Skip to main content
Search form
  • 00:12

    CARL MILLER: I think that, to begin,the good news is that I think that all of us in this roomare all part of the most exciting and most rapidlychanging part of social science.And exactly why?I felt this for seven years now--that what we're all doing is the most important thing

  • 00:34

    CARL MILLER [continued]: that's happening within social science,and it will completely transform social science,and it will completely transform the role of social sciencein society.And I've been trying to think about--or trying to encapsulate or articulate exactly why Ithink that's the case.And I think it is because we're trying to study society--or the way that we're trying to study society

  • 00:56

    CARL MILLER [continued]: is changing through something whichitself is changing society.So we're kind of trying to look through social mediato try and understand how another society works,what the processes that it operatesand how it changes who we think we are--the kind of relationships that we cherish,the reasons we get up in the morning,the things that we value, the problems that we see

  • 01:18

    CARL MILLER [continued]: in society, the ways we go out and try and solvethe problems-- all of that seem to bebeing changed by social media.And social media itself, then, opens up all the new waysin which we can study it.So it seems to be this really powerful combinationof both new and powerful research methodologies

  • 01:40

    CARL MILLER [continued]: being applied to the most powerful agentof social change which is currently happening in society.And I think those two things together, I think,combined make this the most important thing that'scurrently happening.So you don't need me to go through all the opportunities,which I know that we all can see and we all act on every day.

  • 02:00

    CARL MILLER [continued]: Clearly, the data, clearly the scale of the data--the richness of it, the linkedness of it--you know, as we've kind of moved more and more of that kindof intellectual, and political, and romantic, and social livesinto these digital platforms, all the activity which was oncelost to history, all the activity that,in a previous life, me, and Dian,

  • 02:21

    CARL MILLER [continued]: and others would go through the archivesto try and recover, and excavate,and analyze is now kind of being collected and producedin forms which are inherently amenable to collect analysis.And that, I think, is itself kind of like opening upa step change--the amount of data which we can use.

  • 02:42

    CARL MILLER [continued]: But I think especially exciting is, for someone like me--so for someone who works for a political think tank--especially exciting is the way in whichthat data and the research methodswe can use to understand it are opening up the opportunityfor real-time research.So social sciences has, very often,

  • 03:03

    CARL MILLER [continued]: within policy-making and decision-making, beenused as a kind of, lessons learned,what should we do better next time kind of analysis.You know, it's taken us weeks, if not months,and it did at Demos--to kind of get surveys designed, get them out in the field,get the results back, analyze them, interpret them,and then get back to the policy makers.

  • 03:24

    CARL MILLER [continued]: And now, we can see that researchcan be done within a matter of literally seconds.And that allows us to, for the first time,I think, use social science to intervenewithin the very phenomena which it'sgoing out to try and study.And that opens up completely new ways in which social sciencemight be helpful, might contribute to public goods,might make public decision-making better,

  • 03:44

    CARL MILLER [continued]: might make it more data driven, might make it more data rich,and might change the role then of social scientists,and researchers, and think tanks, and academics,and everyone else to actually contribute to those decisions.In addition to the way in which we study society,the things that social listening or social media research

  • 04:06

    CARL MILLER [continued]: allows us to study, I think, makeit the most important thing currently happening.So we're currently clearly livingthrough a kind of techno backlash, a technolash.Clearly, this kind of Utopian idealof the '90s and early naughties, that this technology wouldset us free has dissipated.And instead, I think we've almost too far kind of lashedback at those same platforms and technologies

  • 04:28

    CARL MILLER [continued]: and focus on all the social ills and harms which they create.And our form of research, social listening,allows us to actually put kind of cold, hard objectiveempirical facts around these different kinds of panicsand problems which have become absolutelystaple topics within the press every single day.

  • 04:50

    CARL MILLER [continued]: So echo chambers, they're everywhere.They're polarizing society.They're destroying everything.Well, we can, using social media data,using social media research methods,actually begin to say, well, OK, where are the echo chambers?How do we measure them?How do we define them?How do we understand the sociocognitive orsociopsychological effects of living within them?

  • 05:11

    CARL MILLER [continued]: How do we understand how they operate in different places?How do we understand what those effects are?And therefore, how can we begin to actually createinformed and research-driven ways of trying to do somethingabout the problems that we can measure and seethat they create?Social media elections-- you know,I mean, the number of times I've been asked,is this the first social media election, I couldn't tell you.

  • 05:34

    CARL MILLER [continued]: But again, we can say, OK, so apparently, social media is nowcausing different political figuresto rise up around the world.Well, let's try and understand that.How do we correlate the electoral outcomes of electionswith the social media data in which we can studyand which we can measure?How do we understand how social mediadiscourse might be changing?How do we understand, in this particular circumstance,how echo chambers might be havingeffects on the political decisions which people make?

  • 05:56

    CARL MILLER [continued]: Islamophobia-- hate online-- again, one of the thingsthat Facebook, Twitter, have been accused ofand dealing with themselves is the way in which,amongst all the amazing things thathappened in these platforms, have alsobeen ways for radical left, radical right groups,hate groups, to attack others and, in some cases,

  • 06:19

    CARL MILLER [continued]: to attack minority groups-- religious, or ethnic, or sexualminority groups.And again, we can say, OK, well, we can actually study that.We can try and understand when does Islamophobia actuallyhappen?What seems to drive it?How much of it is actually happening?Who is it directed to?And again, that's what's needed to--for instance, to try and help the police--try to intervene, or try to help government intervene,

  • 06:40

    CARL MILLER [continued]: or even civic society intervene and tryand confront that problem and try and minimize it.I can go on.You know, fake news--like, God, like that needs an academic definition of whatfake news is-- something you see every single day,or, like, the problem of lonelinessin the most networked age that we live,or why depression is going up, or why

  • 07:05

    CARL MILLER [continued]: people seem to becoming more anxious,or why disinformation seems to be spreading.We can use ideational propagation or informationdiffusion-- like methods and researchto understand why and when that seems to be happening.So all these different problems, all these different problemsthat society seems to be facing--some of the biggest problems that weseem to have the fewest number of answersto are all the problems with social media research,

  • 07:27

    CARL MILLER [continued]: by its very nature, allows us to go out and try and solve.So that, again, I think, in additionto all that data and all the amazing new researchmethods that social media research allows us to useand to leverage-- that's also whythis is the most exciting part of social scienceand most exciting part of research.

  • 07:48

    CARL MILLER [continued]: Another kind of keynote is, basically,kind of tee up the issues of the day, and so these are mine.I think that there are a number of challenges to what we'redoing, which militate against those exact opportunities whichit presents.And I don't think we're anywhere near close to solving them.And I want to just lay out what I think

  • 08:10

    CARL MILLER [continued]: those biggest challenges are.Because I think that it is only by gathering peoplelike us together that we're actuallygoing to even hope to try and solve them.I think, firstly, it's telling.When we set up CASM, Centre of the Analysis of Social Mediaat Demos, like seven years ago, I actuallymoved out to the university in order to do that.And I think there were some things which

  • 08:31

    CARL MILLER [continued]: are specific about the way that universities are set up,which still will make it very, very difficultto do the kind of social media research whichneeds to be done.So here are my challenges.Fracturedness.So we've seen social media research definitelyemerge as a practice, and it's a testament to allthe different people in this room.You can see how diverse the practice actually is.

  • 08:53

    CARL MILLER [continued]: So it's happening within academia, of course.It's happening within think tanks.It's happening within boutique consultancy companies.It's happening within the tech platforms themselves.It's happening within in-house corporate research teams.It's happening within the public sector.It's happening within the police.It's happening within the central government departments.It's happening in those different places.So we've seen it emerges a practice, for sure.

  • 09:17

    CARL MILLER [continued]: But I think we haven't seen it emerge yetas a kind of coherent discipline.So ironic, then, that the second themeis coherence in a talk which seems to lack so much of that.But we've kind of--we haven't seen, I think, those things--

  • 09:37

    CARL MILLER [continued]: all be pulled together into a kind of common pool of learningapproach method technology.And I think that is the source of manyof the key weaknesses with what we're currently doing.So we haven't seen coherence in methodology.Dina alluded to this.This is one of the most formidable methodological

  • 09:58

    CARL MILLER [continued]: challenges that I think social science has ever faced.Because in the face of all that data,all the social science approacheswhich we've all been schooled in and used,mainly go out the window.So you can't do qualitative grounded theory methodologyor questionnaires on the millions of data pointsin which we've currently got.So across loads of different kinds of social media research,

  • 10:19

    CARL MILLER [continued]: you kind of see the same thing happening,which is the importation of new muscular data analyticstechnologies-- the things that are capable of handlingthat scale of data.And that has-- every different stageof the research cycle created problems, whichI don't think have been solved.So sampling-- social scientists freak out about sampling bias.

  • 10:41

    CARL MILLER [continued]: We think really, really hard about howthe way in which we get the data might injectvarious kinds of artifacts into the research which wemight put out at the end of it.Well, I think, actually, rigorous sampling methodologiesof social media research are almost nonexistent.So we've searched for keywords.We've searched for hashtags.We might search Facebook groups.

  • 11:03

    CARL MILLER [continued]: We might scrape forums.How on earth do we know that those things arerigorous and representative of the populations in which we'resupposed to draw them from?We really don't, very often, know that at all.Interpretation-- so, again, social scientistsand researchers-- we freak out about the factthat we might be injecting artifacts

  • 11:24

    CARL MILLER [continued]: into the data which we create.So we might be kind of shoehorning the datainto various kinds of categories which it doesn't actuallyreflect.And now, again, we have to use usually algorithmic approachesto try and interpret that data for usif we're doing content analysis-- so to find patternsin it in various ways, and then heinjects artifacts into the data in ways which we can't often

  • 11:47

    CARL MILLER [continued]: tell and often are unsure they're actually happening.And I could keep going on-- interpretation of allthe data at the end of it.It's hard for us to know how on earth and when the outputsof this kind of-- this new form, these new forms of research--sometimes agree with, sometimes contest,sometimes cooperate, sometimes don'twith conventional social science.

  • 12:09

    CARL MILLER [continued]: And I think the actual techniques and beginningto fetter together all these different waysof looking at humanity into one overall picture, again,in my eyes, anyway--and these are all things which I imagine the panelists mightagree with me or disagree with me on,and I hope they say either.I don't think we've come up with a way of doing that yet.And so that's coherence of methodology.

  • 12:31

    CARL MILLER [continued]: Coherence of technology--I think there are, basically, two different kindsof social media research teams which have emerged.You've got a small number of social media research teamswhich build their own tech.And this is especially true within the tech platformsthemselves, a few university outfits, and a few companies--

  • 12:54

    CARL MILLER [continued]: present a big gap, and then the social media researchers--their desperate to do this kind of stuff--can see so much value in looking at social media for their ownresearch questions that are trying to but don't necessarilyhave a tech team that builds social media researchanalytics--analytic technologies.And those researchers, I think, simply

  • 13:15

    CARL MILLER [continued]: do not have the options of technology to use.You see, social media researchersare basically using technologies whichwere designed for the marketing and advertising disciplinesand absolutely don't give researchers what they need--things like sentiment analysis is great for a brand.It's not very useful if you're trying

  • 13:35

    CARL MILLER [continued]: to understand textured, subaltern narrativeswithin Middle East or North African dialogues on a forum.Researchers need more power than that.They need more flexibility.They don't want a plug-and-play platform.They need something else, somethingwhich might take 'em time to learn to use but which actuallyallows them to wrap the technologyaround their research question rather than, I think,

  • 13:58

    CARL MILLER [continued]: what currently happens, which is often shoehorning a researchquestion into what the technology allows them to do.So I think there are witnesses around technologyas well and actually quite often,I think very little idea of what the right technologies areto use.And I think, within the public sector,this is especially true.I think there's not great visibilityamong central government departments, for instance,

  • 14:19

    CARL MILLER [continued]: about what's the right platform, what'sthe wrong platform, which ones are better than which others.And I don't actually know the answer to that.Because you know, you could say, well, OK, the research councilsneed to fund technology development,but you actually need technology which is constantly changingand constantly developing.It's moving forward in such an unbelievably rapid rate

  • 14:42

    CARL MILLER [continued]: that it's hard to see how public funding will keep a platformtop of everything.So open question.Data, coherence of data--so in addition to the fact that thereare some research teams that build their own techand some research teams which don't, thereare some research teams that have accessto-- much more privileged access to data than others.

  • 15:05

    CARL MILLER [continued]: And I mean, we are one of them in a sense.I mean, we work with the social media platforms.And through that, we often get data which is not open,and that presents loads of problems.How on earth do you check?How on earth do you replicate the study?How on earth do you make sure that the findings are accurate?Replicability is a really massive problem, in general,I think in this field at the moment.

  • 15:27

    CARL MILLER [continued]: Because its, firstly, quite difficult to replicate studieson the basis of data, especially when that data has come outof a proprietary platform.And secondly, its really hard to replicate the technologiesused.Like, how do you replicate the algorithm that somebody used?You try to train your own algorithm?That might, for very legitimate reasons,throw out a completely different answer,which brings me to publishing.

  • 15:49

    CARL MILLER [continued]: One of the reasons I moved out of King'swhen we first started CASM was that I simplycouldn't see a yearlong publishingcycle as ever being the right way of publishing social mediaresearch.I know that the publishers are thinking about this.I know there's lots of innovation happening in termsof trying to make it quicker.

  • 16:10

    CARL MILLER [continued]: In many ways, I think, you know, we're getting to the stagewhere academic publishing should belike a real-time analytic output with contextual informationaround it.I don't think at Demos, that what we currently output--these kind of static PDFs--that's not the right way of presentingthis kind of research at all.And we could be far more innovative in the way

  • 16:31

    CARL MILLER [continued]: that we present the outcome of our research.And then, likewise, with the publishing,how do you publish an algorithm?There are conventions within computer sciencefor doing this.There are very few conventions within social scienceor within the social science journalsfor publishing an algorithm and allowingthat to be checked independently by researchers.

  • 16:53

    CARL MILLER [continued]: And then we have some real experts on the next challengehere in this room, so I'll keep my remarks on this brief.Obviously, ethics.We've been worried about ethics since the very beginningof this.I still am.That's not to say there aren't researchers working on it,but I don't think we've got coherent, ethical frameworksacross the whole of the field or we're anywhere

  • 17:15

    CARL MILLER [continued]: close to having that yet.Partly this problem isn't really necessarily justa problem of what we're all trying to do;it's a problem with the big data generally.I just think that we're increasinglyseeing in attitude poll after attitude poll and behaviorafter behavior this sense that most people don'tfeel like they're invited to the big data party.

  • 17:36

    CARL MILLER [continued]: They feel like the big data revolutionis something which is happening to them,not something which they are a part of.And obviously, any field which uses big data in orderto do what it does is wrapped up in this particular problem.And we've got GDPR--so the General Data Protection Regulation coming in very soon,which is going to levy all different kinds of liabilities

  • 17:59

    CARL MILLER [continued]: on data processes, which I imaginemost people in this room will come under thatin one way or another.That is that thick--that regulation-- it's going to be really hard, I think,for independent academic researchers to work outhow on earth they don't get fined 4%of that global revenue, which would be, like, 0 pounds.But yeah, it's a massive problem,

  • 18:20

    CARL MILLER [continued]: and having coherent ethical guidelines in placewould be really useful.One story-- I did promise this was incoherent,so I'm now going to jump back to data again.One thing we haven't seen academia door even government do is come together and actually lobbythe tech giants for some kind of common data access.

  • 18:44

    CARL MILLER [continued]: So you see very little kind of collective bargainingpower in play at the moment where I reallythink there should be.So that's an incoherent point that I just jumped back to.OK, and then, finally, final transition--and Dina mentioned this as well- multidisciplinarity--

  • 19:05

    CARL MILLER [continued]: the researcher of the future willbe someone that can both hold social scientific principlesin their head and is also really at home with machine learning--someone that kind of understands what latent Dirichletallocation is but then also understands why sampling biasmight be a real problem.And certainly, for people like me, we're basically too old.

  • 19:28

    CARL MILLER [continued]: We've missed the boat for this.I mean, we weren't trained.We weren't trained in how to rap togetherthese different methods.So kind of looking, I suppose--looking for this whole new generationof a completely different kind of researcher to come together.At the moment, the best that we can dois stick together social scientists and technologists

  • 19:48

    CARL MILLER [continued]: in the same team and try to get them to cooperate.But I think, really, the next generation, itwill be people that have both the things in their own headtogether, which will make the real breakthroughs.Quite often in academic projects,multidisciplinary means a lawyer and a historiandoing legal work and historical workand then kind of putting their results

  • 20:08

    CARL MILLER [continued]: into the same publication.It's unbelievably difficult to actually dogenuine multidisciplinary work--different languages, different values,different things out of the research.And I do worry sometimes--this is turning into a bit of a bad-tempered moan, isn't it?I do worry sometimes that multidisciplinaritydoes come up as kind of more of a watchword in funding

  • 20:30

    CARL MILLER [continued]: proposals than something that you genuinelysee happening on the ground--bad-tempered moan over-- but it is super, super difficultto do.So finally, to indulge me briefly,I just wanted to say a word about power.Power has been the thing that I'vespent most of this year trying to pursue and understand,

  • 20:51

    CARL MILLER [continued]: trying to understand how power's changing was the result of allthese new digital technologies we use-- the revolution we'reliving for.The thing that-- whether it's in a cybersquat in Berlin,or in South Korea, or an intelligence agency,or a cyber criminal's home, or all the different places thathave kind of found power, apparently changing,

  • 21:13

    CARL MILLER [continued]: apparently transforming, apparently redistributing--there seems to be this thing in common,which we seem to be living through a time of both morecontrol and liberation than ever before.And it's really not clear which way everything is going to go.But I think the one thing that we do knowis that academic research, research being

  • 21:34

    CARL MILLER [continued]: done in private companies--understanding how society works--has always been part of the way in which we've tamed powerand civilized it, and ensure that its worstabuses are identified and mitigated against.And power is changing, and I thinkit's getting rawer, and less tamed, and less understood,

  • 21:56

    CARL MILLER [continued]: and less familiar than ever before.And my own personal take on this is that social media listening,social media research-- whatever it is that we're doing--that has got to be part of the picture in the future in termsof identifying power and working outwhere it's most abusive and working outwhere things need to change.

  • 22:16

    CARL MILLER [continued]: Because that is going to be the heart of how politics respondsto this, how the legal system responds to the changesthat we're going through, how our norms change,and all the different kind of squishy, societal thingsthat change around the technology.And if those things can ever hope to keep up with tech,it will only be because we keep upwith tech within our own discipline

  • 22:38

    CARL MILLER [continued]: so that we can keep everything powered forward and, I think,eventually build a kind of digital worldthat we want to live in.


Director of Research at Demos, Carl Miller, outlines some of the challenges facing social scientists and social science research in age of social media and big data.

Looks like you do not have access to this content.

Carl Miller on "Social Listening" in the Past, Present, and Future

Director of Research at Demos, Carl Miller, outlines some of the challenges facing social scientists and social science research in age of social media and big data.

Copy and paste the following HTML into your website