Malte Ziewitz: We thought that as we are wait­ing for the cof­fee to kick in, we should at least give you a brief intro­duc­tion to the con­fer­ence theme which main­ly involves three things. One, to briefly recap how this con­fer­ence came about, what was it that puz­zled us ini­tial­ly and that led us to think about some­thing like Governing Algorithms. 

And then sec­ond, share five select­ed scenes from the provo­ca­tion piece that some of you might have read. This provo­ca­tion piece actu­al­ly start­ed as a record that Solon Barocas, Sophie Hood, and I just put togeth­er for our­selves ini­tial­ly, but then we thought it’s actu­al­ly a nice thing to share. And now it’s become quite pop­u­lar, at least by…when you count the down­loads. So we thought while it is impos­si­ble to share real­ly every detail and to give a com­pre­hen­sive overview, we should at least share five themes, five algo­rith­mic moments that will prob­a­bly pop up at dif­fer­ent stages dur­ing the day. 

And then final­ly the third thing I’d like to do in the next fif­teen min­utes is to actu­al­ly just walk you through the pro­gram, the plan for today and make sure that every­body’s com­fort­able and feels fine.

So how did this start? Actually all of us—Solon, Sophie, and many oth­er fel­lows and research, not just at PRG, the Information Law Institute, but also at MCC—we’ve been study­ing com­pu­ta­tion, automa­tion, and con­trol in dif­fer­ent forms for quite a long time. But it was only at the end of last sum­mer real­ly that we real­ized that there’s this new notion of the algo­rithm gain­ing cur­ren­cy, right. So at that point we noticed well there’s trade books com­ing out, right, by Christopher Steiner, Automate This: How Algorithms Came to Rule Our World. There’s a book called Nine Algorithms That Changed the Future by John MacCormick. There’s hard­ly an edi­tion of The New York Times which has not the word algo­rithm” in its title and also oth­er news­pa­pers. They are Tumblr blogs on the top­ic of Algopop, where peo­ple just col­lect stuff that’s relat­ed to algo­rithms. Boing Boing has posts on stuff like algo­raves, which is basi­cal­ly a very bor­ing video of peo­ple to elec­tron­ic music, but actu­al­ly it’s called an algo­rave,” which is quite amaz­ing. So why should this be called an algo­rave? And final­ly, maybe the ulti­mate har­bin­ger of some­thing becom­ing real­ly hot, there’s a TED Talk, and the one by Kevin Slavin some of you even sug­gest­ed in the open read­ing list that we have on the web site and that Helen mentioned. 

So, there seems to be this big buzz around algo­rithms. And of course that’s some­thing that as researchers we are quite sus­pi­cious of, right? And so, we felt increas­ing­ly uneasy about this. And the uneasi­ness came from a ten­sion, which is the fol­low­ing. On the one hand, there are increas­ing­ly claims that algo­rithms gov­ern, shape, man­age, con­trol, select, sort or oth­er­wise shape our lives, as you can see in some of these titles here. But then at the same time, it’s real­ly dif­fi­cult to pin down how they do it, what these algo­rithms are, and how these things work, and what actu­al­ly is the point of focus­ing on algo­rithms when you could focus on so many oth­er things. 

And so this was the ten­sion between the rather hyper­bol­ic, or you could cyberbol­ic, claims in these cas­es. And then on the oth­er hand this uneasi­ness about well, what is this thing, actu­al­ly, and what work does this con­cept do? 

So then the next thing we did, which is an obvi­ous thing when you do research in the social sci­ences, you look at what oth­er peo­ple at that time, at that point in time that was last sum­mer, have writ­ten about it. And then we thought well a use­ful way to think about this is is think about it in terms of three strands. 

So there’s a lot of peo­ple in soci­ol­o­gy, for exam­ple, who start­ed the­o­riz­ing the phe­nom­e­non of the algo­rithm. So Scott Lash, for instance, talked about algo­rithms as path­ways through which cap­i­tal­ist pow­er works. Grand claim. 

Nigel Thrift has writ­ten about the tech­no­log­i­cal uncon­scious,” right. So all this stuff that is some­how hid­den some­where that does some­thing, and we don’t quite know how it works but it is pret­ty impor­tant, apparently. 

There’s Lorraine Daston, a his­to­ri­an of sci­ence, who’s talked about algo­rithms as rules of ratio­nal­i­ty that replaced the self-critical judg­ments of rea­sons.” So she makes this broad his­tor­i­cal­ly to think about and make sense of them.

And then there’s legal schol­ars like Ted Striphas, who talked about the black box of algo­rith­mic cul­ture which wraps abstrac­tion inside of secre­cy.

So then our next strand, which in many ways there’s a com­mon dynam­ic, which then is often a reac­tion to the first one, is that peo­ple then start to ask Well, these are all grandiose claims. But how do these things work in prac­tice,” right. That’s when they start empiricizing. 

And then there’s peo­ple like Scott Graham, for instance, who said It is…time for a con­cert­ed mul­ti­dis­ci­pli­nary effort to try to open up the black box­es’ that trap software-sorting.

Then an arti­cle that prob­a­bly many of you know by David Beer, a recent one, sug­gests What is need­ed are descrip­tions of pow­er through the algo­rithm,’ devel­oped by focus­ing on those work­ing with…applications and soft­ware, by focus­ing on the appli­ca­tions and soft­ware as mate­r­i­al enti­ties, or by focus­ing on those who engage with the soft­ware.

And when you then read this, it makes sense, it sounds like a good way to study algo­rithms. But then, on sec­ond thought we thought, well what is this oth­er than actu­al­ly study­ing engi­neers, tech­nol­o­gy, and users? And what’s the point of under­stand­ing algo­rithms in there?

And then final­ly a third strand, which is also quite com­mon in these dynam­ics and often even detached from the oth­er ones, is a strand that has to do with politi­ciz­ing. So peo­ple then become con­cerned with algo­rithms as gate­keep­ers of pub­lic com­mu­ni­ca­tion.” The European Commission has been wor­ried for a while about search algo­rithms oper­at­ing with­out suf­fi­cient trans­paren­cy, calls for open­ness, dis­clo­sure, and so on. Or gen­er­al­ly, a con­cern with the pow­er of algo­rithms needs to be assid­u­ous­ly addressed.

So what we found inter­est­ing about these things is that in vir­tu­al­ly all of these cas­es algo­rithms are used as what Ian Hacking once called a bit of an ele­va­tor word.” It’s a word that you can use and it just takes you anywhere…you want, real­ly. So, in many ways algo­rithms here appear as a resource. 

So we thought well, what hap­pens when we turn this into a top­ic, where we don’t make claims about how algo­rithms change our lives, world, and so on. But what hap­pens when we turn this idea, this notion of an algo­rithm into a top­ic and invite you all and then have a great time for one or two days. 

And so then the idea is to think about algo­rithms as a top­ic of empir­i­cal study, or of a social sci­en­tif­ic study, and as a theme that is mobi­lized to orga­ni­za­tion­al and polit­i­cal pur­pos­es. And at this point, you arrive at slight­ly dif­fer­ent ques­tions. So you don’t ask any­more nec­es­sar­i­ly what is an algo­rithm, or what do we need to do to under­stand one. But you ask well what is it that algo­rithms actu­al­ly do,” and what are they,” any­way? You ask well, what issues and con­cerns do algo­rithms” allow us to address across such diverse fields. Which is quite amaz­ing, because when you look at the pro­gram today, we have papers from air­port secu­ri­ty, from pla­gia­rism detec­tion, from the finan­cial mar­kets, and from social media I guess, right? So what is it that makes this con­cept use­ful still across these fields, even though it seems to be a bit of an ele­va­tor word. 

And final­ly, how can we account for the grow­ing inter­est in algo­rithms in fields beyond com­put­er sci­ence? That’s some­thing we start­ed to dis­cuss yes­ter­day in the exchange I think between Chris and—the lady’s name I unfor­tu­nate­ly for­got and who’s prob­a­bly not here today. In short, the dis­cus­sion already start­ed yesterday.

So, just to get us in the mood we thought we’d just extract five very quick things from the provo­ca­tion piece, and maybe sharp­en your sens­es and also ask you to keep these in mind as we go through the day, through the presentations. 

So the first theme that we strug­gled with was real­ly to think about algo­rithms, the very idea algo­rithms. And the exam­ple we’re going to use here, just for sim­plic­i­ty because every­body knows it, has to do with web search. It’s a bit lame [indis­tinct], maybe. But it’s just some­thing that’s easy to explain. 

So in that case of search there’s often claims, as you all know, that the search algo­rithms rank eval­u­ate these things and then deter­mine rel­e­vance. So rel­e­vance is algo­rith­mi­cal­ly devel­oped and deter­mined. So then the trou­ble starts when you think about how this actu­al­ly hap­pens. So then top­ics come up that already sur­faced yes­ter­day. So what about the data, right? When you do search, you don’t search the Web but an index of the Web. So there’s a huge data­base which is itself updat­ed all the time and eval­u­at­ed, a point that will also pop up in one of the respons­es today. 

What about the mate­r­i­al enti­ties that are involved? What about stuff like data cen­ters, that some of you are study­ing like Kate or Megan Finn. What about even mate­ri­al­i­ties that you use when you use a com­put­er. Like a key­board, a desk, you sit in front of a screen, right. What about the inter­face and its design? I mean, pro­duc­ing top ten list of links is just one way of think­ing about these things. You could do this many oth­er ways in prin­ci­ple, in many oth­er col­ors, forms, arrange­ments. And final­ly what about the user, right. When you do a search, you come up with a search query. You might not search just once but twice, or three times. And then what you do is actu­al­ly out­side as well. But still when we talk about search there’s often this reflex to think about the algorithm.” 

And so, we thought well what do we actu­al­ly talk about when we talk about algo­rithms in these con­texts. And what is it that we can under­stand by look­ing at these ques­tions. So this is this weird thing, is that it’s easy to make a claim that involves the word algo­rithm” in the gen­er­al, but the moment you try to be spe­cif­ic you kind of lose it. Which made us a bit ner­vous. Also about discomfort. 

The sec­ond theme has to do with secre­cy. That’s some­thing that also comes up very quick­ly, right, when peo­ple talk about algo­rithms. Especially researchers are end­less­ly wor­ried about try­ing to under­stand the inner work­ings of the algo­rithms. So how does this actu­al­ly work. And then they try to get access, and try to talk to peo­ple, and try to have some­one explain it to them. They turn to com­put­er sci­en­tists, they turn to peo­ple who work at some of these com­pa­nies. And to stay with the exam­ple of Google, what hap­pens a lot of the time is this. This is a quote by Udi Manber. He’s a vice pres­i­dent of engi­neer­ing at Google, and he writes in a blog post, 

Algorithms are in many ways Google’s crown jew­els. We are very proud of them. For some­thing that is used so often by so many peo­ple, sur­pris­ing­ly lit­tle is known about rank­ing at Google. This is entire­ly our fault, and it is by design. We are, to be hon­est, quite secre­tive about what we do. There are two rea­sons for it: com­pe­ti­tion and abuse.

Introduction to Google Search Quality

So this is the com­mon sto­ry you get, and most peo­ple know. So it needs to be secret because com­pe­ti­tion and abuse, and they’re hid­den. Which also sug­gests actu­al­ly that there is some­thing to learn. 

Now, this is from field­work with the search engine opti­miza­tion indus­try. And when you jux­ta­pose what Udi Manber said with what a senior SEO con­sul­tant said, you kind of start get­ting into trou­ble because this guy said,

If you try to solve the rid­dle, if you try to come up with your own math­e­mat­i­cal ver­sion of Google, you’re dead, you’re dead in the water. You can’t do it, you know, nobody at Google could do it. Larry and Sergey would not be able to sit there and say, Here is my algo­rithm.” There’s too much stuff in there. They might know all the ele­ments of that algo­rithm but there’s dif­fer­ent influ­ences and factors.
SEO Consultant 48

So this then makes one even more ner­vous, right? So on the one hand there’s wide­spread talk about secre­cy, and we have to under­stand how these things work, and this might be a ques­tion of access. Some peo­ple dis­cuss this as a ques­tion of exper­tise. This might be a ques­tion of know­ing. But then it may turn out that the secret is that maybe there is no secret. Which then brings up all kinds of oth­er inter­est­ing ques­tions. That is, what work does this secre­cy, this spec­tac­u­lar secre­cy, do if there is not much to know. So what is it to know” an algo­rithm in light of wide­spread con­cerns about secre­cy, obsuri­ty, and inscrutability?

A third theme which I’m sure will also be dis­cussed today had to do with prob­lems and solu­tions. And the rea­son is that also again, a lot of talk about algo­rithms is framed in this rhetoric. So algo­rithms are often con­ceived as techno­sci­en­tif­ic solu­tions to pub­lic prob­lems. This is an exam­ple again, to stay with our exam­ple of the Google search engine from an expla­na­tion about how search works on the Google cor­po­rate page. It says,

You want the answer, not tril­lions of web pages. Algorithms are com­put­er pro­grams that look for clues to give you back exact­ly what you want 

I mean, already this phrase, they give you back exact­ly what you want” is…you could write an arti­cle about that. Which is quite amaz­ing. But what’s inter­est­ing here, there’s a prob­lem embod­ied, right, in this descrip­tion of the algo­rithm, which is an infor­ma­tion over­load prob­lem, which makes a lot of sense. But then again when you start think­ing about it, when was the last time you actu­al­ly faced tril­lions of web pages? Well, you could say that actu­al­ly I only face tril­lions of web pages because there is the search engine. 

So, to what extent, you can ask, can algo­rithms solve” prob­lems, or maybe are they just per­form­ing them? And there’s def­i­nite­ly a cou­ple of peo­ple in the room who’ve tack­led this spe­cif­ic issue, like Evgeny Morozov has talked about solu­tion­ism, or Daniel Neyland has actu­al­ly a project which I think is about can mar­kets prob­lems, is that cor­rect?, which is start­ing up. So there is lots of rela­tion— Especially when you think about the mar­ket and the invis­i­ble hand and the algo­rithm as the invis­i­ble hand. So there’s lots of inter­est­ing par­al­lels to think about. 

Which might go into the direc­tion of mag­ic, to a cer­tain extent. 

The fourth theme of course, it’s a very impor­tant one, is the theme of ethics, fair­ness, and account­abil­i­ty. There’s a lot of con­sid­er­a­tions in the provo­ca­tion piece. This is just a case we thought would be good to illus­trate that. This is actu­al­ly the ex-wife of the ex-president of Germany. And she actu­al­ly sued Google Germany because when­ev­er you type in her name, which is Bettina Wulff, the auto­com­plete func­tion would sug­gest escort,” this [pros­ti­tu­ierte] is the German word for pros­ti­tute, rotlicht” is red light” like red light dis­trict, and height, which is how tall you are. And she was not very hap­py with— [audi­ence laughs] I’d rather explain it. 

And she was not very hap­py with it, so she actu­al­ly sued Google. And this is still pend­ing but there’s anoth­er law­suit which has been decid­ed by fed­er­al court in Germany just three days ago, a very sim­i­lar one, where a com­pa­ny did exact­ly the same thing because when­ev­er you type in the name of the com­pa­ny, the words Scientology” and fraud” would come up. And in this case the fed­er­al court in Germany actu­al­ly decid­ed that Google needs to sort this out. This is defama­tion, at least in the German con­text. This is libelous. And so they need to do some­thing about it. 

Which then brings up inter­est­ing ques­tions. So how are algo­rithms impli­cat­ed in claims about ethics, fair­ness, and account­abil­i­ty? You could ask well what do algo­rithms have to do with this case. And actu­al­ly what they have to do with this case is that Google Germany said Well it’s not— We did­n’t do any­thing, right. It’s just the auto­com­plete algo­rithm. It’s what users search for. It’s the algo­rithm.” So this is a nice exam­ple of an algo­rithm becom­ing an object of the defer­ral of account­abil­i­ty. So you could say algo­rithms can become account­abil­ia, right. And think­ing along those lines might be quite fruit­ful we thought, and opens up lots of oth­er ques­tions about what counts as an eth­i­cal or not-so-ethical algorithm.

And final­ly, the theme of the con­fer­ence Governing Algorithms, who, which, or what is actu­al­ly gov­ern­ing what, which, or whom. And is algo­rith­mic gov­er­nance” a use­ful project to engage in? Like in the exam­ple before, is it the German fed­er­al court gov­ern­ing the algo­rithm but by man­dat­ing that Google needs to sort out their auto­com­plete results? Is it the engi­neers at Google gov­ern­ing the algo­rithm by doing such eval­u­a­tion? Or is the user expe­ri­ence,” this mys­te­ri­ous user expe­ri­ence, that is mobi­lized to opti­mize an algo­rithm and gov­ern it? Or is it the oth­er way around, that algo­rithms gov­ern us with a mal­ware warn­ing? This is a mal­ware warn­ing from two years ago which was on Yochai Benkler’s home page. He had a virus, and so when you want­ed to go to his home page there was this virus warn­ing. Or is some­thing like auto­com­plete, actu­al­ly, that gov­erns us? These are all inter­est­ing, incred­i­bly dif­fi­cult questions. 

And the best thing when you face all these incred­i­bly dif­fi­cult ques­tion is to ask incred­i­bly smart peo­ple to answer them. And that’s what we did for today. So this is the plan for today. 

As you can see, the day is orga­nized around four ses­sions which are high­light­ed in orange on this slide. We asked four very brave schol­ars to reflect on some of the themes men­tioned in the provo­ca­tion piece and come up with their own take on some of the issues, some of the con­fus­ing aspects of algo­rithms. And for­tu­nate­ly they’ve done so and they’ve writ­ten papers which you can find on the web site under Resources->Discussion Papers. 

In addi­tion, we have asked, for each of these ses­sions, two even braver schol­ars to respond to these main papers. And you can also find their respons­es on the web site and they will also talk about this today. 

And final­ly, we have some­thing we call the clos­ing pan­el, with Evgeny Morozov and Paul Dourish. And these two guys will lis­ten very care­ful­ly all day to every­thing; they will write down every­thing you say. And then at the end of the day, they will basi­cal­ly reflect on themes that have come up, themes for future research, stuff they found weird, strange, inter­est­ing, or worth shar­ing. Then we’ll have a quick dis­cus­sion at the end.

Another impor­tant thing to men­tion is that we will have lots of breaks in between the ses­sions. Many of them are cof­fee breaks. There will be cof­fee. And I think also still for a bit longer, break­fast out­side all day. Not the break­fast but the cof­fee and the tea. There will be a lunch break where you’re on your own. We have pre­pared a list of places around here where you can go. And so just make sure you’re back by 1:30 because we’ll go on then quick­ly quickly.

And then final­ly, on the culi­nary part we thought maybe just if any­one is still up for any­thing tonight, we thought we could have drinks at the B Bar. We can tell you lat­er on today where it is and how this works. 

One impor­tant thing, one last thing I for­got to men­tion is that of course the for­mat of the ses­sions fol­lows the paper for­mat. So there will be the main pre­sen­ters speak­ing twen­ty min­utes about their paper, mak­ing their point. Then the dis­cus­sants will each have ten min­utes to respond to that. And then the mod­er­a­tors will open it up. And in order to par­tic­i­pate in the dis­cus­sion, there’s two ways, same as yes­ter­day. One way is to send a tweet with the hash­tag #gov­al­go and share you ques­tion there, or your com­ments, or your con­cern. And Dove, who’s stand­ing in the back, will mon­i­tor these tweets and try to feed them into the dis­cus­sion. The oth­er way is to just line up behind one of the micro­phones in the room and make your com­ment here. The advan­tage is that you’re not restrict­ed to 140 char­ac­ters if you do it right here in the room. So you might take advan­tage of that. It’s also more per­son­al because we can see you and know who you are, and we would like to ask you to intro­duce yourself. 

With that I think that’s it. Have I for­got­ten any­thing? Solon, Sophie? Helen? No. Okay. So then, I think we can start with the intel­lec­tu­al fire­works we’re going to burn today.


Help Support Open Transcripts

If you found this useful or interesting, please consider supporting the project monthly at Patreon or once via Cash App, or even just sharing the link. Thanks.