Bestand wählen

Eratosthenes for the 21st Century: Inventing Cyber-Social Geography

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Erkannte Entitäten
the part of the if it if it if it was not and the to
build that indeed indeed to and but
get well thank you and for having me this is actually my 2nd Republic I spoke at Republican 2009 are in fact I think my only picture in Wikipedia is in German language of Wikipedia on
the Republican page so I it's great to be back and had to say the whole shows a lot better now than it was then on the so let me I can kick off this talk I you know the title of it is maybe a little
bit mysterious and it's because I had a short time so that the abstract after being invited sizable they put down something the let me talk about whatever it is that 1 talk about Our but what I really want to do is talk about how we a graphical and be before graphic existed in my own academic work look at online social networks because most of the platforms out the analytics platform serve mainly during social media and online conversations as text in a river of
content and timing you do very 20th century kinds of things with that content or the keywords maybe there's some sentiment analysis but that river of tax that fire hose of tweets and that's the byproduct of what
social media truly is this really is a network of interacting connected agents and that network as shape and structure and so what I'd like to talk about is how we go about mapping out that structure how we think about this problem in a very think 21st century way the so just to quickly go over the agenda I wanna talk about this idea of social Cyber Social geography but then talk a little bit about social networks and machine learning how we analyze these things and then get into kind of the substance of the talk around how do you figure out what's real or organic versus what's in organic or manipulative because I think is we are all certainly aware in 2016 2017 are these networks have been hijacked and used to manipulate the voting public skew-symmetric all kinds of things and what we're doing a lot of work on that by the way so I had where they had a CEO of graphical we also have a relationship with the Berkman Center Harvard supporting a lot of the work in a graph we actually sell data to a lot of different academic organizations as well as some of the major platforms they're trying to grapple with this problem online propaganda so this is something that we are working very uh our work on every day talk about the different levels of sophistication thank you out with these manipulation efforts at different levels of control and bot networks because they've got extremely sophisticated and then talk about the Battle social influences problem of fake news and propaganda and then hopefully suggest and things that we think about what we can do about it the the so this is a matter of a lot of real-time global scale connectivity in 1901 so more than a hundred years ago it was possible to communicate to the other side of the planet in real time just had to be a king or a president or the owner of a major company to do it but think about how impactful the telegraph was to changing human society so the reforming of markets the projection of power the way the empires were structured the way you could give commands to our country that you ruled on the other side of the world the way generals could control armies the structure of all these large organizations the structure of human society was
transformed by this technology given that nobody had 1 in the house except maybe Abraham Lincoln actions say during the civil war abraham lincoln spend more time in the telegraph office at the White House in any other place but I think of how this radically change society and
let's fast forward to now I'm sure most of you have seen this image about 2010 Facebook connectivity so this is what
our global scale real-time communications networks look like right now part but I would like to suggest that the impact of these technologies is going to be as strong as the telegraph and it's not just the difference in what you do with this in your home it's all the social changes are happening on the back here how this is done how countries are
on and help publics communicate so the thing about that map of the Facebook map is that is shown cyber connections but is still projected on a geospace you still looking at how these connections projected GO-mapped but I would suggest that the patterns of connections in online conversations are their own type of terrain all these decisions people make about who to follow what the like put aside in the blog posts all these millions and
millions millions of individual decisions former structure because people the patterns in the way people make these choices sports fans follow their favorite team people follow their favorite politician all these of large scale patterns and connections can be
analyzed as networks so not just streams of text and time but you can analyze it is a kind of cyber social terrain with the patterns of density in this network are meaningful and that's what we do a graphical so this is a picture of a kite 1 piece of cyber social training this is the global
blogosphere about 200 thousand dots on the screen and usually most connected weblogs in the 14 topological languages the way we just described as a lot of these kind of bubble diagrams it is what
it described ever read them what's happening here every dot is a node on the network and it's being told is the 0 by a little spring the force of gravity based on the connections so a weblog network like this is who's being cited in the blogs from the Blogger role as like a little spring the polls folks together so the thoughts down here are all densely interconnected dots here densely interconnected in these network neighborhoods people know with each other are information travels quickly you can think of it as a kind of picture of the pattern of influence and information flow in this network and this is what it looks like for the global blogosphere and that in this case all the colors are different languages so he's scholars language anyone care to guess what the 1 in the middle that's the English so most of our countries weblogs they don't connect directly to each other they connect through English so you have a prominent technical bloggers abiding English and they could be cited in French German Italian Chinese language part but they're very few citations directly across the languages they do exist but if we look at specific countries the online topologies of these networks are extremely sophisticated and extremely detailed so so it's a little hard to read here but some of error
of the these are weblog networks in 12 different languages and each 1 has a different shape but different structure because what people cite what they're interested in these patterns connections differ from society to society and we can see with these languages are so I suggest that this is a kind of cyber social terrain and it's natural these things exist are created organically by all these choices that individuals make that idea of organic Cyber-Social train something very important when we start to think about manipulation so that more this idea of how we think about cyber social trained on the right you're saying that same weblog network are this time rotating in 3 D and I just wanna point out through the folks were not happy that English is the connector that these coming around decided European languages you'll see French Italian German actually connect directly without going through the english so this is maybe a map of kind of post Greg's at the European cyber social trends so on the left is a guy named Eratosthenes uh he was the head of the great library great Museum of Alexandria roughly 3rd century BCE position for the foremost scholar in the known world and he is most famous for this for calculating the circumference of the Earth to within 3 % of its modern known value are amazing intellectual
achievement time is what made him famous but is less famous for this as being the
father of the field of geography she wrote a book called Geographica which are looking at is at his map of the ancient world donors meridians red lines was the
1st person to take seriously that we could measure Q. terrain that's something that you could impose intellectual order on an actual scientific measurement he generated this idea in his book is called Geographica so people often ask me about why a company is called graphic and I say well we have the same ambition for mapping Cyber-Social train so read kind of G yeah so mapping of the 21st century's come a very long way we have much more sophisticated techniques for mapping geo terrain and we're trying to bring the science of mapping Cyber-Social terrain forward so that we can do the same level of sophistication covering platforms and languages across the world so we apply a set of
techniques in multiple languages revered structural analysis with of matter where were analyzing German or French or Chinese and we analyzed Facebook we analyzed where we analyze the web analyze youtube networks working across multiple platforms to do this create maps around particular pieces of cyber social trains this is the example of pure map around ISIS accounts also really
interesting presentation earlier on allies doesn't online influence campaigns of the ISIS accounts often you'll find people with a list here we've got to 200 accounts we've
identified the so we always that's question 200 out of how many and it's not like there's always clear water between a counselor messaging prices and sort of other groups a similar so our mapping will end up with 10 thousand individual accounts this is the Twitter network where the accounts in Arabic categorizes message and we have the western militant accounts and all kinds of other folks the part of the same piece of cyber social trained as a front the Saudi dissidents all these folks are having the same corner of the network as the size accounts OK now I wanna get not more serious suspect but serious about something that I know is of great concern a concern in
Germany Europe certainly United States right now which is the manipulation of the cyber social trends so how is it that our state and non-state actors are able to get control of this network to bring their ideas
out into affects the way people think what trends in the news what ends up being reported on this extremely sophisticated so I say the coordinated actors and not just state actors and commercial actors to I have gotten very very good at manipulating the social media this have strategies that
operate across platforms it's not like there's a our kernel for manipulating Twitter different kernel for manipulating Facebook the search of
cross-platform strategies that that match up with other kinds of pure campaigns are and it's also got to a point where the strategy of the tactics 11 to blend in a lot of grass parts of the network so that cyber social train we're looking at it is natural that emerges that has its own integrity has been our captured in some way infiltrated by these folks are able
stand up sets of accounts that infiltrate the healthy living all parts of the network and blend in and can basically get the signal injected of the strategically most watched network the yeah so I talk about the evolution of
social media manipulation we see really 3 levels in our work so level wonder what we call the eggs your baby but Caesar usually single-purpose plots the stood up there's no effort made to construct a a believable profile and they might just read the same message 10 thousand times they have very simple programming and they're trying to affect some kind of a bond in some particular metric this
also on the troll side yourself motivated trolls people they're going online because again personal interest in causing trouble or somehow manipulating at the 2nd level is more sophisticated this is where you have crafted profiles you can tell whether the account is a real person or a part of the past the so called Turing test it's rather
easy to know and you also have the often under state power of the construction of 50 cent armies with that in China and other countries of the same thing commercial firms do the same thing receptor lots of
people that are paid to go move a message in 1 accounts and then 3rd level of sophistication which is what we've seen a lot in the past couple years is really uh yet more impressive and dangerous which
is a hybrid human and automation we might have thousands of accounts that are mainly controlled and automated ways in which humans take the reins at key times our 4 key moments of the key things so you have a combination of human automation and some the accounts are extremely realistic you really can't tell the not real people and these in social identities exist across platforms will be at work account of your Facebook page the B medium account of be blog or combinations of these things you really can't tell that sort of person yeah yeah so I give a level 1
example so we did some work with the S P we got a call from ESPN are a few months ago and have enough you know in the United States the NBA has a fan voting for the All-Star Game so every year there's
an All-Stars game in fact our team members were selected from across all the teams and the selected on the basis of a combination of fan voting so who's just got the most of motivated fan base and also on expert voting so kind of experts in statistics and all the you know more official sorts of things 1 input to so ESPN had noticed an unusual number of votes for a certain basketball player was very mediocre player and yet was coming in 2nd in the voting and so they track down graphic and said can you help us figure out what's going on here we think he's been supported by a box and it turned out to be much more interesting he
was not being supported by parts just happened to be the 1st NBA player from another country and yet the entire country behind so everyone is home country including the people of the ministries that in the government rather there's a key trended upward from the sky
and it was wholly organic and completely reject it just wasn't American Indian dance but we discovered that 1 of the other players in fact had a series of parts that were tweeting and outrageously high number of times every day votes for the and it was this guy so this is the network of fans around Kevin direct user the basketball player and did have it does have some organic community random but you'll see that his network is basically dominated by his own team and the local bay area
california sports fan so these are the words this is the network of folks out there voting for him and doing it normally but there were about 200 box there were not connected to anything not this network at all follower were not followed by a single legitimate Bay Area sports fan they were tweeting on average 7 thousand times a day
each voting for our cabin so that's what we found instead Our solo Bundesliga star power so this is the network press Paul and in online influence celebrity Kanellakis product to tweets to support from his friend Justin Bieber who as 1 of the largest most active on Twitter footprints in the world and you can see
that the network this is his home sports to network and this is the uh the fans voting from Justin Bieber's network so online influenced object when these kind of fossil evidence sources but Alice look at level 2 so this is
where you have accounts that actually look real this is the Twitter network around German lifestyles we have segments of here around food travel fashions Pylyshyn entertainment news and politics but this is roughly the 12 thousand most influential Twitter accounts in Germany and here's a
whole segment of paid social marketing the out there are striped across the middle this stuff so again
we're using advanced machine learning and analyzing patterns connections in order to surface the segmentation and we can pull out stuff like us another uh may be more worrisome
of level 2 examples so this is Twitter around a major conference that between Russia and Ukraine so these are a roughly 12 thousand Twitter accounts active on the Russian Ukrainian prices all the ones on the right the Ukrainian so Ukrainian business and culture in Ukrainian politics and activism inside these maps the actually usually 50 to 60 individual microsecond so these are just groupings on the left you have a Russian accounts and includes both pro-government of these with 1 supporting the Russian government as well as the activist rather poses the Russian government
but then you have not just 1 but 2 very very large are spam but network so our each 1 of these has
roughly a 1500 Twitter accounts the tightly connected and doing all the same thing if you look at any of the profiles you couldn't quite tell it's not a real person they're fairly convincing so this is level 2
that is to say a word about level 3 which yeah
so I think we all know what has been so scarily happening in western democracies these last the last couple years especially last year it's really the
moved to the forefront are in terms of how the state actors as well as a certain political actors are able to manipulate these networks I would say it's not a new thing we've actually been studying political extremist networks for a very long time but we do a lot of this work now some of that with the organization of this initial for the online civil courage initiative are run by the Institute for Strategic Dialogue in London and sponsored by Facebook where we map that these networks around certain extreme groups hate speech groups and we do this in order to generate information that can help by NGO scanner message and fight back against this kind of stuff on and in this work as well as other work where we study the fake news problem academically and
commercially working with a couple of the very large platforms as well as the finishes at Harvard and Oxford George Washington University and Stanford so we support a lot of
this work I like to say that in comes to studying these kinds of things were working with so many customers and are collaborators across that were like the proverbial man of feeling all parts of the elephant and when such a transfer foot in where the ones that are going to touch the whole thing in our work so so I think we if we bring this idea of cyber social train it helps us think about this problem of fake news bottom trawls in a way that we're really looking at a battle for influence across the cyber social trends I would ask ourselves what can we do to maintain the integrity of that terrain so there's
not being manipulated and shaped by the manipulative forces everything back to 2007 a ever think there's a lot of very optimistic thinking about the impact the Internet would have on democracy and our public conversations I remember when the Howard
Dean campaign in the United States use blogging to help it that candidate uh advanced in the primaries and was the 1st time that you don't have to own a television station or a newspaper to have a public voice in the US and it seems so long ago now of course later Twitter and Facebook and all
that sort of made our blogging what Queen outdated but remember how transformational that wasn't it wasn't that long ago only about 10 years and at the time I think we really thought that this democratization of voice to be afraid that in fact you could have publics sharing more information more points of view represented and this would end up going
well and then 2017 were facing parts trolls fake news the hijacking of this whole system of public communication and subverting it out for the interests of State actors and uncertain portion political spectrum so and we've gone from 1 end of the spectrum of our of optimism to this other and a kind of pessimism and so I think we have to think about what we can do about this but 1st let's look at just saw how bad it hits from this is the network pushing fake you use the United States are right before the election so we took a list of several dozen are known propaganda and fake news websites and this is under the unsatisfactory from January 20 17 are just the beginning of this year and we said we're gonna take every Twitter account that treated at least 10 pieces of fake news from 1 of these websites and we're
gonna make a map of that's what's the network that's out there pushing this stuff and the stuff that you see up there the trunk or unless far right
that's what the story has been in the press that the far right in US the Donald Trump supporters of them a leveraging fake news and pushing it forward and is not untrue but that's only 1 part of what's going on you also have accounts that will look very much and seemed to be uh before you must follow so the occupy movement in the United
States some of the Bernie Rhodes some of the black flies matter stuff so these accounts are also moving fake news there's a whole large segment of this network which is just open the moving progression of the contents supporting a sob opposing any kind of Western intervention serial are as well as promoting a lot of information from wiki links in sort of a loving Julian a sort of and then there's the international right so the user clusters of a counselor our everything from date misogynists out to the
white power movement for white identity movement so this is the entire network moving fake news in the United States it's not just that truck Donald Trump of a much more complicated
in network and it's also international because Cyber-Social train does not respect political borders right that's 1 of the more challenging things about it that of sovereign democracy can have its home native language of information environment now manipulated from folks overseas and here's a
German figures network so does they're working with us and collaborators here in Germany the of the summary of the order of she's here so the Markus Becker adults helped us out with this also we've been doing the same kind of methodology we take note German language of propaganda and conspiracy warranted sites we draw the network of
his pushing that stuff and it's just as complicated in fact more complicated here than it is in the United States and we've got the German right this is what folks may expect who got the left that the European right we got the US right we've got the US laughter and we've got
the Russian focus stuff all of these folks are moving German-language fake news and conspiracy right contact right it's not just the folks in German but this is we can look a lot of Twitter this is those of Facebook map of or these are pipe the public pages were not doing anything phase but private data of the public pages and looking at patterns of connections and likes across them but this is German and non-mainstream politics you can see you've got the right and the left here and right in the middle of it as a cluster that is basically trading in conspiracy theories and all kinds of fake
news right at the middle of the network is connected but sites such as the you know so if you think about the operations of might want to undermine the central institutions of state and empower the political French with a lot of other strange theories about new world and the house a group of
bankers around the world and what you're democratic institutions are illegitimate and don't work this might be exactly strategy you would employ so what do we do about this the so we can find ways to use technology for good in the main thing is creating transparency if you have any state actors
or other active political bottles manipulating the environment I'm not saying cut off anybody from having the legitimate voice right
foreign state actors don't need to invent right wing Americans electing Americans right in terms of I all people out there and their voice in a democracy deserves to be heard to but when you were leveraging foreign state power and capabilities in combination with troll forms and what networks to purposely manipulate the information environment of highly structured propaganda
designed to undermine people's confidence in the system and promote certain cantons over others as we've seen a lot lately are you can create transparency around that and that's the 1st step to open source but as a we can use technology to detect and counter this propaganda and we can do the because things were during the online civil courage initiative to empower the forces of good and and then we also have to talk about our expectations for the dominant platform right it's not like the salt water for office with floral will fall but everybody's been manipulated
so what can a public expect from the platforms in terms of reasonable efforts to secure their own platform from this kind of relations so that's my thought recipe for things that that we can be thinking about for how to solve this problem and thank you very much you want of you know me there's my address of we launched the
German website today of working of more more actively here in Germany as well and so is convergence of questions opponents thank you few John Canny edge u thank you do what do we have any questions yes we have 1 question so when we have time for 1 question so I can you see a running thank
you we have the question sigh that also spontaneous higher minimum sun-diamond freelance
understand thanks for that fascinating talk um I don't weak-effect see um secrets or anything that I'm just wondering if you could give us this make it what is complex thing but how do you get this map of states I mean do you have a machine that will be you have forced programs that look through to itself example for ISIS on what's something I mean and then do you look at the location where people are hot how does that work and some that the 1st part of 1st and 2nd is how do you differentiate among and between real people and machines is that it's really difficult so I have there are some indicators that way you can always so that's definition thank you
so the 1st question we use publicly available
API as a publicly available that as a result the kind of sneaky going on here in terms of what we collect our just leveraging data I think and much more effective ways to to bring out more intelligence because the mode of analytics is more sophisticated so we're using made most people use that same Twitter data or same Facebook data from the API to do more content and all that stuff and we're all we're just interested in the patterns of connections so most people don't touch that they and terms of the underlying methods we use a lot of machine learning it's a it's very large-scale sophisticated analysis of patterns of relationships but to get to get to the mountains all mythological talked to give about that on and then on the was the 2nd part you have it's the a less
realistic at all so there are few
kind of easy things you can do you know so the posting frequency
which doesn't look humans and usually concentrated or in a robust and sleep as there's a lot of fairly easy stuff you of the you know this is going to standard out we're doing 2 things the over about that 1 is in our mapping the algorithms are extremely good at picking up coordinated patterns of activity so if you think about it a whole bunch of real people no matter how similar they are and the way they think are going to be a lot more diverse behaviors and the same number of accounts controlled by a group of robots right which is not going to be the performative diversity and there are so we see that structurally now and we're working on a whole new generation of yet more sophisticated measure of measurements for this and I think we have the 1st academic paper being published with the research collaborator in montages stated and a little bit more of the other the thank you OK thank you
John Kelly you
the right or if I
Quelle <Physik>
Nachbarschaft <Mathematik>
Umsetzung <Informatik>
Streaming <Kommunikationstechnik>
Gruppe <Mathematik>
Metropolitan area network
Shape <Informatik>
Güte der Anpassung
Profil <Aerodynamik>
Dichte <Physik>
Generator <Informatik>
Kollaboration <Informatik>
Rechter Winkel
Ordnung <Mathematik>
Selbst organisierendes System
Analytische Menge
Räumliche Anordnung
Virtuelle Maschine
Lesezeichen <Internet>
Arithmetische Folge
Fächer <Mathematik>
Reelle Zahl
Inhalt <Mathematik>
Cluster <Rechnernetz>
Algorithmische Lerntheorie
Binder <Informatik>
Wort <Informatik>
Abstimmung <Frequenz>
Web log
Formale Sprache
Kernel <Informatik>
Figurierte Zahl
Nichtlinearer Operator
Zentrische Streckung
Konstruktor <Informatik>
Plot <Graphische Darstellung>
Twitter <Softwareplattform>
Strategisches Spiel
Projektive Ebene
Extreme programming
Web Site
Kombinatorische Gruppentheorie
Physikalische Theorie
Bildgebendes Verfahren
Leistung <Physik>
Einfach zusammenhängender Raum
Graphische Statik
Matching <Graphentheorie>
Einfache Genauigkeit
Physikalisches System
Objekt <Kategorie>
Mapping <Computergraphik>
Serielle Schnittstelle
GRASS <Programm>


Formale Metadaten

Titel Eratosthenes for the 21st Century: Inventing Cyber-Social Geography
Serientitel re:publica 2017
Autor Kelly, John W.
Lizenz CC-Namensnennung - Weitergabe unter gleichen Bedingungen 3.0 Deutschland:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen und das Werk bzw. diesen Inhalt auch in veränderter Form nur unter den Bedingungen dieser Lizenz weitergeben.
DOI 10.5446/33131
Herausgeber re:publica
Erscheinungsjahr 2017
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract Social media networks have meaningful and somewhat stable macro-structural topologies, which constitute a kind of terrain subject to mapping and dimensional analysis. Therefore one might view “sociocultural terrain” as organized in two intersecting planes: geographic and cyber-social. Throughout most of history, the geographical plane was where people associated, organized, traded, and struggled. In the 21st Century, the cyber-social plane is rapidly becoming equally if not more important as the dynamic locus of human relationships.

Ähnliche Filme