Ed Finn: Jennifer Golbeck is an Associate Professor in the College of Information Studies at the University of Maryland, College Park. She also directs UMD’s Human Computer Interaction Lab, and stud­ies how peo­ple use social media, and thinks of ways to improve their inter­ac­tions. Ian Bogost is Ivan Allen College Distinguished Chair in Media Studies, and Professor of Interactive Computing at the Georgia Institute of Technology. Founding part­ner at Persuasive Games, LLC, and a con­tribut­ing edi­tor at The Atlantic. So, wel­come both of you.

Alright. So, our top­ic is ​“What should we know about algo­rithms?” What should we know about algo­rithms, Jen?

Jennifer Golbeck: You know, so I talk to peo­ple a lot about algo­rithms, and the ones that I work on as a com­put­er sci­en­tist are build­ing algo­rithms that can take the dig­i­tal traces you leave behind, whether it’s from the Fitbit, espe­cial­ly social media. But any of these traces, and use them to find out secret things about you that you haven’t vol­un­teered to share. Because all kinds of things about you come through in those pat­terns of behav­ior, espe­cial­ly when you take them in the con­text of hun­dreds of thou­sands, or mil­lions of oth­er people. 

So when I go talk about this, the thing that I tell peo­ple is that I’m not wor­ried about algo­rithms tak­ing over human­i­ty, because they kind of suck at a lot of things, right. And we’re real­ly not that good at a lot of things we do. But there are things that we’re good at. And so the exam­ple that I like to give is Amazon rec­om­mender sys­tems. You all run into this on Netflix or Amazon, where they rec­om­mend stuff to you. And those algo­rithms are actu­al­ly very sim­i­lar to a lot of the sophis­ti­cat­ed arti­fi­cial intel­li­gence we see now. It’s the same underneath.

And if you think about it, most of the time the results are com­plete­ly unsur­pris­ing, right? ​“You bought this Stephen King nov­el, here’s ten oth­er Stephen King nov­els.” Sometimes they’re total­ly wrong, and you’re just like, why would you ever think to rec­om­mend that to me? And then some­times we get this sort of serendip­i­ty that you men­tioned, these great answers. And my favorite exam­ple is that I had bought The Zombie Survival Guide, which is exact­ly what the title sug­gests, like an out­door sur­vival guide but for zom­bies. And I read it very quick­ly, and the next day I go back and Amazon is like, ​“Oh, you know, since you bought The Zombie Survival Guide you might also like…” and it has oth­er books by the same author, World War Z, which was made into a Brad Pitt movie which you maybe saw, some oth­er zom­bie books, a cou­ples zom­bie movies, and then this camp­ing axe with a pull-out 13″ knife that’s in the han­dle? And I was like, ​“That’s exact­ly what I need.” The book was telling me this. And then I was like, okay prob­a­bly not some­thing that I need. But I bought it any­way. I thought it was just such a great exam­ple of like, I nev­er would have gone look­ing for it, but it was such a cool thing to recommend. 

And so, I think the thing to know about algo­rithms is that that’s gen­er­al­ly what they do. They usu­al­ly tell us stuff that’s not super sur­pris­ing, or that we kin­da could’ve fig­ured out on our own, but some­times they give us great insights, and some­times they’re wrong. And just like you don’t watch, in order, every­thing that Netflix rec­om­mends, or buy, in order, every­thing that Amazon sug­gests that you should buy, the thing I think we real­ly need to keep in mind with a lot of algo­rithms today is that they’re going to tell us stuff but we absolute­ly have to have intel­li­gent humans tak­ing that as one piece of input that they use to make deci­sions, and not just hand­ing con­trol over to the algo­rithms and let them make deci­sions on their own, because they’re going to be wrong a lot of the time, or they’re… They’re not going to do things as well as a human would do.

Finn: Ian, what do you think?

Ian Bogost: I’ve become real­ly inter­est­ed in the rhetor­i­cal reg­is­ter of this word, algo­rithm. How we use it And I did this piece for The Atlantic ear­li­er this year called ​“The Cathedral of com­pu­ta­tion” in which I sort of said any­time you see the word ​“algo­rithm,” espe­cial­ly in print, in the media, if you try replac­ing it with ​“God” and ask if the sen­tence kind of works, it usu­al­ly does. So there’s this anx­i­ety we have, you know. ​“Google has tweaked its algo­rithm” or ​“What are the algo­rithms doing to us? How are they mak­ing deci­sions on our behalf?” and in what we are we sort of pledg­ing feal­ty to these algorithms?

So there’s a sort of techno-theocratic reg­is­ter to the con­cept of the algo­rithm. And there’s this mys­ti­cal notion about it, too. I think one of the rea­sons we love ​“algo­rithm” instead of ​“com­pu­ta­tion” or ​“soft­ware” is real­ly we’re talk­ing about soft­ware, is what we’re talk­ing about. When we say algo­rithm, we invest this kind of Orientalist mys­ti­cism into fair­ly ordi­nary expe­ri­ences and ser­vices and so forth. 

And you know, this idea of the poet­ry of com­pu­ta­tion is inter­est­ing because I think it helps us kind of get under the skin of the rhetor­i­cal nature of the word algo­rithm, and not just the word but how we use it. When you think about that idea of the poet­ry of com­pu­ta­tion, it should kind of ter­ri­fy you that okay, if we’re going to run our lives, our air­planes, and our auto­mo­biles, and our busi­ness­es on poet­ry, on these sort of poet­ic modes— It’s not because we dis­trust poet­ry, or because poet­ry isn’t good at what it does. It’s because what poet­ry does explic­it­ly is to defa­mil­iar­ize lan­guage. To take ordi­nary speech and to show us some­thing about that speech. To recon­fig­ure the words that we nor­mal­ly use in a dif­fer­ent way.

And this aes­thet­ics of the algo­rithm com­mon in com­put­er sci­ence of ele­gance, of sim­plic­i­ty, of tidi­ness, of order, of struc­ture, ratio­nal­ism, all of those sorts of fea­tures, are fan­tasies. To some extent, these are messy, dis­as­trous­ly com­pli­cat­ed com­pu­ta­tion­al and non-computational sys­tems. Like Amazon has a logis­tics sys­tem, and ware­hous­ing, and all these fac­to­ry work­ers and ware­house work­ers they’re abus­ing and so forth. And all of that stuff, we’d like to kind of cov­er over it. But when we’re able to sim­pli­fy it, to kind of point to this mys­ti­cal God-like fig­ure and say, ​“Oh, the algo­rithm is in charge,” then we feel bet­ter about that gesture.

So maybe one way of think­ing about algo­rithm is as a kind of synec­doche, you know that rhetor­i­cal trope where you take a part and you use it to refer to the whole. So, we talk about Washington instead of the fed­er­al gov­ern­ment. And when we do some­thing like that, we kind of black box all this oth­er stuff. And we pre­tend like we can point to Washington and that that suf­fi­cient­ly describes the way that the fed­er­al gov­ern­ment does or does not func­tion. Which of course it does­n’t do. It allows us to sim­pli­fy the abstract.

So yeah, the tech­ni­cal aspects of algo­rithms, I think have become much less inter­est­ing, cul­tur­al­ly speak­ing, than the rhetor­i­cal func­tions of algo­rithms. How we see this term and this con­cept weav­ing its way into our per­cep­tions. Into the media, into ordi­nary peo­ple’s con­cep­tions of the things that they do, and kind of— ​“Oh, Fitbit knows some­thing about me, and so I’m going to use it.” I think those are some­what under­served perspectives.

Finn: I think yeah, as we engage with these sys­tems more, they become more and more impor­tant for every­day indi­vid­u­als, no longer sort of tech­ni­cal experts or some­body who’s design­ing an air­plane or fly­ing an air­plane. And we’re all depen­dent on algo­rithms in many ways, now. And in many new ways that we weren’t even say, ten years ago.

I’m real­ly inter­est­ed in this notion of defa­mil­iar­iza­tion that you both brought up in dif­fer­ent ways. In part this is about black­box­ing things, and abstract­ing things. In part, it’s also about sort of the unin­tend­ed con­se­quences, you might say. You were talk­ing about the dig­i­tal traces that we leave online, which is a top­ic of great inter­est for me as well. 

And one thing I think about is all the copies of our­selves, or the ver­sions of our­selves, that are cre­at­ed. These pro­files that are aggre­gat­ed by dif­fer­ent com­pa­nies and then poten­tial­ly sold. How lit­tle access we have to them. And so, is that some­thing that you think about as well, Jen? Or do you think that there are— Is the con­ver­sa­tion mov­ing for­ward about that? Are peo­ple learn­ing how to read these dig­i­tal ver­sions of our­selves more effec­tive­ly? Or is this a morass we’re just sort of begin­ning to work through?

Golbeck: Yeah. I mean, I want to say that we’re get­ting more sophis­ti­cat­ed about it. But then if you actu­al­ly look at it, I’m not sure that we are. And there’s so many facets to this. But I guess a cou­ple that I think are interesting. 

One, I like to start with that Netflix/Amazon exam­ple because it’s a way that we’re all in inter­act­ing with this tech­nol­o­gy that, if we talk about it it sounds like these ter­ri­fy­ing black box­es who maybe are so much smarter than us, and we don’t even know how to han­dle it. Except we total­ly do, because we use it on Amazon and Netflix all the time, right? And that’s exact­ly the same thing as the scary AI that Stephen Hawking says is going to ruin human­i­ty, right? We actu­al­ly know real­ly well how to deal with it when it’s pre­sent­ed in that way.

On the oth­er hand, if we look at the kind of vir­tu­al ver­sions of our­selves, I think we can look at our own vir­tu­al ver­sions and under­stand and process those. And when I talk about the kind of algo­rithms I make, I get a lot of push­back. Like, ​“Well you know, the ver­sion of myself that I have on Twitter, that’s a real­ly pro­fes­sion­al ver­sion. That’s not how I actu­al­ly am, and so maybe it’s not going to find the right things out about me.” And maybe that’s true and maybe not. Sometimes, depending.

Finn: Are you say­ing that the axe did not fea­ture promi­nent­ly in your Twitter persona?

Golbeck: Actually, you prob­a­bly could total­ly find the axe, look­ing in my Twitter per­sona. I talk a lot about zom­bies online. 

But you know, we can say that for our­selves, right? But then, if you look at how we treat oth­er peo­ple online, these dig­i­tal ver­sions, and espe­cial­ly when peo­ple get them­selves in trou­ble, the one bad thing that some­body does online becomes the entire­ty of that per­son, as we view them. And algo­rithms can see beyond that. But we as humans often can’t, where this per­son put out a tweet that seems racist. And then that per­son starts get­ting death threats, and gets fired from her job, and all of these bad things hap­pen because the one bad thing that you did that gets shared wide­ly and that there’s a record of becomes the rep­re­sen­ta­tion of you as a per­son to the Internet.

And so we have all these dig­i­tal traces, but it’s real­ly hard for us as humans to process those. And as just one one more exam­ple, we’re doing a project now look­ing at peo­ple on Twitter who have admit­ted that they got a DUI. And we’re look­ing at what sorts of things they say, and can you check if they’re kind of chang­ing their ways or what­ev­er. As I had my stu­dents pre­sent­ing this week, ​“Here’s the peo­ple we found who said they had DUIs. And here’s this guy who got a DUI.” And then the stu­dent was like, ​“Actually, he seems like a real­ly good guy, you know. Here’s this stuff with the base­ball team he vol­un­teers for. And here’s these things with his kids.” 

And I was like oh, we have to be moral­ly ambigu­ous? Like, we can’t just hate him because he got a DUI and admit­ted it? Like, there’s all this oth­er good stuff? And we’re so used to kind of see­ing these dig­i­tal traces and mak­ing our own infer­ences like oh, because this is there, that’s a bad per­son, or that’s a good per­son. And actu­al­ly, we’re all very com­pli­cat­ed peo­ple, and we all do bad things and good things. But we’re not great at judg­ing it when we have a full record of peo­ple. And I think that that’s a prob­lem that comes with all this, is that we don’t for­get, and things don’t fade. Everything is there, and we have a hard time deal­ing with that. Algorithms can kind of deal with it a lit­tle bit bet­ter, or we can pro­gram them too. But as humans we have a hard time han­dling that.

Bogost: We also take com­put­ers to have access to truth in a way that we don’t take poet­ry to, for exam­ple. So to kind of come back to this poet­ry busi­ness, if the pur­pose of poet­ry is to defa­mil­iar­ize words, then the pur­pose of algo­rithms is to defa­mil­iar­ize com­put­ers. They show us how com­put­ers work, and they don’t work, kind of. Or they work bad­ly, or they work in this very wonky, strange way. And you see it when you go to Amazon. You see that you ordered some but­ton cell bat­ter­ies because you need­ed two of them. And then it’s like, ​“Oh, per­haps you’d like these oth­er but­ton cell bat­ter­ies.” And no, no, but I see what you’re doing. I see the car­i­ca­ture that you’ve built of me, and ha ha that’s inter— 

But then we flip that on its head and we’re like oh, actu­al­ly this is truth. Amazon knows some­thing about me. Google knows some­thing about me that’s true, and there­fore I can know some­thing about you by see­ing the way that Twitter or Facebook or what­ev­er is re-presenting you to me. Whereas we tend not to do that with poet­ry if you wrote— You know, here’s your book of high school poems. It’s like oh yeah that’s a sort of car­i­ca­ture of you at a par­tic­u­lar mo—ha ha ha, we’ll look at that and then put it aside and under­stand that you as an indi­vid­ual are more than that set of words, right.

Golbeck: If I can give you a quick exam­ple on that, my dis­ser­taon work was on com­put­ing trust between peo­ple online. So, if we did­n’t know each oth­er, could I guess how much I trust you? And I was pre­sent­ing this—this is like 2004, 2005, so ear­ly in the social media space. And I was giv­ing this talk like yeah, you know, we tell if our algo­rithms are good because you’ll say how much you trust me, and then I’ll com­pute it, and I’ll com­pare what the algo­rithm said to what you did.

And I would get these answers from these old­er com­put­er sci­en­tists who were like, ​“Well, if the algo­rithm says on a scale of one to ten you should trust me to three, but you said a sev­en, maybe you’re wrong.” Like, the the algo­rithms says a three, so that’s prob­a­bly right, as opposed to all of our per­son­al his­to­ry of inter­ac­tions let­ting you make this very human judg­ment. Like oh, but the algo­rithm says three, so maybe you, human, are wrong.

Bogost: It’s super inter­est­ing to think, ​“Well, what does the com­put­er think about me?” But not so inter­est­ing to think, ​“I absolute­ly trust the com­put­er to make deci­sions about me.”

Finn: Yeah, I think that bat­tle of trust is real­ly inter­est­ing, and the ways in which we now— The space of human agency and the space of shared agency, where we’re sort of col­lab­o­rat­ing with com­pu­ta­tion­al sys­tems. And then the space where we just sort of trust a com­put­er to do some­thing. Those are all mov­ing around in real­ly inter­est­ing ways.

For exam­ple, now I find myself ques­tion­ing my fre­quent­ly blind­ly obey­ing the instruc­tions of Google direc­tions about which way I should dri­ve home. And then some­times ques­tion­ing my pathet­ic slav­ish­ness to this sys­tem that obvi­ous­ly does­n’t get it right all the time. And then paus­ing because of who I am, won­der­ing to what extent I’m just a guinea pig for them to con­tin­ue test­ing that this isn’t actu­al­ly the fastest route, this is just that I’m in Test Group B, to see whether that road is a good road.

So, this pos­es a ques­tion I think also comes out of ​“The Cathedral of Computation,” Ian, that we need to learn how to— So, see­ing is one metaphor. I also tend to think of it in terms of lit­er­a­cy and learn­ing how to read these sys­tems. So, how do we begin to read the cul­tur­al face of computation?

Bogost: Yeah. It’s a great ques­tion. It’s an impor­tant prob­lem. So, the com­mon answer, let’s start there, is this sort of ​“every­one learns to code” non­sense that’s been mak­ing the rounds? Which, it’s not—I mean, I call it non­sense just to set the stage, right. But, it’s not a bad idea. You know, why not? It seems like it’s rea­son­able to be exposed to how com­put­ers work, and to some extent you learn some music, you learn some com­put­ing. Great. 

But real­ly the rea­son to do that is not so that you can become a pro­gram­mer, but so you can see how bro­ken com­put­ers real­ly are. And you put your hands on these mon­strosi­ties, and just like any­thing they don’t work the way you expect. There’s this library that’s out of date and some ran­dom per­son was updat­ing it but now they’re not any­more. And it was inter­fac­ing with this sys­tem whose API…who knows how it works anymore? 

And once you kind of see the messi­ness, the cat­a­stroph­ic messi­ness of actu­al work­ing com­put­er sys­tems, then it’s not that you trust them less or that now we can unseat their revolt against human­i­ty. Nothing like that, but rather it brings them down to earth again, you know. But in addi­tion to that, the way that we talk about these sys­tems, and the fact that we talk about them, that we talk about them more is also impor­tant. That moment with Amazon is a moment of lit­er­a­cy. It’s a moment of you as an ordi­nary per­son rec­og­niz­ing, ​“Okay, I see the way that Amazon is think­ing that it has knowl­edge,” and then work­ing with that, and think­ing about it, and talk­ing about it. That kind of lit­er­a­cy is just as, maybe even more impor­tant, because it’s right there on the sur­face, and we can read it.

And then I think there’s a third kind of lit­er­a­cy that’s impor­tant to cul­ture, which is the way that we dis­cuss these sub­jects in the media. It real­ly does mat­ter. And the more that we present the algo­rithm as this kind of god when we write about it, espe­cial­ly for a gen­er­al audi­ence, then the more we don’t do our jobs of explain­ing what’s real­ly going on and how a par­tic­u­lar sub­sys­tem of a com­pu­ta­tion­al ele­ment of a very very large orga­ni­za­tion that has all sorts of things hap­pen­ing, we do a dis­ser­vice to the pub­lic in that respect.

Golbeck: I agree with every­thing you said. And I think this lit­er­a­cy of just being able to under­stand what we know and what we don’t is so crit­i­cal. Because when I talk about this arti­fi­cial intel­li­gence that I do, it’s com­plete­ly unsat­is­fy­ing, whether I’m writ­ing about it or if I’m talk­ing to peo­ple to say you know, what we do is we took all this data, and we put it in this black box, and we basi­cal­ly have no idea what goes on in there. And it spits out the right answer, and we kin­da know it will do that in pre­dictable ways. But we can’t tell you what it’s doing on on the inside. We spent a cou­ple decades research­ing that, and we can’t. That’s a com­plete­ly not-exciting article. 

So what we do is we say, ​“We put your stuff in this…box, and it may be a black box. And it spits out this answer, and look, here’s some stuff that we kind of com­put­ed inter­me­di­ate­ly that sounds like it’s some insights that make you feel like you’re get­ting a story.”

So, the exam­ple that I use most is we take your Facebook likes, and they put them in this black box, and it can pre­dict how smart you are. And that’s not too sat­is­fy­ing. And so we say, ​“Yeah, and if you look at it, here’s the things that you like that are indica­tive of high intel­li­gence. Liking sci­ence and thun­der­storms and curly fries.” 

And every­one goes, ​“Curly fries?” 

And then when I talk about it—especially like, mar­ket researchers—people get real­ly angry. ​“How can you know that’s going to be true? And it’s going to change.” And it’s like, I’m just telling you that for a sto­ry. We don’t use that. We don’t care about that. It’s not part of the com­pu­ta­tion­al pic­ture, but it allows us to tell a sto­ry that makes it feel like there’s some­thing human going on in there. And that is a strug­gle for me, because you want to tell this sto­ry, ​“Here’s what these algo­rithms do, and it’s unpre­dictable and crazy.” But you can’t tell a sto­ry with just like, ​“black box spits out answer.”

Bogost: Yeah, but we can reframe that sto­ry. I don’t know if this is the best exam­ple, but it’s a kind of infor­ma­tion deriv­a­tives trad­ing that you’re doing, right?

Golbeck: Right.

Bogost: Which, I mean, I don’t know that that’s the the way to talk to the everyper­son about the exam­ple that you— But it does­n’t have to be reframed as com­pu­ta­tion, right. There are oth­er touch­points we have in the world, where like, you know how there’s infra­struc­ture? There are all these high­ways, and you did­n’t build them but they were here before you. There are cer­tain com­pu­ta­tion­al sys­tems that were there before us, and we come to them and we actu­al­ly have no idea how they work. We lit­er­al­ly have no idea. So, the work of explain­ing how com­pu­ta­tion­al sys­tems work that does­n’t rely on this appeal to mys­ti­cism, I think is super important.

Finn: I think this ques­tion of sto­ry­telling is real­ly impor­tant. Not only because this is all an elab­o­rate ploy for me to do research on my book project about algo­rithms, but also because humans are sto­ry­telling ani­mals. And sto­ry­telling is essen­tial­ly of a process of exclu­sion, right. It’s select­ing the telling exam­ple that may or may not rep­re­sent the broad­er his­to­ry, but you have to find the exam­ples in order to tell a sto­ry because humans aren’t going to sit down and read the phone book, right? We’re not going to sit own and read the database. 

And so my ques­tion is, how do we grap­ple with sto­ry­telling as…is sto­ry­telling a fun­da­men­tal­ly dif­fer­ent way of know­ing than what we might think of as com­pu­ta­tion­al knowl­edge? You know, when you’re talk­ing about…the com­pu­ta­tion­al approach is the process of inclu­sion, right. We want to include as much data as pos­si­ble to make the data set as rich as pos­si­ble so that the solu­tion will be more com­plete. Is that a total­ly alien way of know­ing? Are there ways to bridge that divide?

Golbeck: I mean, it’s so hard, right. For the com­put­ers, you absolute­ly want to give it every­thing. And then when you’re talk­ing about what the com­put­ers do, gen­er­al­ly when you’re work­ing with this huge amount of data, which is the excit­ing thing now, you’re end­ing up with not log­i­cal insights but sta­tis­ti­cal insights. And any human can look at the con­nec­tions that are formed and go, ​“That does­n’t make any sense to me except that it tends to work most of the time.” And so we want to tell a sto­ry that says here’s some sta­tis­ti­cal insights, and and let me tell you a few.

But that does­n’t real­ly give a pic­ture, and it’s hard to give a pic­ture, of ​“here’s how sta­tis­tics work,” and lit­tle pat­terns emerge as impor­tant from this big mass of data. It’s a sto­ry that I try to tell all the time. But peo­ple, I have found, latch onto the spe­cif­ic exam­ples and have a hard time grasp­ing the big­ger thing. And I think in terms of com­put­er lit­er­a­cy that that is so much more impor­tant than being able to pro­gram. Programming is great, and you will see what a mess it is. But being able to grasp that this is a sta­tis­ti­cal insight and the indi­vid­ual exam­ple does­n’t mat­ter, that’s the thing that I would like to be able to do better.

Bogost: Yeah. I mean, com­put­ers are more like mar­i­onettes, or like table saws or some­thing than they are like sto­ries. They’re these machines that pro­duce things. And you design this machine such that you can then design things for the machine. So you have your table saw, and you make a bunch of jigs so you can get the right cut. And you build this pup­pet, then you have to kind of manip­u­la­tive it in this per­verse way that you can’t real­ly even explain, in order that it pro­duces an effect that appears to give life to the creature. 

It’s a dif­fer­ent way of think­ing in the sense that whether it’s a sto­ry, whether its an out­come, or a busi­ness result. Whatever it is that the par­tic­u­lar com­pu­ta­tion­al sys­tem is doing, it’s not doing delib­er­ate­ly, and it’s not doing it in a sin­gu­lar way. It’s a sys­tem that’s been designed to pro­duce many sim­i­lar kinds of out­comes. And this is a kind of weird way of think­ing about behav­ing in the world, espe­cial­ly since we ordi­nar­i­ly think in and talk in specifics. In sto­ries, in exam­ples, in indi­vid­u­als. And that’s also still how we write about every­thing, includ­ing computation.

And you see this when you see com­pu­ta­tion­al arts, and you see the aes­thet­ics of com­pu­ta­tion if you look at Twitter bots or gen­er­a­tive text, or any kind of gen­er­a­tive art. You know, the results are ter­ri­ble when com­pared with hand-crafted sto­ry­telling, or humor on Twitter, what have you. What’s remark­able about them is not their indi­vid­ual utter­ances or indi­vid­ual effects, but that there is some sys­tem pro­duc­ing many of them, and when you look at it holis­ti­cal­ly you can appre­ci­ate it in a dif­fer­ent way. And kind of get­ting that aes­thet­ic abil­i­ty, right?

I mean, we talk about ethics a lot when it comes to indus­try and to com­put­ing. But we don’t talk about aes­thet­ics enough. Like, one oth­er way into this lit­er­a­cy prob­lem is through aes­thet­ics. Understanding how com­put­ers pro­duce results on the artis­tic reg­is­ter, right. Even if we kind of hate those results, or we can’t rec­og­nize them as art, and say­ing, ​“Actually, some­thing just like that is hap­pen­ing inside of Facebook or inside of Google.”

Finn: Yeah, I think that notion of aes­thet­ics is real­ly impor­tant because I think it’s one of the ways that we can con­front very inhu­man or very alien ideas and sys­tems, method­olo­gies, with­out nec­es­sar­i­ly hav­ing the lan­guage to artic­u­late what it is, right. Aesthetics can be a non-verbal way of engag­ing with these questions.

So, I think there’s a con­nec­tion between aes­thet­ics and what you referred to as illu­sion before, as well. And so my ques­tion for you both now is, are the illu­sions nec­es­sary? Or we could talk about it as that kind of faith, and you know maybe it’s a bank­rupt faith or a mis­placed faith. But is that some­thing we have to have? Is that the only way that humans are going to inter­act with these systems?

Bogost: No, it’s start­ing point. It’s the thing you do when you don’t have bet­ter options. And then you real­ize oh, this is insuf­fi­cient. And this is a good start­ing point. And then you rec­og­nize also the intrin­sic flaws of the illu­sion. And you seek more knowl­edge and deep­er under­stand­ing. And then you real­ize this [has] sort of know been demys­ti­cized now.

And you can do this his­tor­i­cal­ly. Maybe that’s one con­crete exam­ple of some­thing we can do. Go back and and unpack any his­tor­i­cal com­put­ing sys­tem, and see the bizarre rea­sons why it was con­struct­ed in the ways that it was. What it did. How it had an influ­ence on lat­er sys­tems. Then you’re just, ​“Oh, okay. This is just like any­thing else.”

Finn: The Atari, for example.

Bogost: The Atari, for exam­ple. Yeah, I’ve writ­ten a book on the Atari that tries to do exact­ly this. So, com­put­ing his­to­ry has a role to play here. And as a kind of very quick aside on that mat­ter, com­put­er sci­ence as a dis­ci­pline is one of the most ahis­toric that I know of. Just com­plete­ly unin­ter­est­ed in his­to­ry. It’s just bar­rel­ing for­ward, right, mak­ing that last algo­rithm slight­ly more effi­cient so they can do some­thing slight­ly different.

Golbeck: Yeah. I think you’re mar­i­onette exam­ple that you gave. I’ve nev­er heard that exam­ple before, but I think it’s so spot on, and gets to all of these issues that we’re talk­ing about. Because if you’re watch­ing this mar­i­onette per­form, that’s one thing that you can see, right. And then if we try to explain it, ​“Oh, if I pull this string, this thing hap­pens,” we can have all of these debates about why does that thing hap­pen? And why isn’t this thing? And can’t you do it this oth­er way?

But that’s dif­fer­ent than the thing that is being pro­duced for you to look at. And which of those con­ver­sa­tions do we want to have? Maybe both. But they’re two real­ly dif­fer­ent con­ver­sa­tions. And I think that’s part of the strug­gle, that as a com­put­er sci­en­tist I always want to talk about both. Look at this amaz­ing thing that you can see that it’s doing. And then also here’s all these crazy things that make that work.

But it’s real­ly two dif­fer­ent sto­ries, and I find it’s hard to say, ​“Here, you pull the string and this hap­pens.” And peo­ple say, ​“But how do you get this big com­plex thing at the end?” And it’s just too com­pli­cat­ed [crosstalk] to tell it all the way through.

Bogost: Because it’s a lot of strings

Golbeck: Yeah, there’s a lot of strings.

Finn: Yeah, I think that the sort of unan­swer­able ques­tion about whether it’s real­ly a mar­i­onette unless you’re see­ing that com­plex­i­ty at the end, right? And that’s the thing that you focus on. Which I think is about aes­thet­ics and kind of notions of per­for­mance, or when an algo­rithm or a sys­tem becomes a cul­tur­al thing.

We just have a cou­ple of min­utes left. So, what would be some, just to sum up, a cou­ple of prac­ti­cal things that you would sug­gest if some­body wants to actu­al­ly under­stand algo­rith­mic sys­tems better?

Golbeck: Oh gosh, that’s so hard. So, com­ing back to a point that you raised before about algo­rithms as poet­ry or algo­rithms as beau­ti­ful things. I’ve absolute­ly had that thought, that I’ve looked at algo­rithms and I’ve gone, who­ev­er wrote this had this new insight to the prob­lem that I did­n’t have. You can learn about algo­rithms with­out hav­ing to learn about com­put­er sci­ence. And so I guess if some­one want­ed to do, that some­one like, ​“I don’t real­ly know any­thing about com­put­er sci­ence. I just want to start get­ting in to see what that is,” that you might start with some kind of basic tuto­ri­als on the Turing Machines. 

You men­tioned Alan Turing at the begin­ning, and he kind of put for­ward this fun­da­men­tal notion of all com­put­er sci­ence that says you can have a piece of paper and basi­cal­ly a lit­tle pen­cil that can write a one or erase a one, and that can rep­re­sent all com­put­ers every­where. And you spend a lot of time as an under­grad­u­ate doing that. It can get very com­pli­cat­ed, but it is an acces­si­ble con­cept. And I think if you spend a cou­ple hours play­ing around with that and see­ing how you can do actu­al­ly sophis­ti­cat­ed math and all kinds of inter­est­ing things, with this real­ly sim­ple machine, it starts to give you an insight into the process that we use to devel­op these much more sophis­ti­cat­ed algorithms.

It won’t help you fig­ure out all of the strings and the train­ing that you need to manip­u­late those strings in the right way to get the pic­ture, but it starts to help you see like okay, these algo­rithms, it’s not this myth­i­cal thing, it’s like a bunch of peo­ple who were beat­ing on this real­ly hard prob­lem, who kind of manip­u­lat­ed into doing the thing. So I think as a start­ing place for learn­ing how the algo­rithms work, it won’t get you into all the com­plex algo­rithms, but it gets you in the space of think­ing about them in the right way. 

Bogost: Yeah, I mean, com­put­ing his­to­ry is what I think we’re both point­ing at. If we’re liv­ing in this deeply com­pu­ta­tion­al­ly age where com­put­ers are inside of and run­ning so much of our lives, maybe we should know where they came from.

Finn: Thank you both so much. That was great.

Golbeck: Thank you.