Algorithms as Political Power: Dr. Taina Bucher on Tech & Society

[00:00:00]very often we think about algorithms as technological entities when i think in fact it's not self-evident that they are before i get into sort of how i see algorithms i think that it might be useful to just say a thing or two about um or address certain prevailing myths um about algorithms so in general so first of all i think that algorithms are not just math code or a set of instructions telling the machine what to do clinging on onto notions of for instance recipe um which is which is one of the most sort of used analogies for explaining what algorithms are while they might be true for sort of rule-based algorithms i think we don't get um too far by sort of um leaving it at that right there they're much more than recipes algorithms are multiples in the sense that how they emerge in specific situations really varies so let's imagine that i'll have to make my mom's famous lasagna dish from memory since i've lost that recipe what i'll do is probably to try it out a few times perhaps over dinner parties with friends um in non-coveted times of course and then sort of little by little perfect the dish um try it out incorporate the feed back of of my friends and and sort it by trial and error and then in the end i hopefully i'm able to make a sort of fairly decent dish which personally is quite a feat because i barely know how to boil an egg but that's another story um so let's imagine then that you and your friends uh will have to sort of make the same lasagna in the same way um now i'm sure that we will pretty much end up with with different dishes they might be called lasagna but they're different and i think that's exactly how sort of machine learning works in that it's not so much the recipe but a procedure for recipe what the end result is really depends on how it was trained the kind of examples that were given how long it took to perfect the kind of feedback the procedure and so on and so in the end you'll end up with sort of an acceptable answer and the the dish will be sort of the same it will be called lasagna but it's not the same it's not the same in each instance so in a way i would rather think of algorithms as more of an assemblage or arrangement specifically in terms of machine learning algorithms then i think they're much more generative agents in and of themselves so not none of these rule-based instructions but really phenomena and entity that also evolve in terms of what else they they get in touch with um the the sorts of histories and legacies and moments that linger within them and the kinds of future that they will open up or preempt thinking about it relationally and as as sort of events um i think has been more fruitful and more more useful because it does not sort of assume an inside or an outside of algorithms but really thinks of it as something that contributes to making the world as assets sort of works on the world very often we think about algorithms as technological entities um when i think in fact it's not self-evident that they are right and because they're per default treated as if they're purely technical um and the idea of the technical object as something mechanical is very very often um what's emphasized humans or the sort of the human aspects of these entities is very often sort of left outside of these formal definitions of what an algorithm is and to me there's no obvious reason for that in one of my case studies in the book i talk about journalism and news media as sort of thinking about algorithms as journalistic events in that they help to problematize an existing set of of understandings and and of what journalism is an ought to be um so so here i'm really thinking about the power of algorithms uh in terms of generating new ways of being in the world of reframing what journalism could be through its encounters with algorithms right so how for instance notions of news and what journalism um sort of the purpose of journalism and what it means to be editorial responsible for instance how those notions are also changing uh by being touched by these ways of of new inclusions of algorithms how how it sort of uh plays a part in in challenging existing boundary boundaries and limitations right so everything from how newswork is conducted sort of how the journalistic institution itself is reorienting towards these new actors if you will so here's sort of this this idea of the algorithm as an actor that influences sort of that that shapes how how journalism is um is viewing its purpose in the world and how certain actors so you know the journalists themselves the editors the readers the investors um even the platforms themselves um how they all sort of have a have a part in in a way playing a role in this technological drama to use sort of uh pfaffensburg's notion um right so if you think about this sort of almost like you play almost sort of a like a theatrical stage then then we might begin to sort of map out how how the the algorithm in conjunction with all of these different actors also shapes new ways of thinking around it shapes new strategies shapes new business plans um shapes new worries and challenges so it's sort of full of opportunities and challenges right so it's not sort of necessarily and not at all necessarily in a negative sense right so it's not just thinking that these algorithms shape sort of um shape institutional logics in a negative way a lot of the people that i spoke to for for this book we're also seeing sort of the opportunity of having to reorganize having to sort of um think about what journalism should be what their businesses should be right so one of them sort of mentioned that that he couldn't remember the last time they were actually having sort of this more ontologically uh oriented discussions uh in their boardrooms right so um just think just having all of these very fundamental questions um revisited about sort of what is the purpose of of our work um how sort of how can we use it to our gain while still keeping our journalistic integrity um sort of all these negotiations i think in a way for me that's what what what is really um so exciting about this is not so much sort of thinking about it as it change actors but really sort of the kinds of discussions and and negotiations that you're able to study because of it right so how are how are these notions of news how are notions of editorial responsibility changing i talk less about politics with a big p in terms of politicians or parliamentary politics or electoral politics and really think about politics in a very broad sense and basically i use politics uh in an in an ontological sense meaning that i'm i'm interested in in the world making capacities in these generative capacities of algorithms and um and so this understanding is very much influenced by by the anthropologists um anna maria mull and her notion of ontological politics specifically which is basically just the idea that that reality is never given but made so it's a way of thinking about this making um this making of the world as sort of a political act and i think in that sense algorithms are political in as much as they as they really help to make the world appear in certain ways rather than others so it's i guess less politics or it's not so much a politics concerned with who gets to speak um but rather a kind of reality that takes shape around and because of algorithmic systems and right so and i think that's also one of my core arguments is that parallel politics of algorithms reside not so much in what algorithms do um or what they are in technical terms right but that that sort of the political lies in when they come to matter uh on the work they they do on the world um the kinds of actors that get deployed and and and entangled within and for what ends uh so who doesn't find foucault useful right he is one of those uh um i mean he's one of those rare figures and and and scholars that sort of seemingly has said it all um right he offers such a rich conceptual uh vocabulary so for foucault building on jeremy bentham's um writings the panopticon really offers sort of this diagram or map of thinking about the regulatory force of power um and it's not so much power then as a power over um which sort of is also in line with foucault's um thinking about power in general right so but more about power exercised through um certain concerted um distributions of bodies so lights and gases and thinking about power of in terms of how things are arranged in a certain order and in this case how how this arrangement or a way of organizing the social then sort of produces the relations in which individuals are caught up and so essentially foucault shows that spaces are always designed and they're designed to make things seeable and seeable in a specific way um and this of course um this this relates whether we're talking about prisons or digital spaces algorithmic spaces um the world is always sort of the built world is always made in a specific way and so while foucault used the panopticon to talk about this there's a regulatory um force of power within this specific arrangement in the present and talked about the threat of of visibility right of these prisoners not knowing whether um they were being watched and therefore sort of internalizing that uncertainty and having to act as if i was thinking about sort of how how an algorithmic social news feed in the case of facebook is of course also an a specific arrangement if we think about it as a digital space designed in a certain way designed to make certain things more or less visible then i was sort of seeing as you say the inverse logic not so much the uncertainty of of sort of being watched as it were but the uncertainty of whether what you're saying or what you're contributing is deemed important enough being on top of the news feed is sort of a reward it's not a punishment like it's the system works in order to um to instill in a way this sense that that contributing something important to your friends and family is something to um to strive for and it can only per definition be rewarded to those whose contribution is the best what does this way of arranging bodies and visibility suggest about the ideal user right because of course most people remain so co-lurkers uh we we still sort of have immense problems with privacy and and people not want to contribute uh to begin with so so it's not that the sort sort of forces people to participate um but that that this way of working helps to shape and arrange that digital space in this specific way that then again suggests uh what is the ideal form of behavior on this site according to the ways in which the algorithm um is organizing that space you can definitely uh read and um [Music] sort of find traces of how a platform or a system envisions what an ideal user is what is sort of the suggested behavior what are the values what sort of what sort of reward systems are there right so what what kind of content is is valued rewarded whether it's whether it's rewarded by heightened visibility on top on top of feeds uh or whether it's even rewarded monetary in a way these systems work by and when i say these systems these i'm thinking about sort of commercial social media platforms that their their only goal is to to keep users to offer offer a service that makes people return and we return on a regular basis we what they do is perhaps um less important but having sort of offer something that feels um important to users something they have to get back to um because at the end of the day that's sort of uh how they how they sell their their advertisement right these are are basically advertising driven uh systems whose only goal is really to uh keep keep that uh advertising complex going and for that they just need they they need users [Music] and they need sort of returning returning users many of the people that have spoken to the so-called users in this book they are they're users who i came across because they voiced an interest in the workings of these platforms right so i that needs to be said um and perhaps those then would be um sometimes called very sort of media savvy users people who think carefully about the systems that they're engaging with and wondering sort of why something is shown to them or why it works the way it does all of them had sort of this experimental attitude towards these systems coming up with techniques that would help them practically navigate in their everyday lives um and so um to me that was very generative for thinking about this concept of what i call the algorithmic imaginary as a very important aspect to knowing algorithms and in some cases as important as sort of knowing the mechanics if you will so that the the algorithmic imaginary is really just an attempt to sort of articulate what i mean by this generative power also on behalf of the public's belief and experiences and expectations towards these systems but i think that the ways in which people encounter and experience algorithms um shape their ways of thinking and talking and feeling about them right so it will definitely also shape their discourses and um and yeah ways of talking about algorithms that in turn shape how they're oriented and act on these systems you