Sareeta Amrute: Hi every­body. Welcome to Data & Society. My name is Sareeta Amrute. I’m the Director of Research here. It’s my sin­cere plea­sure and hon­or to wel­come you to data and Society for this dis­cus­sion inspired by Mary Gray and Siddharth Suri’s recently-released book Ghost Work: How To Stop Silicon Valley From Building a New Global Underclass.

Mary Gray is Senior Researcher at Microsoft Research and Fellow at Harvard University’s Berkman Klein Center for Internet and Society. Mary also main­tains a fac­ul­ty posi­tion in the School of Informatics, Computing, & Engineering, with affil­i­a­tions in anthro­pol­o­gy, gen­der stud­ies, and the Media School at Indiana University. Her research looks at how tech­nol­o­gy access, mate­r­i­al con­di­tions, and every­day uses of media trans­form peo­ple’s lives. And today she’ll be talk­ing to us about her lat­est book, writ­ten with Siddharth Suri who’s based in Seattle. Take it away Mary.

Mary L. Gray: Thank you. Thank you every­one for com­ing out. And I see some famil­iar faces and I just real­ly want to voice my appre­ci­a­tion for all the sup­port I’ve had over the years doing this work. First and fore­most to my coau­thor Sid Suri, but to all the work­ers who have giv­en their time and let us into their lives to learn about their expe­ri­ences. This work would­n’t be pos­si­ble with­out the time that they’ve giv­en to us. 

So with that I want­ed to start by giv­ing you a sense of where this work came from. And for me, I was think­ing about my own research ques­tions before com­ing to Microsoft Research. Most of them cir­cled around the ques­tion of how do we become more or less seen? How are we known and val­ued as peo­ple? And what role do tech­nolo­gies play in that? And much of the herald­ing of the Internet was that we’re going to become more vis­i­ble. We’re going to be able to speak our truth, hear all voic­es. And much of my ground­ing in anthro­pol­o­gy and crit­i­cal media stud­ies brings me to the ques­tion of how so? When is that not true? And what are the con­di­tions under which peo­ple make that more or less true? 

So, I come to this project with that back­ground. And in many ways What I’m hop­ing to do is incite you to care about this world of work that is more or less seen, known, and val­ued, depend­ing on where you are in this uni­verse. It real­ly start­ed with com­ing to Microsoft Research and ask­ing a basic ques­tion about how arti­fi­cial intel­li­gence is made. I had no idea. 

And so when I start­ed ask­ing com­put­er sci­en­tists and engi­neers in my lab what goes into devel­op­ing algo­rithms and the mod­els that are built to be able to advance arti­fi­cial intel­li­gence, it turns out that there are a lot of peo­ple involved in that work out­side of the coders and the engi­neers and com­put­er sci­en­tists that are the­o­riz­ing these tech­no­log­i­cal inno­va­tions. It’s a lot of peo­ple who are effec­tive­ly clean­ing and man­ag­ing data, the train­ing data that become the mod­els for build­ing algo­rithms out. And there isn’t a case of any arti­fi­cial intel­li­gence that exists that does­n’t depend at some point in some­one touch­ing that data, curat­ing that data, and tak­ing some­thing that’s oth­er­wise kind of struc­tural­ist non­sense and putting it into some struc­tured sense that a com­pu­ta­tion­al process could then mod­el and learn from. 

So the goal of this book, if there’s noth­ing else you take from this book, is to under­stand that arti­fi­cial intel­li­gence always has human hands in it. That we are ben­e­fit­ing from a lot of peo­ple con­tribut­ing to advanc­ing these tech­nolo­gies. Even in cas­es where we might ful­ly auto­mate one process along the way, par­tic­u­lar­ly as its impact or its appli­ca­tion to a domain it was­n’t expect­ed to enter. Say lan­guage, like text trans­la­tion that’s done in real time. If you’re a speak­er of mul­ti­ple lan­guages and you’re code switch­ing, odds are pret­ty good that the AI isn’t going to be able to keep up with you. So look at those cas­es where you then have to bring peo­ple back into the mix to be able to devel­op a mod­el that would be able to cap­ture what kind of exchange is happening. 

That’s real­ly the begin­ning of this book, is to under­stand who are the peo­ple who are doing all of this work. And it turns out that when you ask com­put­er sci­en­tists and engi­neers, often their respons­es are, ​“I don’t real­ly know.” I’ve nev­er real­ly met these peo­ple. The beau­ty of this tech­nol­o­gy is that I don’t have to meet them. And I say that now with all seri­ous­ness. The sense is that this is a tech­no­log­i­cal inno­va­tion often called ​“human com­pu­ta­tion” or ​“crowd­sourc­ing.” The tech­nique of being able to thread a per­son into a moment of judg­ment where you need a per­son to be able to eval­u­ate or decide some­thing that a com­pu­ta­tion­al process can’t quite fig­ure out?, bring­ing that per­son into that moment, that judg­ment, and then thread­ing them into a com­pu­ta­tion­al process, an auto­mat­ed process so that you can car­ry on with an output. 

So, we’re some­what famil­iar with some of the appli­ca­tions. They’re are lit­tle bit vis­i­ble to you today. This is an ice­berg. Most of you are famil­iar per­haps with your Uber dri­ver. You’ve met them. Maybe you’ve chat­ted with them. You might be famil­iar with oth­er plat­form ser­vices, on-demand ser­vices, that effec­tive­ly are using the same tech­nol­o­gy of human com­pu­ta­tion to match a per­son who’s able to deliv­er a ser­vice through a mix of appli­ca­tion pro­gram­ming inter­faces that calls that per­son to the job, what­ev­er it might be. Whether it’s to pick you up at the air­port, or to pick up some food and bring it to your door. Or if it’s a con­tent moderator. 

And I think what’s fas­ci­nat­ing is two years ago if I’d said the phrase ​“con­tent mod­er­a­tor” or ​“con­tent mod­er­a­tion,” I would’ve just got­ten blank looks. How many of you know what con­tent mod­er­a­tion is today? It turns out they’re doing an incred­i­bly impor­tant job. They are peo­ple who effec­tive­ly curate, look at pieces of text and images that are beyond the capac­i­ty of any com­pu­ta­tion­al process to ana­lyze and eval­u­ate and say, ​“Ah. That’s pornog­ra­phy.” Or that’s spam ver­sus some­one shar­ing information. 

So it turns out that we’re not that far along in being able to eval­u­ate text and images to fig­ure out is that con­tent that should or should not be there. You could say any­thing that’s hard for a human to eval­u­ate and decide, ​“Is that mis­in­for­ma­tion or just a fact that I’m not famil­iar with yet?” the odds are very good a com­pu­ta­tion­al process isn’t even close to being able to fig­ure that out. If it’s hard for a per­son to fig­ure out it’s going to be intractably, a tech­ni­cal­ly hard process for com­pu­ta­tion to mod­el. You have to have real­ly cer­tain this or that, yes or no, to build code with accu­ra­cy to be able to auto­mate something. 

So again, take away how much this world that’s com­plete­ly depen­dent on hav­ing peo­ple at a moment of judg­ment enter the scene, like con­tent mod­er­a­tion, and then look below that sur­face. And that sur­face below that ice­berg, that is this spi­ral­ing, grow­ing, expan­sive world of ser­vices that effec­tive­ly are build­ing to keep a per­son in a com­pu­ta­tion­al loop. Because it turns out it’s much more effi­cient and effec­tive to be able to match a per­son to a task like cap­tion­ing and trans­la­tion, or a task that might be image tag­ging for a new set of images that you’re try­ing to eval­u­ate, whether it’s for train­ing AI or that you want to do a mar­ket­ing project. In all of those cas­es, all of these busi­ness­es that’re prob­a­bly unfa­mil­iar to you that are on this slide, are quick­ly mak­ing the best of a busi­ness mod­el that brings contract-driven, task-oriented work to peo­ple most­ly doing work in their homes, or if they’re in a set­ting they’re cov­ered by what are called ven­dor man­age­ment sys­tems. And again, they’re peo­ple that you will nev­er meet as an end con­sumer, but that you ben­e­fit from every day. 

So when I’m ask­ing engi­neers and com­put­er sci­en­tists about this work of human com­pu­ta­tion and the role of peo­ple in the loop, it turns out that most of these busi­ness­es are effec­tive­ly doing what these engi­neers are doing, which is bring­ing peo­ple in as quick­ly as they can and then mov­ing on to the next project. They’re not ask­ing who are these peo­ple? Under what con­di­tions might be be work­ing? And in most cas­es they’re work­ing on con­tract for that spe­cif­ic task itself. So the moment of engage­ment might not last more than a few min­utes at best. So it’s a pret­ty kalei­do­scop­ic world. 

So, at Microsoft Research I feel incred­i­bly lucky to be around peo­ple who do reflect on this ques­tion of what are they build­ing for the rest of the world. And in many cas­es when I meet a group of peo­ple who are say­ing, ​“I don’t real­ly know who are the work­ers who are here,” there’s at least a sub­set of those folks who will say, ​“I don’t exact­ly know but you know, the tech­nol­o­gy real­ly keeps me at a dis­tance.” And then there was a third set that would answer fair­ly reg­u­lar­ly, ​“I don’t know and I don’t know if I want to know.” 

And as you can imag­ine, for any anthro­pol­o­gists in the room that’s just…that’s…you real­ly want to pur­sue that ques­tion. What makes some­body uncom­fort­able about know­ing who is on the oth­er side of a screen? What makes it seem an intractably, socially-uncomfortable ques­tion to find out about their work conditions? 

So when I met Sid Suri, he was real­ly one of the first peo­ple who gen­uine­ly com­ing out of com­put­er sci­ence want­ed to not only know what work con­di­tions peo­ple might be engag­ing but what their lives were like. And so we start­ed on the jour­ney, and I don’t use that word light­ly. It took us five years to devel­op a method­ol­o­gy for being able to bring the val­ue of qual­i­ta­tive crit­i­cal work that engages peo­ple in their every­day lives and fig­ure out where you could inte­grate mea­sure­ment and com­pu­ta­tion­al analy­ses to build out a pic­ture of this world of work. 

So, often will get this ques­tion of well, how big is this mar­ket? Underneath that ques­tion is often peo­ple who feel like why should we both­er car­ing? This is work that’s going to be auto­mat­ed any day. If you buy the begin­ning premise of this book, and I hope you do, this work isn’t going away. The tasks will change. But in fact we’re build­ing towards a world of a ser­vice indus­try, infor­ma­tion ser­vices, knowl­edge work, that isn’t a niche job. This is the dis­man­tle­ment of full-time employ­ment. The dis­man­tle­ment of full-time employ­ment for any­one who does cre­ative work. 

So we might not be able to see how large that mar­ket is. Arguably we’ve nev­er done a head count. There is no effec­tive way to do a work­er cen­sus of an envi­ron­ment that is by design dis­trib­uted, glob­al, and often does­n’t have a cat­e­go­ry of work that peo­ple would rec­og­nize and res­onate with where they could say, ​“Yeah, that’s me. I do that work.” So this is both a world in which our old cat­e­gories of ​“what job do you do” is being blown apart. And, it’s a world in which we don’t have any mech­a­nisms for track­ing and hold­ing account­able the sup­ply chain that’s going into this world of work. 

But let’s take some guess­es here. And some of this is draw­ing on eco­nom­ics and the sec­ondary lit­er­a­ture about the pos­si­ble size of this mar­ket. Right now we know that there’s about an esti­mate of about 5% of the US pop­u­la­tion alone, accord­ing to Pew, that’s doing some form of online work. So at least in part, the work is sourced, sched­uled, man­aged, shipped, and built through an appli­ca­tion pro­gram­ming inter­face (an API) and the Internet; 5%. That does­n’t nec­es­sar­i­ly mean they’re doing their entire job online. It means that a form of income that’s impor­tant to them is com­ing from one of these jobs. 

Now this is real­ly strik­ing if you take into con­sid­er­a­tion we’ve only had the pos­si­bil­i­ty of mak­ing an income from this form of work for about a decade. So to have 5% of the US pop­u­la­tion already doing this work, start think­ing through the size of this mar­ket, the growth of this market. 

And if that’s not quite enough, think about how large the glob­al mar­ket for the busi­ness­es gen­er­at­ing val­ue will be by next year. It’s a $25 bil­lion dol­lar indus­try already. And that’s not a small num­ber if you think about how it com­pares to oth­er indus­tries that are fair­ly mun­dane. So that ice­berg that I showed you, all of those busi­ness­es that are spin­ning up below the sur­face of our vis­i­bil­i­ty as con­sumers, is build­ing an incred­i­ble amount of eco­nom­ic val­ue that up to this point does­n’t seem to be mov­ing to the oth­er side of the screen, to the work­ers them­selves who are doing the work.

Another pro­jec­tion here of think­ing about the impli­ca­tions of dis­man­tling employ­ment is to imag­ine this is not the dis­place­ment or the full automa­tion of occu­pa­tions and work. It’s the semi-automation of work and being able to task­i­fy it that is the tar­get of most of these indus­tries and the tech­nolo­gies that com­pa­nies are build­ing. Everybody would like to fig­ure out how to break down things like sched­ul­ing, man­ag­ing any sorts of appoint­ments, any of your work­flow. Figuring out how to break that down and turn it into a task that you can hand off to some­one else so that you can up what­ev­er it is that is your main point of view or val­ue, right. That this is again the object of most of the indus­try on build­ing out the tech­nolo­gies. At the rate we’re going, we’re look­ing at 38% of jobs with­in the US mov­ing to being semi-automated by the 2030s. 

Now, that might sound like a shock­ing num­ber but let’s just make it mun­dane. That means tak­ing most of office work, knowl­edge work, infor­ma­tion ser­vices, and turn­ing it into con­tract work. That’s already hap­pened, in a lot of places. So it’s not as though this is so futur­is­tic. And in many ways the goal of this book is to say it is not too late to rede­fine what this world looks like. We’re real­ly just at the begin­ning, but note that it’s mov­ing quickly. 

So the kinds of work that this entails… Again, because you can’t see it it’s often real­ly hard to describe. It’s every­thing from edit­ing, copy edit­ing, con­tent cura­tion. If any­body par­tic­i­pates in these kinds of jobs, you might see your­selves in these these tasks. Taking sur­veys, mar­ket­ing design, any sort of graph­ic design. Any sort of data entry. And label­ing, which is a pret­ty labor-intensive, cognitively-hard job. If you’re con­stant­ly look­ing at again, data that’s com­ing in that scraped from some­where with not a lot of con­text and you’re try­ing to fig­ure out what would you call this mate­r­i­al? Analyzing some­body’s atti­tude about a prod­uct; how would you assess that atti­tude? It’s actu­al­ly pret­ty chal­leng­ing work. And it goes very quick­ly, kind of task by task. 

We were study­ing very spe­cif­ic com­pa­nies. I’m going to just go through a bit of the method­ol­o­gy that we used. But we were look­ing at com­pa­nies that gen­er­ate sales leads. So you can scrape the Web and get an idea of who your con­tact per­son might be if you sell air con­di­tion­ers. But you’re going to do much bet­ter doing your sales if you know who you should call in that office. Turns out gen­er­at­ing sales leads, that’s a par­tic­u­lar kind of on-demand ghost work. Being able to take what is oth­er­wise just a web scrape of peo­ple’s con­tact infor­ma­tion, and curat­ing that list and fig­ur­ing out who should I con­tact, and then hand­ing it over to a busi­ness that wants the best con­tacts. That’s a very spe­cif­ic ver­ti­cal with­in this industry. 

And then trans­la­tion. TED has one of the largest open trans­la­tion projects that was the begin­ning of a vol­un­teer com­mu­ni­ty invest­ed in mak­ing videos avail­able for hard of hear­ing com­mu­ni­ties and for lin­guis­tic diver­si­ty. And it was the the heart and soul of ama​ra​.org, which is one of the orga­ni­za­tions that we studied. 

And then the oth­er kinds of tasks that again are becom­ing more famil­iar to some of you, con­tent mod­er­a­tion, clas­si­fi­ca­tion tasks, that are meant to opti­mize your search query expe­ri­ence. So if you’re typ­ing in something—and this will hap­pen hap­pen every elec­tion year. If you have a new can­di­date up for elec­tion, odds are pret­ty good if you’ve nev­er seen that can­di­date before they had to do some work to make sure that when peo­ple were search­ing that term, search­ing that name, that it matched to the prop­er biog­ra­phy or per­son­’s offi­cial pres­ence online. So it’s kind of this renew­al of a need for mak­ing sure the infor­ma­tion is relevant. 

I love the exam­ple of— How many of you remem­ber a moment dur­ing the past elec­tions when Romney made a ref­er­ence to binders full of women? Right. To be able to fig­ure out should that be a trend­ing top­ic? Because if you just think about that phrase, that’s a non­sen­si­cal phrase. Aside from who said it. And so real­ize that took a lot of con­tent mod­er­a­tors work­ing very quick­ly to be able to iden­ti­fy oh, the con­text for that. Oh, it’s an elec­tion cycle, a can­di­date debate. Yes, trend­ing. Makes sense. That’s hap­pen­ing below the sur­face. You will nev­er see it hap­pen, and it’s not some­thing that can be auto­mat­ed, any­time soon would be my argument. 

And then last­ly think­ing about these mun­dane uses of loca­tion ver­i­fi­ca­tion. How many of you had a favorite restau­rant that went out of busi­ness last month? Odds are if it moved some­where the loca­tion need­ed be rever­i­fied and updat­ed with­in search queries. That’s still very much the hand­work of peo­ple below the API. 

So, there’s a lin­eage here. This is not so new. And we take great pains to point out that the ten­den­cy to treat con­tin­gent work that seems like it’s going to go away any­time as there­fore not that valu­able and some­thing we don’t need to care about in terms of our employ­ment rela­tion­ships. That’s old news. 

So if you think about the expe­ri­ence of the Industrial Age and piece­work and the work that was lit­er­al­ly some­thing that could­n’t be accom­plished by the new­ly quickly-moving loom but that could be tak­en to most­ly fam­i­ly farms in the United States con­text, and be able to share that raw mate­r­i­al and the mate­r­i­al that’s been cre­at­ed, say a shirt. And to have the but­ton or the flour­ish of bows that could then be attached by a per­son. That’s the kind of work that the entire time the hope was even­tu­al­ly that would be auto­mat­ed away. And yes, even­tu­al­ly the machines were able to attach the flour­ish, the but­ton, the bows. That did­n’t mean that the work entire­ly dis­placed oth­er kinds of work that need­ed to be put on the table. 

This could go on for gen­er­a­tions, and did. In man­u­fac­tur­ing arguably. The only rea­son automa­tion can knock it out of the park is pre­cise­ly because you can build the fac­to­ry around the auto­mat­ed mechan­i­cal process­es and get peo­ple entire­ly out of the build­ing. But in any case where you’re work­ing with peo­ple, and effec­tive­ly when you’re try­ing to serve their inter­ests and antic­i­pate their needs, you’re in an entire­ly dif­fer­ent world of required task on a per­son­’s time and cog­ni­tive ability. 

So, if you think about the next gen­er­a­tion of lin­eage here, and the com­put­ers behind being able to put peo­ple in space or to be able to do some amaz­ing tech­ni­cal achieve­ments, we’ve always had these moments where the assump­tion was there’s some­thing rote and uncre­ative about this work. It can be done by, and often is done by, the same sus­pects gen­er­a­tion to gen­er­a­tion. But their work is not seen as inte­gral to what is real­ly val­ued and worth retain­ing or under­writ­ing through full-time employment. 

So for much of the women who were involved in the Cold War projects through NASA and through oth­er aero­nau­tics insti­tu­tions, they were on con­tract. They did­n’t have full-time jobs. They could be released at any point, and par­tic­u­lar­ly before it was ille­gal to fire women because they were preg­nant, women could be dis­missed as soon as they mar­ried because odds were good they would get pregnant. 

So think­ing about this lin­eage it’s impor­tant to imag­ine who is often invit­ed to take on these con­tract posi­tions pre­cise­ly because into the 60s they were imag­ined to be the per­fect tem­po­rary work­force. Both able to do the work, as in Kelly Girls, and sell­ing Kelly Girls as the oppor­tu­ni­ty for busi­ness pro­fes­sion­als to have some­one take care of their needs and then quick­ly exit to bring in fresh minds, fresh bod­ies, for the work that needs to be done around the office. So, you might see a pat­tern devel­op­ing here in this lin­eage of who is seen as replace­able or less valu­able and there­fore ripe for con­tract work. 

As we move into the ear­ly 80s to late 90s, and the Internet and con­nect­ed com­mu­ni­ca­tion devices allow the work­flow of office work that oth­er­wise seemed the domain of pro­fes­sion­als, from account­ing to human resources, to any sort of finan­cial ser­vices, it becomes quite easy to take that work and move it to oth­er con­ti­nents where you have enough lin­guis­tic capac­i­ty to be able to take advan­tage of labor arbi­trage, a cheap­er work­force, and still be able to get the work done. 

So, what I’m hop­ing you see in this lin­eage is pre­cise­ly the set of assump­tions that say who’s not so valu­able here? Who is the per­son that should hold this tem­po­rary job? Because we don’t nec­es­sar­i­ly need to care about them too much. And what are all of the ways in which con­tin­gent work, par­tic­u­lar­ly in the United States, set in motion a fram­ing of con­tract work as dis­pos­able, less impor­tant? Contingency becomes a val­ue propo­si­tion to the busi­ness, not to the worker.

So, the way in which we went about study­ing this, and I can go into this in the the Q&A, it’s real­ly hard to fig­ure out how to find peo­ple behind a dis­trib­uted sys­tem who are work­ing glob­al­ly in their homes. I’ll be the first to con­fess that. I’m some­one who real­ly likes to find the peo­ple that peo­ple assume are oth­er­wise real­ly hard to find. Because it turns out if you just ask peo­ple, ​“Hey, where are those peo­ple?” they’ll quick­ly iden­ti­fy them­selves and say, ​“Oh I do this work.” This was a whole oth­er lev­el of challenge. 

And so, the way we went about it was to find four insti­tu­tions, orga­ni­za­tions, that were pro­duc­ing this kind of labor and then to fig­ure out ways in which we could meet the work­ers who were par­tic­i­pat­ing in these labor mar­kets. We work agnos­ti­cal­ly with the assump­tion that this is con­tract work unless it’s oth­er­wise called some­thing else. And we worked with the terms of the peo­ple who were engag­ing in these projects. 

So we stud­ied Amazon Mechanical Turk, which set the base­line for how most of task-based work is framed and treat­ed, and the side­step­ping of any legal frame­works or clas­si­fi­ca­tion that go with it. 

And then we also looked at the Universal Human Relevance System, which is the inter­nal plat­form that Microsoft has. Please note, every large tech com­pa­ny has an inter­nal plat­form, that you can’t see, that’s larg­er than Amazon Mechanical Turk, that has far more work than you could ever track. Because when it comes to the account­ing of this work­force, it’s effec­tive­ly the equiv­a­lent of paper. It’s not con­sid­ered labor cap­i­tal. It’s just an asset being bought and sold through a pro­cure­ment farm, or through a pro­cure­ment office. 

So I say that cold­ly because impor­tant­ly, that’s part of the legal frame­work and arrange­ments that exist between busi­ness­es hir­ing out­side of their com­pa­nies to be able to have full-time labor on tap through oth­er firms that are the employ­er of record. So it’s a real­ly impor­tant lin­eage to under­stand how it pipes into the back of sev­er­al of these companies. 

The oth­er two com­pa­nies we looked at, or orga­ni­za­tions that we looked at, LeadGenius is a social entre­pre­neur­ship… It was a start­up found­ed in Silicon Valley that gen­er­ate sales leads. And they have a glob­al work­force, with a lot of recog­ni­tion that try­ing to do the work that they want to have done in the United States is not some­thing they could legal­ly do with­out being clas­si­fied as for­mal employ­ers. So they’ve moved a lot—almost all—of their work­force out­side of the United States. 

And then a fourth orga­ni­za­tion that we stud­ied, Amara, Amara On Demand. I’m fas­ci­nat­ed by Amara On Demand. And one of the cofounders and orga­niz­ers of it, Dean Jansen’s going to join us for a con­ver­sa­tion lat­er. But what fas­ci­nates me about Amara is that it start­ed out as a vol­un­teer com­mu­ni­ty, again doing cap­tion­ing and trans­la­tion of video, which is tech­ni­cal­ly a very hard prob­lem to solve. To be able to look at video in any robust way and inter­pret what are the actions that are hap­pen­ing in that video, that’s way beyond com­pu­ta­tion right now. 

And this com­mu­ni­ty of vol­un­teers effec­tive­ly became a mag­net for com­pa­nies and orga­ni­za­tions that want­ed to be able to trans­late and cap­tion their videos in oth­er lan­guages. So com­pa­nies start­ed approach­ing them say­ing, ​“Can we just pay you to do this fast?” So it cre­at­ed a labor mar­ket that from it’s very begin­ning was orga­nized by the vol­un­teer ener­gy and the atten­tion to the work­ers doing the work for this com­mu­ni­ty. And I think it’s a won­der­ful exam­ple of what we could be doing dif­fer­ent­ly in terms of orga­niz­ing these worlds of work around work­ers themselves. 

So, in look­ing at these four com­pa­nies it became an oppor­tu­ni­ty to start from there and to put out sur­veys on each of those plat­forms to be able to reach work­ers them­selves and ask them about them­selves. And at the end of those sur­veys to be able to ask them, ​“Would you be will­ing to meet in per­son for an inter­view?” And the map that you’re look­ing at… For work that is in the­o­ry work that can go any­where, that’s avail­able to any­one, you should see some pat­terns. There’s some­thing that maps onto the infra­struc­ture of out­sourc­ing, of places where there are not many job oppor­tu­ni­ties that are com­pa­ra­ble to ser­vice work at retail stores or in oth­er set­tings where the pay­ment is about the same as what you might be able to get in these online mar­kets. So note the pat­tern, because that should tell us there’s some­thing struc­tur­ing again, who does this work, where they do this work, what are the oth­er oppor­tu­ni­ties that it fore­clos­es or sug­gests are not available? 

And on top of those, once we had enough peo­ple inter­est­ed to do the inter­views it just turned into old-fashioned anthro­pol­o­gy in a lot of ways. Of go to where the peo­ple are, meet peo­ple, see who would be will­ing to allow us into their lives for a long enough peri­od of time to be able to under­stand the ebb and flow of their engage­ment with this work. So it becomes real­ly impor­tant for exam­ple to be able to be in India and see what hap­pens when the mon­soon sea­son hits, and how peo­ple then whether (no pun intend­ed) the work that they have to do, which is effec­tive­ly being online and hav­ing to be hyper­vig­i­lant to pick up tasks that are then delayed by what­ev­er might be get­ting in the way of their Internet access, for example. 

The lay­er… And you know, hats off to Sid for fig­ur­ing out the ways that he would be able to mea­sure this world. Because I have to con­fess, I did­n’t real­ly care that much about mea­sur­ing. I cared about if there’s any­body expe­ri­enc­ing this world that’s enough for me. And thank­ful­ly, he helped me see there’s a lot of val­ue in being able to under­stand the dis­tri­b­u­tion of this work, the oth­er pat­terns. And so what I want to share with you is real­ly the out­come of merg­ing these two approaches. 

What we found was, and this is to me one of the most strik­ing find­ings in the book, is that there’s a real Pareto dis­tri­b­u­tion of par­tic­i­pa­tion. And when I say that, like so many oth­er pow­er laws that are out there it turns out there’s a con­cen­trat­ed few that are pick­ing up and doing most of the work in these mar­kets. So a good…depending on the plat­form, a good 10 to 15, 20% at most are pick­ing up most of the tasks to be had. 

And then there’s this core group of peo­ple that we call ​“reg­u­lars.” That first group, we call them ​“always on.” And they lit­er­al­ly are. They’ve turned this into an income stream that main­tains their liveli­hoods. And they might have oth­er income streams and often they’re work­ing on mul­ti­ple plat­forms. They’ve turned it into full-time work for themselves. 

The sec­ond group. There’s about 30% at most but clos­er to 20, are what we call reg­u­lars. They’re step­ping into this, they’ve sunk the costs of fig­ur­ing out how to make these plat­forms pay off. They’ve learned what they need to learn. And impor­tant­ly they’ve con­nect­ed with peers. Much like the always-on have, they’ve con­nect­ed with dis­cus­sion forums, oth­er peo­ple who help them man­age and fig­ure out how to reduce their costs get­ting this work done. They are the bench, the deep bench, that is always able to step up into this labor mar­ket and pick up a task and do it. And it is what allows any­body who’s always on to step away and not have the entire mar­ket just fall apart. The argu­ment we have here is there’s no way to turn this into ful­ly on time, always on work. To turn it into full-time work pre­cise­ly works against what it is that peo­ple who have entered this world have said is impor­tant to them about enter­ing this labor mar­ket. I’m going to come to this in a moment. 

But last­ly and most impor­tant­ly, there’s a good 70% of peo­ple who walk into this—we call them ​“experimentalists”—try it, and they’re like, ​“Peace out. Don’t wan­na do this.” They have a range of rea­sons they decide they don’t want to do it. All of them are still pro­vid­ing val­ue, both to the com­pa­nies that are able to claim that they have 500 thou­sand work­ers on demand. So think about any time you’ve used Lyft or Uber, being able to see enough of those lit­tle cars that tells you okay, I’ll both­er. That’s the val­ue all the exper­i­men­tal­ists are bring­ing to this market. 

And pick­ing up one task or two tasks, it is pre­cise­ly being avail­able that they are offer­ing. Being will­ing and avail­able. That’s the most valu­able thing that they’re doing. And I think it rais­es this ques­tion of, ​“And yes, isn’t that valu­able?” Why isn’t that con­sid­ered valu­able? They’re bring­ing an incred­i­ble amount of val­ue cer­tain­ly to the busi­ness­es. They weath­er all of the costs. These are in most cas­es the peo­ple who could not fig­ure out how to tap into a com­mu­ni­ca­tion net­work to make this man­age­able, often felt iso­lat­ed, and alienated.

So, col­lab­o­ra­tion is key in this envi­ron­ment. The thing that allowed peo­ple to find their foot­ing; to be able to make enough mon­ey to make this worth­while; and again, the trade-off being being able to do it on par­tic­u­lar terms, was tap­ping into a net­work. This is from an exper­i­ment that Sid Suri and one of our coau­thors Ming Yin devel­oped to be able to iden­ti­fy what were the com­mu­ni­ties that peo­ple were engaged in—discussion forums. The dif­fer­ent col­ors are the dif­fer­ent dis­cus­sion forums for—this is just for Amazon Mechanical Turk. So an incred­i­bly robust, rich, nuanced, com­pli­cat­ed envi­ron­ment of inter­ac­tion. And all of the small dots are the soli­tary workers. 

So the vast major­i­ty of peo­ple who haven’t con­nect­ed with some­body are there, still pro­vid­ing val­ue, but you’ve got this tight clus­ter and groups who are orga­niz­ing around spe­cif­ic com­mu­ni­ties who again real­ly scaf­fold with each oth­er and make this man­age­able work. 

If you have to work through the night, you plug in ear­phones, put the phone to charge, and talk all night.
Akbar, an on-demand work­er in Hyderabad, who talks with oth­er work­ers while he works nights. [slide]

I want­ed to share this quote from one of the work­ers talk­ing about how impor­tant it becomes to be able to con­nect with oth­er peo­ple doing this work. I think the deep irony here is that the plat­form builders assumed this is great work because you can do it alone, and you don’t have to inter­act with any­body, you don’t need any help. what they had­n’t antic­i­pat­ed, per­haps because they did­n’t have enough anthro­pol­o­gists and soci­ol­o­gists in the room, was that peo­ple might still be invest­ed in hav­ing social con­nec­tions. The social con­nec­tions are actu­al­ly incred­i­bly valu­able to get­ting work done. It’s just immea­sur­ably valu­able. And that’s part of this envi­ron­ment, is fig­ur­ing out how to rec­og­nize the val­ue of that connection. 

The moti­va­tors that came up most often, and this might sound trite, it actu­al­ly maps on entire­ly to the lit­er­a­ture we have already about how peo­ple talk about their work, what they val­ue from work, par­tic­u­lar­ly when they decide to be self-employed or to try and free­lance. It’s about con­trol­ling your time, and I’d like every­body to stop using the word ​“flex­i­bil­i­ty” if you please. Because this isn’t about flex­i­bil­i­ty. It’s about hav­ing oth­er con­straints on your time and need­ing to con­trol your sched­ule, often hav­ing to do with fam­i­ly care, elder care. child care, oth­er work respon­si­bil­i­ties and oth­er inter­ests. Having inter­est in oth­er edu­ca­tion. Other hob­bies, oth­er joys. And in most cas­es peo­ple doing that cal­cu­lus of ​“how can I make this kind of work sus­tain me so that I can make my life run the way I want it to run?” That’s aspi­ra­tional to be sure but it’s cer­tain­ly part of what moti­vates peo­ple to keep at this. 

The sec­ond thing they’re after is to be able to con­trol what they work on. And I’d imag­ine many peo­ple in this room share that. They’d do just about any­thing to be able to define what their project is rather than have some­body else tell them what to do.

And the third is con­trol­ling your work envi­ron­ment. And if you’re pushed to the mar­gins, con­trol­ling your work envi­ron­ment is not a ​“nice-to-have.” So peo­ple with dis­abil­i­ties, queer-identifying peo­ple, women who felt mar­gin­al­ized by the for­mal employ­ment oppor­tu­ni­ties in their area all talked about this work being away of relief from those oth­er constraints. 

I can make mon­ey wher­ev­er I am and work on things that mat­ter to me. All I need to do is take my com­put­er with me. I’m liv­ing my ide­al life.
Carmela, an on-demand work­er in the US who does ghost work to sup­port her dream of being a chore­o­g­ra­ph­er [slide]

So when Carmela is talk­ing about effec­tive­ly being able to turn to this work because she can then pick up her com­put­er and do it from any­where, I want to take very seri­ous­ly and not dis­miss her claim­ing that this is the kind of work that lets her live her ide­al life. It’s how do we rec­og­nize, take that at face val­ue, and still remain crit­i­cal of a sys­tem that might still take advan­tage of her desire to do this work, with­out some­how think­ing that Carmela is the prob­lem here? 

So, I want to move into think­ing about where do we go from here? What have we learned from the way these work­ers not only sur­vive this work but make it mean­ing­ful, that we could then use to rede­fine this world of work? Because I’ll con­tin­ue to say, this is ear­ly days. We have an oppor­tu­ni­ty to real­ly design this pur­pose­ful­ly, with peo­ple at the cen­ter of our equation. 

So I want to focus on two things. If you get the book, the entire con­clu­sion is just here’s what we could do. It’s my buck­et list, that comes from the buck­et list of the work­ers we engaged. 

But the one I want to focus on for this con­ver­sa­tion is to think about what it means to rede­fine the social safe­ty net and job clas­si­fi­ca­tion. I’m just going to say it, I would like to com­plete­ly blow up employ­ment clas­si­fi­ca­tion as we know it. I do not think that defin­ing full-time work as the place where you get ben­e­fits, and part-time work as the place where you have to fight to get a full-time job, is an appro­pri­ate way of address­ing this labor mar­ket. And par­tic­u­lar­ly if we con­sid­er that glob­al­ly there are so few peo­ple in the world who have ever had access to full-time employ­ment that pro­vid­ed any benefits. 

So let’s start orga­niz­ing our clas­si­fi­ca­tion and treat­ment of employ­ment with all of the con­tin­gent work we’ve seen in this lin­eage and imag­in­ing that we will be those work­ers in the future. And if that is the case, then the best thing we could do is say there are some basics here. That we’re build­ing a com­mons. We have a labor com­mons that every com­pa­ny ben­e­fits from being able to draw from. They can dip in and out of this pool. So, how are we going to sup­port that pool? How are we going to make it sus­tain­able? So that the val­ue propo­si­tion isn’t here, let’s just exhaust this pool and drain it. Because that’s a tragedy. We know of the tragedy of the com­mons. Apply the same log­ic. Imagine if Healthcare4all isn’t just a nice thing to do because it’s char­i­ta­ble, it’s like that makes busi­ness sense. You need a healthy work­force. You need peo­ple to be able to step in and out of it to be able to make it sustainable. 

Continuing ed, there was­n’t a per­son that we inter­viewed who did­n’t talk about how often they were going to online resources to be able to con­tin­ue explor­ing what were the kinds of mate­ri­als they need­ed to fol­low up on and choose to read to be able to do their next project. We all ben­e­fit from being able to do that. Everybody had a base­line of a lib­er­al arts edu­ca­tion. The big news is, a lib­er­al arts edu­ca­tion and learn­ing how to learn is the base­line for every work­er in this mar­ket. Because it’s cre­ative work. All of us had to learn how to learn as that base­line and then build from there. This is not spe­cial­ized work. This is using your brain all the time. So the basic edu­ca­tion that comes with crit­i­cal think­ing is key. And then being able to make every­thing else avail­able becomes critical. 

And then think­ing about cowork­ing spaces, I just want to leave you with this vision. When I saw all of the home office setups of every­body we inter­viewed, and at the end of the day thought of the health and safe­ty admin­is­tra­tion and what our work­place around health and safe­ty was set up to do, it was to pre­vent pub­lic health crises. We have a pub­lic health cri­sis in peo­ple set­ting up their home offices and not hav­ing resources to do it in a way that’s healthy. So I know that sounds small, but every munic­i­pal­i­ty could have cowork­ing space that is think­ing pub­lic health. That’s an inter­ven­tion at the pub­lic health lev­el. Everybody should be able to get to a space where they can get relief for their back, for their neck. Again, sounds trite, but if you’re doing office knowl­edge work, it’s critical. 

And then last­ly I want to dwell on this notion of a retain­er. So, there’s a lot of con­ver­sa­tion about uni­ver­sal basic income as a solu­tion here. What frus­trates me most is the fram­ing of that sug­ges­tion is that these are the poor souls who can’t out­run the robots. That does­n’t real­ly account for how much what we real­ly need is to retain the cog­ni­tive and cre­ative capac­i­ty of peo­ple to be avail­able to busi­ness­es, to each oth­er, for ser­vice work. So peo­ple should be on a retain­er, for sure. Give every­body who’s a working-age adult a retain­er that says here’s your base­line so that you don’t have to think about paid leave or unem­ploy­ment. But that lit­er­al­ly you have what sus­tains you to be able to step away when you need to step away. Have chil­dren, take care of some­one. And at the same time know that you are going to have the finan­cial means to get back into that com­mons, right. 

Very dif­fer­ent atti­tude than uni­ver­sal basic income. The idea that some­how the poor are going to rise against us, and there­fore we need to give them some small amount of mon­ey com­plete­ly miss­es where is this econ­o­my head­ed? It’s a ser­vice econ­o­my. It’s an infor­ma­tion ser­vice and care econ­o­my, where we’re car­ing for each oth­er. And so to be able to do that, we real­ly have to imag­ine how would we give every­body the basic sup­port finan­cial­ly to do that, to be able to come and go in that work? 

So, the last thing we have to do is on us. If you use any of these ser­vices, now is the time, and today is a great day to do it. There’s a strike by Lyft and Uber dri­vers. If you’re a con­sumer of any ride-hailing app, be think­ing about, crit­i­cal­ly, if you’re some­body who con­sumes with care, if you think about where you buy your clothes or buy your food, there’s noth­ing small about that move­ment. Consumer advo­ca­cy and boy­cotting have been incred­i­bly pow­er­ful. They led to the Bangladesh Accord, which was the begin­ning of hold­ing com­pa­nies account­able for the long sup­ply chain involved in get­ting the shirts on our backs and in think­ing about agri­cul­ture and places where know­ing the sup­ply chain became crit­i­cal to mak­ing sure the qual­i­ty of food, but the qual­i­ty of peo­ple’s work con­di­tions grow­ing that food was some­thing every­body could know, and then make your choic­es. And that will nev­er be enough. This is not a land of let busi­ness­es be kind to work­ers, or let con­sumers be self-interested con­sumers. It’s about set­ting up the pos­si­bil­i­ty for the right reg­u­la­tion and clas­si­fi­ca­tion for employ­ment of the future. 

So, the key take­away, oth­er than AI always requires peo­ple so we’re not get­ting rid of them any­time soon, is that this is a com­mons. This is a labor com­mons that relies on peo­ple com­ing in and out. And what will it take for us to be able to sup­port and val­ue labor? No mat­ter where it hap­pens, no mat­ter how many hours some­body puts in. Because in fact, the val­ue of these mar­kets, it’s aggre­gat­ing up what we do. It is lit­er­al­ly the aggre­ga­tion of every­body’s input. And from there, what it’s pro­vid­ing. So think Uber. It’s all of those dri­vers being avail­able to you. It’s not just the one dri­ver who took you to the air­port. So to be able to val­ue all of the aggre­ga­tion of that labor and say what will it take to say each of the peo­ple par­tic­i­pat­ing in that are equal­ly impor­tant to our liveli­hoods, to our lives? 

And with that, I want to thank my col­lab­o­ra­tors. Because there’s noth­ing like a book about this kind of work to make you hope­ful­ly always aware of how much— Everything that went into this book came from an incred­i­bly rich team of peo­ple bring­ing a range of exper­tise. From Greg Minton, who made the beau­ti­ful maps. To all the research assis­tants who were involved in doing the fact-checking. To the key research assis­tants that we had in India to be able to main­tain the con­tacts with the peo­ple we had met through field­work. So there was­n’t a per­son on this team who did­n’t do some­thing inte­gral, that if they weren’t here I don’t know that this book would be here either. So with that, thank you. 

Sareeta Amrute: Thank you Mary. That was amaz­ing. I’d like to invite Dean Jansen to join us here at the front of the room. Dean is the Executive Director and Chief Executive Officer of the Participatory Culture Foundation, the par­ent orga­ni­za­tion of Amara on Demand, co-leading PCF with Aleli Alcala. Amara is fea­tured in Ghost Work and we’re thrilled to wel­come him into the conversation.

So as mod­er­a­tor I’m going take the host priv­i­leges of ask­ing a few ques­tions of Dean and then Mary. And then I’ll open it up to the floor for discussion.

Dean, maybe I’ll start with you if you don’t mind? Okay. If you could tell us a lit­tle bit about what’s changed for Amara since the book was written.

Dean Jansen: Sure, yeah. This was a great ques­tion and one that you know, think­ing about what has changed as an orga­ni­za­tion, we’ve got­ten big­ger. When we start­ed, and when AoD was first start­ing, I think the staff was about nine to…twelve peo­ple? The ini­tial group of folks that were doing the trans­la­tion, around two hun­dred. And today the staff is clos­er to thir­ty. There are thou­sands of dif­fer­ent folks who have signed on and are work­ing with AoD at this point.

And then as far as just the mar­ket­place, that’s some­thing that you know…when we began there was obvi­ous­ly a robust trans­la­tion mar­ket­place. But the audio­vi­su­al sub­ti­tling and dub­bing side of things was still pret­ty nascent. And that’s real­ly start­ing to pick up more and more today. So those’re two areas that have changed significantly.

Amrute: One thing that Mary talks about in the book is this idea of the dou­ble bot­tom line. And she talks about that specif­i­cal­ly when it comes to Amara’s work, which tries to cre­ate fair and col­lab­o­ra­tive labor prac­tices for peo­ple who are doing trans­la­tion and sub­ti­tle work. And so what I want­ed to ask you Dean is, how do you respond to peo­ple who say that the dou­ble bot­tom line is a nice idea but it’s not real­ly prac­ti­ca­ble. That in fact a busi­ness like yours can’t sus­tain itself with that mod­el in place.

Jansen: Yeah, that’s a great ques­tion. And one that I think might be answered well with anoth­er ques­tion, which is what does it mean to be sus­tain­able from orga­ni­za­tion to orga­ni­za­tion? Again, in the space that we’re in, in trans­la­tion, we saw ear­ly on these kind of more-established play­ers had very high mar­gins. And so again just a just a ques­tion of, how do you define sus­tain­abil­i­ty? Sustainability for whom and for what? 

I think if you’re ask­ing more maybe about kind of the lower-margin end of things? We’re talk­ing about some com­pa­nies that are the biggest and most pow­er­ful and prof­itable in the his­to­ry of the plan­et. And so I think it’s a ques­tion less, in our eyes at least, of is it sus­tain­able? More a ques­tion of how can we make it sustainable? 

Amrute: I’ll just name MTurk as one of those com­pa­nies, since Dean was too polite to say it.

Another ques­tion I want­ed to ask about is real­ly play­ing off of what Mary was talk­ing about at the end of her talk. We can think about con­sumer advo­ca­cy. We can think also about reg­u­la­tion. In your mind from where you sit, what are the pos­si­bil­i­ties and lim­its on what a busi­ness enter­prise can do to val­ue and pro­tect micro labor?

Jansen: Hm. Another great ques­tion. I think… Well, let’s see. So, as far as the lim­i­ta­tions go, I think there are a lot of things that orga­ni­za­tions can and should be doing. And we were kind of jok­ing about the—not jok­ing, but a cou­ple of weeks ago talk­ing about how low the bar is when the bar is ​“rec­og­nize that these are human beings that are doing this work.” Early on when speak­ing with Mary, one of the things that real­ly struck me was her describ­ing con­ver­sa­tions she would have, as she men­tioned, with engi­neers who real­ly did­n’t see where all these lay­ers exist­ed with human labor in them. 

So to me in many ways, the work that Mary and Sid and all the peo­ple that have made this book pos­si­ble, all the peo­ple that Mary spoke with, shin­ing a light on it and mak­ing it vis­i­ble as like the first step in fig­ur­ing out… I think I’ve just…skipped your ques­tion, but in terms of like, zoom­ing out and look­ing at soci­etal­ly what can we do and how can we accom­plish some of these things? Just that first step of hav­ing some recog­ni­tion of who and where peo­ple are is real­ly impor­tant. But obvi­ous­ly on an indi­vid­ual orga­ni­za­tion­al lev­el, there is a ton that can be done. 

Amrute: Can you give us just a few exam­ples from your own experience? 

Jansen: Sure, yeah. So let’s see. In terms of… I mean, again, pro­vid­ing peo­ple space to com­mu­ni­cate and work with one anoth­er, one of the things that we found in trans­la­tion was there’s this kind of stereo­type of the lone­ly trans­la­tor. Someone who is work­ing in an iso­lat­ed sort of— And in many ways, that’s…again, I’m not a trans­la­tor so this is just my learn­ing and under­stand­ing of it. But his­tor­i­cal­ly peo­ple do…it has been a more iso­lat­ed kind of work. So for us, bring­ing the col­lab­o­ra­tive side of the tools and plat­form that we’re pro­vid­ing and build­ing, to con­tin­ue lis­ten­ing to what sorts of things peo­ple need to do their work bet­ter, and to col­lab­o­rate more effec­tive­ly. That’s been some­thing that’s been…that’s kind of the bread and but­ter of some of what we’ve been doing. 

Amrute: Thank you. Being cog­nizant of time, I will just ask Mary a few short ques­tions and then open it up to the floor. 

So Mary, one of the key con­cepts in your book is this para­dox of automa­tion’s last mile. Could you pull that out for us a lit­tle bit and maybe relate it to the lin­eage of contract/contingent labor that you put up for us, which in your book actu­al­ly very inter­est­ing­ly starts with the expe­ri­ence of slav­ery in the United States. 

Mary L. Gray: Yeah. I mean, to think about this para­dox in many ways is to grap­ple with what has been the use of the lion’s share of man­u­al labor, for exam­ple. And ear­ly days with you know, if we think about what defines moder­ni­ty, what defines our mod­ern era, it’s imag­in­ing we’re going to get our hands out of the soil and be able to put our minds to work. It’s an eru­dite notion of what does advance­ment look like? What is progress, and progress dri­ven by technology? 

And so the para­dox is that as we strive to pull our­selves out of man­u­al labor, I believe we start rec­og­niz­ing that it actu­al­ly takes quite a bit of cre­ativ­i­ty and com­plex­i­ty to be able to do any enter­prise, to be able to do any­thing pro­duc­tive. So start­ing the lin­eage with slav­ery in the book is to say that was real­ly our first labor law in the United States. It defined who could be owned, who could be used in a way that only treat­ed their bod­ies as valu­able, and did­n’t imag­ine that any kind of work that we do involves a human capacity…not to be too human­ist here…but a human capac­i­ty to be able to bring cre­ativ­i­ty, to be able to bring respond­ing to spon­tane­ity, to what­ev­er we’re doing. 

So that moment of delay­ing rec­og­niz­ing the val­ue, the real deep inte­gra­tion of cre­ativ­i­ty with every­thing that we do, I believe that is this moment. It’s this reck­on­ing with a para­dox we keep intro­duc­ing of think­ing we’re just gonna…we’re going to get the think­ing and the talk­ing out of it. It’s just going to be some­thing we can auto­mate. And then we’re just left with things that are real­ly hard, you know. The pro­fes­sion­al class will be able to do the real­ly hard work. It throws to the wind the idea that tak­ing care of…if you’ve ever tak­en care of an elder­ly par­ent? That takes a lot of think­ing. A lot of creativity. 

There’s a beau­ti­ful book that real­ly for me sets up the dis­cus­sion of the para­dox and it’s by Levy and Murnane. They’re a com­put­er sci­en­tists and an econ­o­mist who talked about the new divi­sion of labor. And they were try­ing to ana­lyze what is it that a com­pu­ta­tion­al process can do, and what is it that a human can do, that are real­ly dis­tinct? What is that division? 

And so we’re try­ing to the­o­rize why it is as we strive to auto­mate things we keep dis­cov­er­ing anew that the capac­i­ty to think cre­ative­ly is in the thick of it? And that as we keep push­ing to auto­mate and reach­ing for hav­ing some­thing else do things for us, we’ll keep dis­cov­er­ing those bits that are real­ly the heart and soul of what what humans do, which is sense each oth­er’s needs, antic­i­pate, and I joke often: be able to apol­o­gize when we get it wrong. Like, com­pu­ta­tion can’t do that. 

Amrute: I’m going to just ask one last ques­tion. I know there’s a lot of peo­ple waiting. 

One thing I want­ed to under­line which I loved in your pre­sen­ta­tion, and in the book, is the way that you use the task of Mechanical Turk itself to set tasks to get research for the book. I think this is a bril­liant method­ol­o­gy. So I just want to put a big fat line under that. And then based on that I want­ed to ask you if you could pull out for us the sig­nif­i­cance for you in doing transna­tion­al, com­par­a­tive work, espe­cial­ly as it relates to iden­ti­fy­ing the chal­lenges and oppor­tu­ni­ties in labor organizing? 

Gray: Yeah. No, I mean I think from the… To be clear, start­ing with India and the United States in some ways was fol­low­ing the labor mar­kets that Amazon Mechanical Turk had cre­at­ed by pay­ing in both cash and rupees. They had built this labor mar­ket by design, with­out prob­a­bly much thought, cer­tain­ly, about the kinds of work­forces it would cre­ate. And in draw­ing that com­par­i­son, it meant we were con­stant­ly able to pull out the places where con­nec­tion broke down. The kinds of nation­alisms that would spark around dif­fer­ent work­groups, for exam­ple. So the transna­tion­al com­par­i­son gives us a chance to see where do peo­ple throw bound­aries back up. 

And I thought it was most strik­ing in India how quick­ly a kind of pan-Indian way of ori­ent­ing to work, across all the plat­forms, also came to the fore­front. So the num­ber of women and men who would start refer­ring to each oth­er as sis­ter and broth­er to nav­i­gate the gen­der pol­i­tics of work­ing in set­tings with some­body of a dif­fer­ent gen­der. The amaz­ing use of English as a way of nav­i­gat­ing lin­guis­tic bound­aries, and what that can mean. The mesh­ing and side­step­ping of reli­gious and caste dif­fer­ences. Those com­plex­i­ties could come out, and in some ways being able to see the com­plex­i­ties around class and race that played out in the United States in dif­fer­ent ways, real­ly came to the fore with that comparison. 

The tech­nique of putting the sur­veys online…I think in many ways I’m real­ly ded­i­cat­ed to us always imag­in­ing that anthro­pol­o­gy, sociology…of engag­ing peo­ple’s lives means get­ting in their lives? I have a dif­fi­cult time feel­ing like stop­ping at a dis­cus­sion forum and let­ting that stand in for peo­ple’s expe­ri­ences can answer the ques­tions I’m inter­est­ed in. I think it has every­thing to do with the ques­tion you’re ask­ing. But for the most part the ques­tion we were ask­ing was ​“What does the rest of your life look like?” And so there was real­ly no way to ask that with­out mov­ing from those plat­forms into their liv­ing rooms, into the cafes they cir­cu­lat­ed in. And that became the the methodology. 

Amrute: Thank you. Who would like to ask a ques­tion? danah? 

danah boyd: This is Mary’s fault, she set me up. She was like, ​“You have to ask a ques­tion.” and I was like—

Gray: A hard ques­tion, please.

boyd: Yeah. Alright, ready? So, much of what you’re grap­pling with is a dynam­ic that we’ve seen iter­ate over and over again through­out his­to­ry. And you’ve point­ed to them by talk­ing about the enslaved peo­ple and as a form of con­tract with­in labor mar­kets, where dif­fer­ent ver­sions of cap­i­tal­ism have evolved in response to these dif­fer­ent gov­ern­ing struc­tures. And we keep see­ing a reg­u­la­to­ry move, and cap­i­tal­ism evolves. We are now at a late-stage cap­i­tal­ism struc­ture, where we’re not only see­ing the effi­cien­cies that are pro­duced by cap­i­tal­ism as an oper­at­ing sys­tem but rein­forced, as you point out, by tech­ni­cal systems. 

But we’re doing it in an envi­ron­ment, to sort of riff off of Sareeta’s point, where we’re not actu­al­ly deal­ing with bound­ed nation-state struc­tures. And I’ve noticed you’re not scream­ing to tear down late-stage capitalism…mmm, maybe yeah—you know I know you. But part of it is how do we actu­al­ly think about the bound­ary work of nation-state struc­tures and the evo­lu­tions of late-stage cap­i­tal­ism as some­thing that can actu­al­ly grap­ple with this so that it does­n’t just keep slip­ping and evolving?

Gray: So two things popped in my head the last week or so. One is think­ing multi­na­tion­als have fig­ured this out in some ways. They know how to cash flow. And so to take at face val­ue that multi­na­tion­als gen­er­at­ing quite a bit of wealth inter­na­tion­al­ly and cir­cu­lat­ing it inter­na­tion­al­ly can’t fig­ure out how to cir­cu­late and dis­trib­ute the val­ue seems—

boyd: [inaudi­ble] say­ing they should govern?

Gray: Uh, hell no. No, they should not gov­ern. No. It is to say there’s clear­ly a way to exchange mon­ey, glob­al­ly. I’ll just…I’ll say it flat­ly like that. And that means in terms of gov­ern­ing the rev­enue gen­er­at­ed, the redis­tri­b­u­tion of it, that is a glob­al con­ver­sa­tion. We know that multi­na­tion­als are reg­u­lat­ed in in spe­cif­ic coun­tries; start there. So, make the United States, make the EU, the bat­tle­ground for say­ing, ​“New clas­si­fi­ca­tion. Everybody gets basics. Go from there.”

I think it’s for me very frus­trat­ing to real­ize how many large companies—and we’re pret­ty much in a monop­o­lis­tic world here. How many of the large com­pa­nies that have merged and acquired each oth­er have been able to stay on the side­lines of a con­ver­sa­tion around uni­ver­sal health­care in the United States. That makes no sense. And it should be on every­body’s mind to be advo­cat­ing to these com­pa­nies, and it can’t be about just advo­ca­cy but cer­tain­ly mak­ing the case ​“Why are you on the side­lines? You ben­e­fit from being able to see this healthy workforce.”

And in fact one way to think about this is sup­ply chains and good work codes for enter­pris­es to say there should be a call for reg­u­la­tion that if you are hir­ing a ven­dor, you should make sure that ven­dor pro­vides health­care, just as you are required or at least ben­e­fit­ing from pro­vid­ing health­care. I think there’s a busi­ness case there. Can it hap­pen with­out reg­u­la­tion? No. We all need to say this is ridicu­lous that we don’t pro­vide basic health­care. And I would just keep mak­ing that case. 

But the sec­ond thing that came to mind is that it real­ly is on all of us to stop let­ting lib­er­al and neolib­er­al eco­nom­ics define the val­ue of labor. To call labor a mar­ket­place and not see the moral­ism with­in that… Like, I’m just going to spell this out just plain­ly. We have allowed peo­ple in elite posi­tions, both gov­ern­ment and pri­vate enter­prise, to say that our labor is the same kind of cap­i­tal as a set of records or a car. We’re not durable goods. Labor means some­thing else. So let’s stop assum­ing that the mar­ket­place sets the val­ue of our labor. My salary is not com­ing from some magically-discerned val­ue of me on a mar­ket­place. It’s com­ing from pow­er. I have a par­tic­u­lar kind of pow­er and priv­i­lege to com­mand a price. And that is a set of irra­tional pow­er moves. It’s not the log­ic of the mar­ket play­ing out. 

So I real­ly want us all to start ques­tion­ing. There should be no turn to the mar­ket ​“what is the mar­ket pay­ing for this task?” It should be, base­line, labor deserves sup­port, because we need healthy work­ers to move the world for­ward. I’m a prag­ma­tist, that’s why I’m not that inter­est­ed in blow­ing up cap­i­tal­ism writ large. I don’t know, you could prob­a­bly get me there, danah. 

But there are alter­na­tive ver­sions of mar­ket­places like coop­er­a­tive mar­ket­places. Marketplaces we haven’t even imag­ined yet because we haven’t let our­selves think that the mar­ket can’t set the val­ue of us. Nor should it. And that’s why that lin­eage start­ed with slav­ery. Why would we have ever imag­ined that the mar­ket, or a mar­ket, should define the val­ue of humans? Let’s stop that, and set a base­line. And then go from there. Perks, sure you get perks if you’re an extra good worker.

Audience 2: I think I might have a sim­ple ques­tion, but I’m sort of think­ing as you’ve been talk­ing about peo­ple’s pow­er and how we might choose a base­line that might be a retain­er or some sort of work for all. Do you think that that might be real­ly dif­fi­cult as more and more peo­ple are doing labor unwit­ting­ly? Like for exam­ple Google uses ver­i­fi­ca­tion to make sure you’re not a robot. And we’re like, ​“Great. That’s great. I want to prove I’m human,” but you’re actu­al­ly doing work for…you’re doing that task and you don’t know that you’re doing it. So do you think that like Hulu or sub­scrip­tion ser­vices are going to start mak­ing peo­ple do work in order to watch a video, and because they want to watch it they’re going to do work like in these micro­trans­ac­tions? Do you see that maybe killing the labor force, too, because all the work’s going to be spread out to every­one that are very will­ing to watch their con­tent to do it? 

Gray: So, love that ques­tion. Two real­ly opti­mistic rea­sons that I think that won’t hap­pen. One is that when peo­ple become more aware that when you check the ​“I’m not a robot” that you’re actu­al­ly doing work for a com­pa­ny that makes a lot of mon­ey and does­n’t actu­al­ly make your ser­vice bet­ter, we should all be say­ing, Hell. No.” And it’s not that we’re going to walk away from it, because we’re all enjoy­ing those ser­vices. It’s that we’re all going to become pub­licly aware that you are hav­ing your time tak­en. You are hav­ing your time tak­en with­out your per­mis­sion. That should nev­er be okay, right. So, one answer to your ques­tion is we absolute­ly have to have more pub­lic aware­ness of the sheer rob­bery of our time that’s happening. 

We make pains—the sec­ond response to your question—we make pains in the book to dis­tin­guish between paid labor and unpaid labor. And the rea­son to focus on paid labor is pre­cise­ly to say that it’s a dif­fer­ent rela­tion­ship. It’s a social— it’s not a char­i­ty when you get a job from some­body. You’re pro­vid­ing val­ue. It’s a busi­ness trans­ac­tion. Be cold about it. So that we can start liv­ing our lives and stop hav­ing work define who we are, right? Like this is— Kathy Weeks is one of my favorites on this. It’s like, why are we not fight­ing for the end of work rather than for a forty-hour work week? Hell, no, right? So let’s go there. And I think the begin­ning of it is with an aware­ness of when we are giv­ing away…and actu­al­ly I think that’s the wrong fram­ing. When our time is being tak­en from us. Let’s stop say­ing we’re giv­ing it away. Because our time is being tak­en from us. Without our per­mis­sion. And we have to say that’s not okay. That’s the first part. 

Amrute: Okay, last question. 

Audience 3: Thank you guys. This is just great. So I want­ed to ask about… I’m look­ing at the cov­er of the book and intense­ly won­der­ing who is this per­son. But it brings back the ques­tion, the top­ic of the vis­i­bil­i­ty of peo­ple doing this work and their rel­a­tive invis­i­bil­i­ty. And I won­dered if you could talk about that and talk about why or how that vis­i­bil­i­ty might be impor­tant to the big­ger equa­tion of change that you imagine. 

Gray: There’s not a day that goes by that some­body does­n’t tell me about their Lyft or their Uber dri­ver when I talk about this book. And so there’s clear­ly some­thing poignant and press­ing for peo­ple when the per­son that they’re wor­ried about is right in front of them. I mean and that’s a good thing. And I think in some ways let’s tac­ti­cal­ly deploy that if you’re an organizer. 

I do think we care. I mean you know, I’m a relent­less opti­mist. I think when we see some­one in pain we want to relieve it? Usually we save that for peo­ple we love the most, but our impulse is there. And so I want to believe that by rais­ing vis­i­bil­i­ty, by see­ing the peo­ple who do this work, that it changes what we want for them. You could just be crass and cyn­i­cal and say, ​“This could be you. So don’t you want to change your cir­cum­stances?” It will absolute­ly be some­one in your life. Do you want to see their life better?” 

So I think it is for me— It’s fun­ny you raised that. The cov­er was actu­al­ly real­ly… I was uncom­fort­able with that cov­er. I found it sen­sa­tion­al­is­tic. I’ll just say this, I love my press, I don’t know if they’re here right now. Clearly what they want­ed was to get this book in peo­ples’ hands and they want­ed some­thing that peo­ple might reach for. And it might have that effect. I hope it does. That was their rea­son. For me I’m now on this mis­sion to find out who is he, you know. And I’ll get back to you. I’ll post some­thing. I’ll find out who he is. 

Audience 3: [I mean?] how else can you make peo­ple vis­i­ble. Other than [indis­tinct] book.

Gray: Yeah. And that is actu­al­ly a real­ly com­pli­cat­ed ques­tion. Because at the end of the day, for the peo­ple I inter­viewed, they’re folks doing their jobs, liv­ing their lives, and they’re not par­tic­u­lar­ly inter­est­ed— And actu­al­ly for the folks I’ve asked, ​“Do you want to be on any of the TV shows I might be on?” And the first thing they say is, ​“No.” And the sec­ond thing they’ll say is, ​“I don’t have time.” And the third thing they’ll say is, ​“I don’t want to be seen as one of those peo­ple who’s a vic­tim.” So what I don’t real­ly appre­ci­ate is most of the cov­er­age out there frames the peo­ple who do this work as vic­tims, or dupes, or unaware of their cir­cum­stances. The peo­ple doing this work are painful­ly aware of their cir­cum­stances. And we would do best by lis­ten­ing to them. I think we could just lis­ten rather than need to see them. This isn’t real­ly about mak­ing you see every­body who’s had a hand in mak­ing your social media palat­able. It’s about know­ing there are peo­ple who are help­ing you. And they’re work­ing. So how do we sup­port them? 

Amrute: Thank you very much. Buy the book. It’s over here. Thank you, Dean.