Open set
Endliche Modelltheorie
Spectrum (functional analysis)
08:43
Different (Kate Ryan album)
Multiplication sign
Order (biology)
Video game
Contrast (vision)
Quicksort
Open set
09:21
Laptop
Multiplication sign
Virtual machine
Mereology
Mereology
Dimensional analysis
Virtual machine
Video game
Right angle
Contrast (vision)
Quicksort
Metropolitan area network
Task (computing)
10:11
Ferry Corsten
Agreeableness
Video game
Contrast (vision)
Office suite
Spacetime
10:46
Theory of relativity
Agreeableness
Forest
Agreeableness
Bit
Contrast (vision)
Quicksort
11:23
Shared memory
Software testing
Quicksort
Contrast (vision)
12:04
Cybersex
Process (computing)
Social software
Weight
Selforganization
Software testing
Repetition
Content (media)
13:06
Cellular automaton
Multiplication sign
Sheaf (mathematics)
Video game
Spacetime
13:55
Cybersex
Type theory
Web crawler
Hypermedia
Moment (mathematics)
Software testing
Bit
14:34
Point (geometry)
Observational study
Point (geometry)
Counting
Information privacy
Cartesian coordinate system
Formal language
Data mining
Uniform boundedness principle
Facebook
Type theory
Word
Digital photography
Facebook
Profil (magazine)
Descriptive statistics
15:33
Web page
Context awareness
Statistics
Building
Information
Weight
Sampling (statistics)
Expert system
Bit
Cartesian coordinate system
17:06
Type theory
Facebook
Word
Exterior algebra
Term (mathematics)
Decision theory
Order (biology)
Bit
Image registration
Image registration
Resultant
Hypothesis
18:30
Point (geometry)
Group action
Observational study
Characteristic polynomial
Heegaard splitting
Facebook
Centralizer and normalizer
Term (mathematics)
Average
Different (Kate Ryan album)
Representation (politics)
Software testing
Curve
Standard deviation
Distribution (mathematics)
Information
Normal distribution
Gradient
Sampling (statistics)
Median
Bit
Line (geometry)
Measurement
Category of being
Arithmetic mean
Pattern language
Right angle
Asynchronous Transfer Mode
20:19
Point (geometry)
Histogram
Standard deviation
Distribution (mathematics)
PASW Statistics
Normal distribution
Mathematical analysis
Sampling (statistics)
Maxima and minima
Median
Skewness
Bit
Shape (magazine)
Measurement
Number
Facebook
Number
Arithmetic mean
Frequency
Normal (geometry)
Software testing
Kurtosis
Resultant
21:48
Point (geometry)
Statistical hypothesis testing
Correlation and dependence
Shape (magazine)
Rule of inference
Order of magnitude
Facebook
Different (Kate Ryan album)
Term (mathematics)
Natural number
Pearson productmoment correlation coefficient
Software testing
Exception handling
Distribution (mathematics)
Graph (mathematics)
Normal distribution
Sampling (statistics)
Mathematical analysis
Plastikkarte
Bit
Variable (mathematics)
Type theory
Personal digital assistant
Normal (geometry)
Central limit theorem
Ranking
Asymptotic analysis
Freeware
Resultant
23:51
Facebook
Arithmetic mean
Observational study
Causality
Mathematical analysis
Bit
Variable (mathematics)
Measurement
Resultant
Crosscorrelation
24:46
Data dictionary
Observational study
Information
Open set
Data dictionary
Proper map
Formal language
Word
Inclusion map
Negative number
Energy level
Family
Resultant
Position operator
25:44
Web page
Facebook
Digital photography
Word
Crosscorrelation
Agreeableness
Term (mathematics)
Agreeableness
Negative number
Energy level
Process (computing)
Position operator
Resultant
26:30
Word
Digital photography
Statistics
Crosscorrelation
Negative number
Mathematical analysis
Crosscorrelation
27:09
Point (geometry)
Statistics
Observational study
Decision theory
Hypothesis
Number
Subset
Facebook
Coefficient
Thermal fluctuations
Determinant
Logical constant
Variance
Correlation and dependence
Crosscorrelation
Extreme programming
Variable (mathematics)
Statistics
Sign (mathematics)
Number
Exterior algebra
Normed vector space
Pearson productmoment correlation coefficient
Nonparametric statistics
Pvalue
Resultant
29:29
Point (geometry)
Optical disc drive
Word
Strategy game
LipschitzStetigkeit
Resultant
Sinc function
30:20
Medical imaging
Elektronisches Marketing
Crosscorrelation
Observational study
Link (knot theory)
Planning
Open set
Graph coloring
Automatic differentiation
31:12
Web page
Uniform boundedness principle
Facebook
Blog
Adaptive behavior
Mathematical analysis
Software maintenance
Theory
Twitter
Vector potential
32:08
Matching (graph theory)
Information
Observational study
Personal digital assistant
Real number
Agreeableness
Video game
Rule of inference
32:55
Group action
Pointer (computer programming)
Agreeableness
Profil (magazine)
Range (statistics)
Social engineering (security)
33:54
Point (geometry)
Hoax
Term (mathematics)
Robot
Videoconferencing
Motion capture
Website
Flag
Quicksort
Twitter
35:14
Process (computing)
Facebook
Statement (computer science)
Social engineering (security)
35:58
Facebook
Facebook
Software testing
Bit
Open set
36:38
Type theory
3 (number)
3 (number)
Bit
37:16
Scripting language
Email
Real number
Multiplication sign
Mathematical analysis
Translation (relic)
Mereology
Twitter
Formal language
Facebook
Goodness of fit
Word
Process (computing)
Googol
Video game
Family
Address space
Social class
39:28
Type theory
Digital filter
Mobile app
Message passing
Process (computing)
Link (knot theory)
Range (statistics)
Bit
Open set
Message passing
Subset
Crosscorrelation
40:40
Area
Facebook
Crosscorrelation
Projective plane
Set (mathematics)
Energy level
Resultant
Exception handling
Crosscorrelation
41:41
Cybersex
Facebook
Message passing
Data management
Game controller
Link (knot theory)
Profil (magazine)
Hypermedia
Decision theory
Control flow
43:12
Facebook
Context awareness
Statistics
Different (Kate Ryan album)
Bit
43:52
Area
Facebook
Statistics
Computer cluster
Code
Coma Berenices
Area
Twitter
45:14
Quantum state
00:00
all right everybody here is Chris Ali alien by weaponizing cyberPsychology all right thanks for turning up we're going to try a little something first so if it fails then it fails we just have some audio can you go what a brother know what's a game that gives the incredible I'm animal began terrible see public I always wanted to do that and public
00:53
enemy could make it well I didn't actually ask them and we discuss this in the bar actually the point of this I'm digressing already I promised I wouldn't is that if you like that kind of music then that probably says you're sort of bold energetic very confident maybe a bit brash something like that and if you fall into the category that doesn't like that well you know it kind of depends whatever sorts of music you like but that's kind of the point of our talk here today it's around weaponizing cyberPsychology subverting cyber vetting
01:31
and really about a facebook piece of research we did looking at what your Facebook activity says about your personality traits so there's three of us here talking today from a small fledgling volunteer organization called the online privacy foundation myself ally an alien who I'll introduces where we're going along so some of you may
01:55
have seen my talk here Def Con last year my name's Chris Sumner also known as soggy from when I was like about eight years old and on Twitter I've got the unfortunate handle that I wish I never had now of the submaster but I can't really change it although I've got like a almost a two year old boy when he gets a bit holder is probably going to be like dad you've got to change that it is so lame or whatever the word for lame is when he's 18 so we're going to talk
02:24
about what I'm going to talk start us off with a quick introduction to personality trait theory I believe the US you know you guys call it like personality 101 or whatever the subject is 101 then going to introduce us to a little Facebook experiment that we we put together our research projects called the Big Five experiment then going to bring Ali to the stage who's
02:49
going to talk about what she did with our statistics all of the data from that experiment and then we're going to get into what the title of the slide is really you know weaponizing cyberPsychology work please okay weaponizing
03:08
cyberpsychology and that sort saw the uses and abuses that you can do if you've got if you're using this sort of information for decisions and finally look at subverting or evading folks that might be trying to use personality derived through facebook activity against you so starting with personality
03:32
one at 101 I give a quick intro as to how i got into psychology and personality in the first place you see like a lot of you guys I was at University doing computer science and if your university was similar to mine there were a lot of people that looked a
03:50
lot like this on your course and I know before anyone says it G Chris where'd they get that nice photo of you I actually probably didn't even look as good as that so it occurred to me when I was at university I learned one very important thing doing computer science it said if you want to meet chicks then
04:08
you need to be doing something like person a psychology or art or fashion or something like that but actually there's something wrong with this picture that's better hot chicks with hot pizza
04:20
brilliant so but something curious happened doing psychology i want on to do psychology at like adult education after I'd graduated from computer science and it actually turned into a bunch of people that looked exactly like
04:38
this and accept older including me and another curious thing happened is I didn't meet any girls but I really enjoyed the subject so so I digress as a starter for 10 if we look at psychology in terms of personality traits it really started in greece's just about
04:57
everything seems to with this dude called theother aptus and I dare anyone to try and say that at eleven o'clock tonight a long time ago and what he was doing he was going to parties that
05:09
looked a lot like this in Greece because they also curiously invented the laser so some awesome tunes bang then he was observing people and you know really looking at what drove some people to behave differently to others so you know there's a whole train of personality trait stuff that goes from there through all sorts of philosophers but if we fastforward a bit to kind of the grandfather if you like of personality
05:40
traits it's this dude called Carl Jung and the book on the righthand side there was really saw this seminal piece on psychological types I was the title of the book actually and why he did is he looked at grouping people into certain traits obviously the the Greeks had done that and his was kind of an extension from what the Greeks of done I'm not going to talk about his work too much other than that he strongly influenced these two who were called
06:09
Myers myersbriggs so actually it's a mother and daughter team Kathryn and Isabelle and they were fascinated by Young's work and looked at how they could apply that in a sort of a practical manner so they developed really a kind of a questionnaire if you like of like putting people into certain characteristics slots so you don't need to know too much about what i'm about to share but you know if you just leave here you know thinking okay these are the traits i can go and read up about them later then that's kind of fine so
06:43
they grouped people into introverts and extroverts they also split those people in two senses and intuitive and into folks who judges perceivers and thinkers and feelers so they got this kind of matrix going on which you could also look at like this and this is one of the
07:07
major criticisms of myersbriggs which some of you may have been exposed to in your organization's corporations is that it tends to pigeonhole people if it's used incorrectly so it can be used not incorrectly but a lot of corporate types especially tend to really misinterpret how to use myersbriggs and end up categorizing people so for an example if we look at just introverts and extroverts which most people are kind of aware aware about anyway you've got this you've got
07:41
this neat graph where you've got introverts and extroverts across a spectrum with people high in introversion to the left hand side and people high in extraversion to the right hand side what myersbriggs does if used incorrectly as it kind of splits those right down the middle so you're either introvert or extrovert it doesn't make any real account for those people who are kind of in the middle further it can change over your lifetime and I think there was a study that basically cited like something like twenty four to forty percent of people change personality types in myersbriggs but it's not to be dismissive of myersbriggs it's just to say that it can and has been used wrongly quite a lot so when we conducted
08:25
this experiment we looked at something called the fivefactor model also called ocean and it's called ocean because it looks at personality traits like openness conscientiousness agreeableness and neuroticism so let me just explain briefly what those things are so that
08:44
you'll know what we're talking about when we look at the data so this is the crazy scientist from back to the future we chose him as representing openness because he's a creative thinker a deep curious sort of fellow who's you know liable to try lots of different things out in his life in stark contrast to
09:04
somebody who's low and openness if you don't know who this is it's a Stepford Wives they're unlikely to do anything wild and wacky and they're probably just thought people that go into restaurants and order the same stuff every time they go rather than trying something new the next tray is conscientiousness these
09:23
overachieving time keeping people that seemed to plague my life or you could look at it as the robocop dimension part
09:33
man part machine or cop robocop is task focused if you need anything doing robocop is probably your man in contrast to uncle book uncle book
09:52
is kind of a scotty sort of dude actually kind of comes good in the movie so you know we could argue that it's not the greatest thing but I kind of like John Candy so I wanted to put that in but you know you probably wouldn't leave him I don't know with your kids or your laptop or whatever of those is more important to you and this the same could be said about this dude as well as right
10:13
on my personal heroes Ferris Bueller who represents extraversion he's incredibly energetic social outgoing really the life and soul of a party and I think in contrast to Milton from the office on
10:30
the office is office space my apologies who's probably more interested in staying in and shining his stapler that's not a euphemism then going to parties I didn't do that in the dry run actually then we've got agreeableness
10:47
and I chose Forrest Gump because Forrest Gump is kind and considerate you know very trusting individual I mean there's not many people that had run back into the jungle to pick up Bubba but forest you could you know his kind of kind sympathetic sort of dude that would do that and then go and see his folks in contrast to Gordon Gekko who is low in
11:11
agreeableness greed is good and that is something that also we've seen we talked a little bit more about in relation to CEOs and what have you then the final dimension that we're looking at really is neuroticism this is woody allen it's
11:25
sort of an anxious fretting worrying sort of dude in contrast to another
11:32
personal hero it's the dude from the big lebowski and i share something with him is my love of white russians but he's not going to get easily ruffled that's for sure so where are we kind of going with this well the myersbriggs these
11:50
people here loved myersbriggs and most of them are a lot of them you know I don't have too much empirical evidence on how much how many of them actually know what they're talking about but a lot of them don't but they think that they do not only that they use that for
12:04
preemployment screening there's a lot of research out there about how many you know personality tests using preemployment screening and what have you and now what we're beginning to see is the introduction of something called cyber vetting okay so what we've seen a companies like this that are cropping up
12:29
social intelligence HR they will actually look at your what's called at your net rep or online reputation to see whether you're you know you're a suitable candidate for a role actually I started off being really skeptical about this organization but the more looked at it I think the more they do a pretty decent job at what they're doing I'd sooner have these guys looking at my online behavior than having some untrained corporate person you know trying to do that for them because they seem to regulate themselves you know a lot better but anyway I digress further I was flying to austin in Texas last
13:07
year and I found myself at the selfhelp section at the airport you know the cell fell in the book shop at the airport and it's not just me or what you know but the you know just anyway how to improve
13:19
your life and be on time for stuff and I picked up this book by a gentleman called Sam Gosling and he wrote a book called snoop and it's about what your stuff says about you how your rooms and your spaces what that really says about you and you know for example if you've got a messy bedroom for example you're probably low in conscientiousness but just because it's neat doesn't mean you're high in conscientiousness what we'd really have to do is see that being repeated time and time again so yeah you might tidy up your room in some mad spring clean but then it gets messy again after a few days that doesn't mean you're high in conscientiousness but in his book he alluded to you know a
13:55
growing over research around social media and personality traits and that's what kind of spider inspiration and also got me kind of worried because you've got personality tests
14:06
that are using so corporate vetting that pigeonhole people if you like you've got
14:12
corporate types that love to use it but really pretty clueless for a lot of them you've got an explosion in social media and now you've got cyber vetting coming
14:23
up so that gave us our you know oh crap
14:27
moment actually now people can tell who I am without actually meeting me which was you know a bit of a concern so a
14:35
friend of mine who i saw cofounded the online privacy foundation with were in a pub having a beer and discussing this actually we ended up having a few quite
14:48
a few beers and we came up with you know let's do this facebook we can do this on
14:54
facebook be a good opportunity to learn how to use facebook can we came up with the big five experiment as a facebook application that we use that took in as actually over a hundred and fifty data points made up of 74 facebook data points which was to our knowledge the largest study of its kind and we also used something called linguistic inquiry in word count which looks at not just the points but the type of language that you're using in your facebook activity your comments your photo descriptions your profile and what have you and if you're new to this this is a UK election
15:37
and they analyze linguistics from the speakers and you ended up here with this chap net nick clegg who it says hopefully is the most vague so there you can see linguistics being used in a you know i guess a practical context now we had a bit of a problem with our application because we wanted to grab quite a bit of data here is the
15:58
application that you probably all love to hate it's called farmville if you can't see at the back and this is what it asks for and that's what we asked for and we asked people to trust us to handle their information appropriately so we had a bit of an uphill battle but be you know we're kind of able to work beyond that through successfully marketing and building trust and we ended up with a relatively good sample set however turns out my drunken matin myself actually I shouldn't have said that didn't have enough information to know what to do with statistics so we managed to recruit statistics ninja although she doesn't like that phrase so I'd like to welcome Ali be to the podium please they can for Ali be need to do the page hello everyone I am the resident stats expert with the team and what I'm going to do for the next 15
17:08
minutes or so is talk you through a little bit about what we did with the data that we had some of the decisions we made about what to do with that data talk you through the results that we found and also try and apply this to what it actually means in the real world so the first thing we had to do we did this we had all this data on your facebook activity and your personality type and so we needed to come up with some hypotheses as to what we thought the data would show us so our null hypothesis was that there be absolutely no relationship between these two things and our alternative hypothesis was that there would be a correlational relationship between your personality and your online activity and so the data really we had to get it to show one or the other so know what I did next just to familiarize myself with the data I looked at some of the demographics so we've got the country of registration
18:01
where people lived when they registered for facebook so as you can see the vast majority of our participants were from Great Britain or the United States which probably reflects our exposure in those countries where we fly at where we advertised and where we could get word of mouth going really and this doesn't actually reflect the Facebook splits I believe the top four countries are the United States Indonesia India and Great Britain in that order so it's not exactly representative but this is what we got in terms of age and sex oh just over two
18:33
thirds of our participants were female which again is not really reflective of the Facebook population because that's more 5050 split albeit slightly in favor of the females but this twotoone ratio has also been found in other studies to do with online personality testing so perhaps it represents their attendance in females to be more likely to respond to this kind of study in terms of age groups that's pretty much representative of the online Facebook population so at least we know in terms of age groups we've got a good good breakup of the data so before I go on
19:11
unfortunately we can have to do a little bit of housekeeping with you guys a lot of what I'm going to talk about in the next few minutes is to do with the normal distribution and standard deviations so apologies if you already know this but if you don't know it yeah I'm not going to make sense to you in the next few minutes so the normal distribution is a pattern of the distribution of data that follows this bellshaped curve and one of the characteristics of a normal distribution is that the mean median and mode the three measures of an average are all the same value that's this central white line right in the middle so anomalous region is actually perfectly symmetrical and a second point is we use a medical the standard deviation to measure the spread of the information across that bell curve and a normal distribution has a property such that sixtyeight percent of all the values in that distribution for within one standard deviation of the mean 95% will fall within two standard deviations and just over ninety nine percent will fall within three standard deviations and you can see that illustrated here with the different gradients shades of blue so after figuring out who was in our sample then
20:22
had to take a look at the actual results the actual data that we got and figure out how to analyze that if you're interested I used this statistical package spss to do all my analysis and firstly i used spss to create measures of the mean median standard deviation minimum maximum values and also to measures called skewness and kurtosis which are measures of the actual shape of the data and then I also created history a mess of that data to make sure i could really fully visualize what the day to look like so here's an example
20:55
this is the number of posts our sample made in February 2001 and as you can see here it's a highly skewed distribution is absolutely no way this is normally distributed as opposed to this distribution which is the personal pronouns that people use in their Facebook posts and as you can see this kind of has the potential to satisfy a normal distribution but we're not quite sure so we need to do some further tests to find out whether it can be reasonably assumed that it satisfies a normal distribution so I use SPSS again to calculate comma Grove smirnoff tests of normality which test the data to see if it can be reasonably obscene assumes that it fits the normal distribution unfortunately with the tests I performed it showed that none of our data points could be considered to be normally distributed so that was a bit of a bummer but why why is this important why
21:49
am I banging on about distributions and normally shape distributions it's because when you're doing a correlational analysis there are two main different types of analysis that can use one of them is the Pearson's test and one of them is a spearman's test now Pearson's is better we like Pearson's because it looks at the actual magnitude nor difference between two data points whereas Spearman just has a rank value so it's just first second and third regardless of the difference between them and so we really want to use Pearson's but the only problem with that is we can only use Pearson's test on data that is normally distributed and and that has a linear relationship between two variables so it was a bit of a bummer that none of our variables were normally distributed but we do have an exception to the rule in the central limit theorem which states that with sufficiently large sample sizes all samples are all samples 10 towards the normal distribution so if we wanted to we could use that as I kind of get out of jail free card and go ahead and use the Pearson's test but I ended up not
22:52
doing that and there are three main reasons for this firstly our karma graph smirnoff tests of normality showed that none of our data was normally distributed so that kind of reinforced the non normal nature of them secondly we don't know anything about the underlying population so for example I have no idea about the distribution of neuroticism amongst Facebook users and I'm pretty sure no one else really does either so if the underlying population isn't normally distributed I'm not really comfortable in saying that the sample we took was normally distributed and third and finally with sufficiently large sample sizes and we had a sample of 537 which in statistical terms is pretty good so is sufficiently large sample sizes the spearman's test is only slightly less powerful than the Pearson's so in this case I would rather have erred on the side of caution and get results that we can actually count on rather than risk using an inappropriate statistical test so our
23:53
study was a correlational analysis looking at big five personality traits and our Facebook activity now one thing it's really important to remember with correlational studies is that correlation does not necessarily mean causation for example there is a very high correlation between ice cream sales
24:13
and shark attacks but it doesn't mean that eating ice cream is going to make you get eaten by a shark it means there's a third variable there the weather or the temperature that we haven't measured and which affects both of those things and also there's not there could only be one extra variable or they could be two extra variable or three extra variables you can't measure everything so you can't assume that just because there's a correlation between two things that that is the you know the reason for the increase so I'm going to talk you a little bit through our results now and the statistically
24:47
significant results of our study indicated that people with higher levels of openness tended to use words more to do with negative emotions and anger and they'd also be more open to talking about potentially to be subjects like money religion and death and they'd also write more about themselves in their BIOS and they'd give them a lot more information about their hobbies and their interests opposed to that people with lower levels of openness and tend to use shorter sentences and talk a lot less about their family people with
25:19
higher levels of conscientiousness tended to be older they use proper words dictionary words and they talk a lot more about their family and use language about centered around positive emotion and inclusion and conversely people with lower levels of conscientiousness tended to talk a lot more about death and tended to swear a lot more intended to use a lot of angry words and words to do with negative emotion people with high
25:44
levels of extraversion tended to have a lot more friends on Facebook they tended to post a lot more photos and a lot more comments and they use words to do with friends and friendships and use a lot more words to do with positive emotion and ascent and conversely they had a lot less books listed on their Facebook page in terms of agreeableness our results
26:07
showed that people with high levels of agreeableness you use a lot longer sentences but they also use non fluency slacker or um so that could account for the longer sentences and they also tend to be older they'd have more friends on Facebook and again they'd post a lot more photos and comments and we actually found no negative correlations with agreeable nurse at all in terms of
26:31
neuroticism people who scored highly neuroticism tended to post a lot more photo albums they tended to have a lot longer posts and they'd swear more they'd use words to do with negative emotion including anxiety anger and sadness and again we found no negative correlations with neuroticism so I've
26:53
talked a lot about stats and statistical significance and what you probably really want to know is so what what does this actually mean in the real world but unfortunately to explain this I do have to go back to the stats so calculating a correlational analysis in SPSS looks
27:10
like this so this is what I get and this is what's called the p value and this refers to our hypotheses so to remind you our null hypothesis said that there was no relationship whatsoever between personality and your facebooks activity and our alternative hypothesis stated that though is a relationship there whatever that relationship might be so the pvalue states that if the null hypothesis is true if there is no relationship between these variables the pvalue is the probability that we can obtain a result at least as Extreme as the one we found in our study so basically it's the probability that the null hypothesis is correct so if it's how it if it's as small as is here it's less than 0 point 0 0 1 percent so we can reasonably say that it's so unlikely that there's no relationship there we can just discount the null hypothesis and accept that there is a relationship there but all that tells us is the probability we've made the right decision but it doesn't tell us anything about the relationship what that relationship is or how strong that relationship is so to figure this out we need to look at this value which is the R value or the correlation coefficient now in statistical analysis what you really want is a value as close to 1 or as close to minus 1 as you can possibly get because this indicates a really strong relationship so already intuitively you can see here although it's highly statistically significant the number itself is only point2 for which you can already tell it's maybe not that strong so what we need to do with this value is square it and that'll give us our correlation coefficient of determination which is the percentage of variance or fluctuation in one variable that can be explained by the other variable so if we square this number you get approximately point 05 and that translates to a five percent so five percent of a person's extraversion can be determined by how many friends they have on Facebook and so in a nutshell
29:11
statistical significance indicates that we're valid in stating there is a relationship there but it doesn't indicate the strength of that relationship and the result can be highly statistically significant but can only explain a very small amount of variance in the data and that's me done so with that I shall hand back to Chris
29:36
so Ali explain this to me again in a bar and I was like well you know I can see your lips moving but I just can't understand a word you just said so we applied it to Vegas strategy and basically what we're saying is that the results will give you an edge but not a massive edge or stated another way if
29:59
you want to make an educated better and a highlighted bet then you'd be crazy to bet against those odds so you know the point is that yeah it does show relationship but of what practical significance is and that was I guess one of the titles we looked at for the talk but since it's def con it was like a practical significance who's going to turn up let's call it weaponizing so one
30:27
of the things we've looked at once you can determine or have an educated guess about people's personality traits is so there are studies for example that show
30:35
links or correlations between people with high openness and their susceptibility to online marketing in fact on the plane over here those a guy was chatting to who was talking about the use of color and images to get certain demographics click and I clicking on particular link so that was pretty interesting but there are studies that show this so if you're an advertiser and the consequences aren't so bad then you know determining people who have got a high higher openness you may want to target your ads at they're more than people who have got lower openness and if you want to see something pretty neat if you go to we
31:13
feel fine org you can see this used in kind of a sentiment analysis of people's blogs and tweets and stuff like that and it's really pretty neat and it's worth going to have and look at it for but you know as we have discussions in pubs kind of turned to looking at well what could
31:31
you do with online dating so for example if you were looking at a potential mates Facebook page you may be able to determine you know whether there's a slight chance that they may be slightly more high maintenance for example or somewhat more misc us in fact there's a book that Sam Gosling references in his books Snoop called the Rachel papers by Martin Amos where the central character of that kind of adapters personality and what have you two to get the girl so this could you know that this could work well in theory until you actually try and apply
32:09
it and actually meet the person in real life and there's a kind of a dating rule that's well known is that you can only really date between plus or minus two of your you know your potential match unless you've got something to trade with like you know a large bank account status or you know something something like that so unfortunately in this case you you're not going to get too far but
32:37
if you don't intend to meet them and you're conducting say a romance scam for example then it could be quite useful to you because you're not using your own picture anyway so you know looking at this we're also okay well what other studies are there that are out there whatever information is out there and now agreeableness is associated with
32:59
gullibility as well so let's say you're conducting some social engineering maybe you're using your favorite tool for trawling through a wide range of you know social network profiles maybe montego if anyone saw the taught last year it might be useful to know who the people who are more agreeable in a group are and target them first because they're likely to be somewhat more gullible than the others and I guess the key pointers remember that this is a bet it doesn't mean they are going to be more agreeable it means that you've got a slight edge over just selecting people at random so in terms of social
33:43
engineering it's a useful tool for social engineers toolkit I guess unless you kevin Mitnick and you you know it's kind of hard wired into you so I don't
33:55
know if anyone saw I guess you all did the HB Gary stuff earlier this year while there was a you know there's a lot of articles written that in this term came up sock puppets which I've never really heard of before but it's essentially the practice of having fake personas lots of fake personas on social network sites and having them do all sorts of you know fun things and to really explain what some of those fun things are is something I really recommend going and checking out
34:24
on Google is by a guy called Tim Wang who conducted a competition called social bots 2011 where they had teams kind of like capture the flag going basically controlling these sock puppets that they've created on on Twitter and going after unsuspecting Twitter users trying to make them do things they wouldn't ordinarily do and they'd score points for that it's a five minute video and it's well worth checking out so I guess I'll be doing that that next year too so with that I'm going to introduce alien out of the podium to talk about subverting and evading mmhmm so hi
35:16
there as you can see I'm a goon I co run DEFCON London with major malfunction I also run fully for Connor thing that's happening in London later this year so why did I get involved well really
35:31
subverting and evading you know as examples of manipulation are really quite interesting I've done it using social engineering on a couple of jobs and when I saw that the press was starting to get involved and they come
35:48
up with craft statements like this that are just wrong and I start to worry not just the Washington Post and ABC were at
35:59
it as well and if they're coming to this conclusion that you can use Facebook as a personality test what you've seen from Ali's data it gives you a slight edge but you really cannot rely on it and so if they're looking at things like openness one thing that's linked to openness is drug
36:17
use so you're going for an interview that they think you're open they're going to put you in the pocket of the all possible drug users well that's just not right the other issue is that they go into a bit more detail it's not just openness you end up with narcissism
36:39
psychopathy and Mac invasive aneurysm which are known as the dark triad now if you've got the CEO of a company these are the traits of a CEO this is why you're not CEOs guys go you're just too open so um you want to subvert this so the first thing is you've got to know
37:03
your enemy you start trying to play with your personality types willynilly and it could get a bit messy so there's one really quite easy thing you could do
37:16
just don't do social networking I mean it's not hard right now you think this would be a common thing but I actually only know a few people that aren't really involved with social networking and one of them is actually on Twitter which they don't classes social networking and the other one is on linkedin oh so you know the Facebook for businessman and it also leads to trouble you generally linked your online activity to your job or to real life activities if you play the job of say one email address for friends and one email address for family that's great until you marry someone that was a friend and they used the wrong email address when replying to your mother it's tricky to separate your life and I'm one example of this I've used the online Nick aliens since 1982 i'm running a conference in the press release we want businessman at the conference could i use my neck alien no so next year if you look on the DEFCON helping thing I'm probably the only goon they're using my neck next year I'll be using my real name so you want to dick
38:45
with this virtus personality thing let's play with language so we were drinking some guinness stuff and i said well how about you know we all know how good google translators and babel fish let's translate our data that's going into the social networking side put it through that wrap it through a few languages and pop it out again because you know that leads to gobbledygook well it did work but a swear word in English actually when you translate it is still swear word in French and part of the lexical analysis was looking at swear words in all languages so we're screwed don't waste your time it's dead easy to script but just don't bother so tweaking your
39:28
personality haha you can link desirable personality traits to a particular type of job so
39:38
you've got openness here that is quite a large range of acceptable values I think you've got extrovert ism that's actually a much narrower range now this is great but how do you know what to say that would influence these values and with a bit more Guinness we actually came up with this how about if we wrote an app
40:02
that you put your status message that you want to actually do you then tweak the sliders so I want this really open or really extrovert ism and then we pop out the status message afterwards now that would be really cool so if you're going to go away and write this and creative commons please this license and second we need a copy because we've got to give it to Ali because she we've actually got a workout whether your correlation of this data is correct and as you've seen that's not necessarily trivial so the last possible thing that we could really think of is let's play
40:41
with some pills and psychotropics yeah great fun a lot of research has happened in Russia and Eastern Europe not so much in the Western world except for things like prozac the problem with something like prozac is your base level that you start at is very variable because prozac is thrown out like smarties it seems so you're not going to end up with a nice correlated set of results and Ally will chew your ass so yeah forget it and it's now back to Chris thank you
41:19
so I get as it gets wrapped up here I guess the the area there of looking at mental health in particular and you know correlations to Facebook activity is somewhere something that has not had as far as i can tell any research now thing would be a very interesting project to to work through so where are we kind of
41:42
going with this well here's the thing that really concerned me is that you know you're your own social networks you know maybe you are in college your digital stuff lasts forever I know if I had you know my pictures and activity from facebook when i was 18 i probably wouldn't be in employment now and you know it doesn't get spent it's always there so if people are going to be trawling that then you know we've got a problem especially if they're going to be trolling that and jumping to incorrect conclusions so what we're not saying is that there is no link here between Facebook and personality what we're hoping the message is is that yes there is a link but don't use it on its own for basing critical decisions so you know we thought about it very briefly what can you do here is ok cyber vetting
42:41
you could probably apply some regulatory control and we all know how well PTI has worked so i'm not sure that that would work particularly well in cyber vetting plus if i'm told don't go and take a look at somebody's social media profile or online reputation as a manager i just say okay i won't and then i'll go and do it anyway I'm not manager but you know there you go that's the kind of thing it could happen what would appeal for folks though is really to sort of embrace
43:13
people's differences you know the edges where the really cool stuff happens so you know kind of being a bit more open to people's differences and the final thing I guess you know for you guys
43:26
really is where you see these statistics in the newspaper saying that Facebook can reveal your personality question and challenge what it says because those articles stem from a piece of research that did not say that this can be used in this context the newspapers grasped onto that so question and challenge everything where you see statistics so we've looked at
43:52
we've looked at an intro to personality hopefully very briefly we told you about our Facebook out Ali told you how about statistics and blew my mind we talked about briefly you know how you can use that information for good and for bad possibly we talked about how you might hide it you know hide from it and then we talked about kind of what you can do so this really kind of concludes our talk almost just want to bring alien to the podium for one final comment before we wrap up okay so this one's a kind of personal you've probably heard about
44:29
barcode and he's got a pretty nasty disease his bone marrow screwed it's killing his red blood cells he's having he's basically living on transfusions so if you're a US citizen your last chance to give blood is 1900 today please please do it the second thing is again if your US citizen please go to the contest area and get tested for your bone marrow because that's the only hope of really curing what he's got a reason we can't do it by the way as us Brits have all got mad cow disease so don't ping him with emails Twitter etc look at bar code status tumblr com and if all this isn't a good enough reason to do it priest says do it so damn well do it
45:16
thank you thanks ray