Bill Kerr: These days the market for freelance talent is global. Employees can work anywhere with an internet connection—so long as the job doesn’t require them to be in the same place as their boss. Companies can search worldwide for the very best talent at the lowest cost.

Until recently, only big companies had the infrastructure and reach to hire overseas. But now a wave of companies like Upwork and Catalant are linking remote workers with potential employers. These companies use a web-based platform—kind of like the Uber for white-collar work—where employers post their jobs and workers list their skills.

Chris Stanton: One of the things that platforms like Upwork allow is for the best person in the world to be working on your task.

Kerr: Harvard Business School [HBS] professor Chris Stanton says these digital labor platforms are meeting the needs of the full range of businesses—from Silicon Valley giants to smaller shops.

Stanton: The mom-and-pop corner store probably doesn’t have an expert web developer in-house, and so they’re going out and finding a new capability. It turns out there are a lot of people who have lots of different skills, and online sources make it very easy to find people who might have skills that your firm doesn’t have.

Kerr: These platforms allow cost savings and flexibility. But they also come with their own challenges for employers, like protecting intellectual property [IP] and monitoring workers.

Welcome to the Managing the Future of Work podcast. I’m Harvard Business School professor Bill Kerr. I’m pleased to be joined today by my colleague Chris Stanton to discuss the rapid rise of digital labor platforms like Catalant and Upwork. Welcome, Chris.

Stanton: Thanks, Bill. It’s a pleasure to be here.

Kerr: So, Chris, we’re going to talk today about digital labor and your research on how firms and employees interact in that space. Maybe you can just begin by describing for us what is constituted by digital labor?

Stanton: Most people when they hear “digital labor” probably have a mental image of pulling their phone out and getting on an Uber app and summoning what used to be a taxi to show up and pick them up at their curb; or to pull out a phone, browse a menu, and have someone come to deliver food. In a B2B context, the same technology that enables you to hire a taxi driver also will allow you to hire someone to do your expenses out of sub-Saharan Africa or to hire a programmer who might be located in India or in Russia. Digital labor 00s, themselves, are markets that enable those transactions.

Kerr: We’re most familiar with the examples like Uber or Lyft, where the service actually comes to you via the app. But you’re describing a whole host of activities, where the real development and delivery of the service happens abroad, and then there’s a digital transfer.

Stanton: That’s correct. These are transactions that might occur beyond your local labor market. You probably used to have your expenses done by someone who sat locally, and now that person might be sitting 5,000 miles away.

Kerr: How big is this market?

Stanton: That’s a very tough question.

Kerr: What makes it such a tough question? It seems like a straightforward question.

Stanton: You would think, but it’s relatively easy to measure shipping containers that cross borders, and it’s relatively hard, at least for national statistical organizations, to measure the content of digital packets that move across the boundaries of countries and international organizations. As a result, what we know about these markets tends to come from selected platforms that facilitate these transactions. But each platform is only capturing a tiny share of digital transactions and services, themselves. I would suspect that, if you had to ask me for a ballpark figure, this is probably a $10 billion to $20 billion per year industry from the United States hiring people abroad.

Kerr: OK. So, it’s the irony of having lots of data bounding everywhere, but our national agencies are not able to tap into it or assemble it in a way that reflects the new knowledge economy.

Stanton: That’s correct.

Kerr: I remember the very first time I interacted with one of these platforms. My father actually asked me to find a typist at Harvard. I kind of had to tell Dad that we really didn’t have the typists around Harvard anymore. But he needed a bunch of things transcribed. I went on all these platforms, and I was amazed at the many opportunities there were to engage people around the world on simple transcription. I also understand that there’s many services beyond transcription that are offered today. Tell us about some of the services that one can find.

Stanton: Really, you can find anything that doesn’t require someone to be physically present to do work. You can find programmers to build the latest concept for an app that you might have. You can find programmers to build yourself a database to manage all of the records that you might need for your business. Or, on the other side of the spectrum, you might find someone to do data entry to put all of the old log files that you have from records that pre-dated the internet or even digital services into a digital format. It runs the gamut.

Kerr: With such a large range, is there one type of company that typically uses these platforms, or is more ubiquitous?

Stanton: As I mentioned earlier, it’s very difficult to get a sense of the overall size or scope of the online digital economy for work. But I have data from one of the largest platforms that I’ve been working with for a number of years. In these data, I can tell you that some of the largest Silicon Valley firms are present, down to mom-and-pop stores that may have had no pre-existing digital presence before coming onto this platform. It just ranges from the highest of the high-tech to people who have very limited pre-existing technical experience.

Kerr: It sounds like, however, the use cases might be quite different; that in the high-tech company examples, they probably have somebody in-house that could do all the types of services that they are contracting, but they need it for the extra work or for the overflow work, whereas the mom-and-pop store wouldn’t have those skills or capabilities in-house.

Stanton: That’s right. That accords very well with at least my mental model of these markets. Some people are coming in, they know exactly what they want, and they just don’t have enough people in-house to do that work, but they know how to get from point A to point B. That might be the Googles or the Microsofts of the world. Others, like the person who runs the mom-and-pop shop, probably have never built a website before and may be relying on an outside contractor to both do the work and to provide knowledge about how to get to a final solution when they couldn’t otherwise do it.

Kerr: You mentioned that this market is very fragmented. Are the platforms competing with each other in different ways? Is one platform known for certain types of services and another for different types of services?

Stanton: Upwork historically has been the largest platform in this market. They do everything from data entry to high-end programming. Other platforms tend to be more specialized. Some platforms have moved toward certifying only top contractors or top freelancers in finance or in programming, and they don’t do the low-end work. Other platforms, like Mechanical Turk, allow you to hire people—without ever actually knowing who they are—to do very, very rote or routine human tasks that one day may be automated but now are somewhat difficult, like image recognition, or CAPTCHA typing, or classification of different types of objects, where someone is just saying, “This is a bird or this is a plane, but both have wings.”

Kerr: So it sounds like this is not going to be a winner-take-all market. Instead, you may have the Upworks being differentiated from the Catalants being differentiated from other providers.

Stanton: That appears to be the way that these markets or these platforms are evolving. In fact, Catalant, which you mentioned, was an HBS company that initially focused just on freelance management consultants and now has broadened their scope to include some programming tasks. But these are really sort of high-end types of services, where most of the people on these platforms have top MBAs or top PhDs. That looks like a much more narrow type of platform, but for a niche or for a specific vertical. I think many of these platforms tend to be evolving based on sort of a horizontal dimension—where there are different types of work being done through different platforms—and a vertical dimension—where some employers really need the top of the market and other employers can just find pretty much anyone to do a job.

Kerr: As we think about an employer choosing to use one of these platforms, wage differences have long existed across countries, but the typical concerns have been: “I just don’t know how good that worker really is abroad,” or “That worker is afraid that I actually won’t pay them once the job is done,” or “I’m worried that you’re going to steal my idea.” How are these platforms attacking these traditional questions about outsourcing work?

Stanton: You might want to take a step back and ask yourself why there’s a platform that’s needed, anyway, because Google is a great search engine, or Bing is a great search engine. You might sort of say, “Why doesn’t every worker who might want to supply services just put up their own little web page, and people can find them individually?” Why do you need a platform like Upwork or like Catalant, or any other type of intermediary to facilitate trade in these services? One of the things that you pointed out is that there tends to be a lot of asymmetric information here, which just means that it’s very hard to know something about someone who might be sitting in the Philippines when you’re trying to hire them from New York or San Francisco or, in our case, Boston. One of the things that platforms have done to attempt to overcome some of this information asymmetry is to implement things that might look familiar to you if you’ve ever purchased something on eBay—like rating sellers to allow you to get a sense of their reliability, or rating sellers based on their skills to give you a sense of …

Kerr: … the number of stars that somebody has.

Stanton: Yeah. Those stars tend to be important. People online end up with stars in very much the same way that a Hollywood movie star might end up being sort of a star in your mind, where you know a lot about that person from past work, because other people leave feedback and ratings. You can get a sense of what type of projects people have worked on in the past to enable you to better evaluate candidates.

Kerr: So we talked a bit about how I’d understand the skills that someone in the Philippines might have. How do they help me with the monitoring of the work that’s being done in terms of facilitating the payments? What are the services after I have decided to work with you on a project?

Stanton: One of the things that I thought was ingenious when I started to work in this area was, I didn’t know how someone in the US would actually trust that the person who claimed to be doing the work would actually be sitting down and working on something, and then billing time. Some of these platforms, especially Upwork in their early days (when they used to be called oDesk), built software that allowed employers to monitor people after they were hired. The way that this monitoring works is that the software, itself, will take a screenshot of what a contractor or what a freelancer is doing, and take this screenshot at random times, and send it back to the employer to see if the person is actually 1) working on the thing that they claim to be working on and 2) actually doing something to where their screen changes, so that it’s not just having something up on the computer while they’re lounging around and billing time. This gives employers quite a lens into what other people they’ve hired are doing, even though they’re not physically co-located with that person.

Kerr: And they measured keystrokes and a bunch of things that can be a little scary at times but also can overcome some of this asymmetric information as you describe it.

Stanton: That’s right. Maybe in the future you’ll even see eye-tracking or other types of more invasive monitoring than just keystroke recording.

Kerr: OK. Go back to also the IP question. You often hear people worried about posting their ideas or their projects up on these platforms because they’re scared the person in the Philippines may take the project idea and run and expropriate it.

Stanton: That is a worry that I hear about frequently. Many people might have different strategies to deal with this. Among the strategies that I hear about are NDAs [non-disclosure agreements] or other legal agreements that attempt to protect IP.

Kerr: Do the platforms provide mechanisms to have NDAs or similar types of legal documents?

Stanton: It varies by platform. It also tends to vary by the location of individual contractors. So Catalant, which sources many of their contractors from the US, will tend to enable more legal work or facilitation around IP transfer.

Kerr: Typically the employer and the employee will be under the same legal jurisdiction.

Stanton: Right. If you end up hiring someone in the Philippines or in India or in Russia, the lack of common legal jurisdiction probably means that you want to use other strategies in addition to just having a contractor sign an NDA. Some of those strategies might be breaking up work to where the individual pieces that a contractor sees aren’t going to add up to the whole work product. So that, if the whole is more valuable than the sum of its parts, this is one way to potentially overcome some of the IP issues that you might be worried about—sending valuable pieces of a company abroad in what may be an unsecured transaction.

Kerr: The task to the manager may change with respect to the platforms, because the platforms are able to take over some of the roles. But then being smart about how you approach them is an extra task.

Stanton: It’s an extra task for the manager.

Kerr: Which brings it into some of your interesting research. I wanted to begin with first a little more background about how are wages set. I’m an employer. I come on the platform. What is it that I see when I post this job to do transcription work or to do high-end app development programming?

Stanton: What would you think that you should pay someone who’s sitting in Manila in the Philippines, which is where a big chunk of contractors come from on Upwork or on other online labor markets? If you ask yourself that question, you probably don’t know. I certainly wouldn’t know if I were a first-time employer going on to one of these sites. It turns out that the way that most wages are set is through contractors proposing a wage to employers at which they’re willing to work, and then an employer can see all of the past work history that a contractor has.

Kerr: You can get a big range of proposals. There could be a 10- or 100-fold difference in the hourly rate that’s being proposed.

Stanton: That’s correct. You might see people proposing hourly rates that are close to the US minimum wage, and you might see people proposing hourly rates for the exact same job with the exact same job description that might be $80 or $100.

Kerr: What is it that your research has found about first-time employers?

Stanton: First-time employers have this problem that I just described, which is that they really don’t know what they should be paying for contractors who are from labor markets that really don’t look like their own. Contractors know that this is someone’s first time, because they can see that an employer has never hired someone else, because employers also get scored on their reputation or scored on their past performance. As a result, first-time employers end up with wage offers that look pretty different from employers who have done this before. What would your guess be about the way that these wage differences for first-time employers look?

Kerr: Well, I guess I’ve seen the research, so I have a little bit of a pre-clue. But I would say in most markets, you would expect that the first time I’m dealing with somebody, I give them a discount. I’d give some sort of incentive to work with me so that I can later build up a relationship. My son’s karate class, you get the first month free before you go and you need to start paying. I know you’re going to say something different. Tell us what happens on the digital platforms.

Stanton: That was my intuition as well. It turns out when you look at the data—and we’ve looked at millions of these wage offers—first-time employers actually end up getting higher wage offers than their more experienced alter egos. To the extent that we think this is being driven by just what a job applicant or a contractor can see about employers, it looks like they shade their bids up or they shade their wage offers up by about 7 to 10 percent when they can see that someone like you, Bill, has never done it before, compared with Bill as an experienced user on the platform.

Kerr: Now, you have the following question, which is: Is this that the employer, the “Bill” in this case, is naïve (and Bill is often naïve, so it works)? Is it that Bill is naïve? Or is it that Bill really doesn’t know what he’s doing, and so I’m going to end up imposing an extra 10 percent labor cost or pain on the employee, and the employee is actually sort of adding that into their estimate?

Stanton: That’s a great question. One of the ways that we have attempted to look at this is to look to see if the inexperienced set of employees tend to treat people differently. That might be some evidence that that inexperienced employer is, in expectation, going to be more costly to work for. Or to produce some pain for a contractor who’s working for them. It turns out that there appears to be minimal differences. The feedback score that is left looks to be the same. The number of hours worked looks to be about the same. It looks like your story that Bill is naïve tends to dominate.

Kerr: It is the more likely story.

Stanton: It is the more likely story.

Kerr: One of the feature of these platforms is they generate such a large amount of data, that you are, as the researcher, able to look at them and understand how the employer and the employee are interacting at many different points through their contract, and can combat this—whether or not I’m asking for repeated clarification questions, or that I have been able to give the employee the right information.

Stanton: That’s correct.

Kerr: Beyond this sort of wage-setting for first-time employers, we think about employers going on the platforms and searching for low price. It’s got to be one of the important dimensions to their choice. How much do they really scan the world for the lowest possible price?

Stanton: That’s a question that I think is super important when thinking about these markets, because one of the things these markets are meant to do is to democratize work and to allow people to hire from anywhere.

Kerr: Global access to digital labor.

Stanton: Correct. The question is, do people who are using these markets actually react to differences in prices, or are they willing to hire people outside of a familiar context? That’s a very hard question to answer. The way that I have thought about it is to use an experiment that looks at shocks to foreign exchange. Every contract in most of these markets is denominated in dollars, meaning that when you hire a worker in India, Bill, you pay that worker in US dollars and the platform translates the dollar that you pay into the worker’s local currency, which in this case would be the rupee. We know that exchange rates tend to fluctuate. One of the ways that exchange-rate fluctuations affect workers is it affects what they’re actually paid in their local currency. As a result, it also changes what they’re bidding for jobs. Using the exchange-rate variation …

Kerr: … in large part because they probably have an opportunity for a local job that is in the foreign exchange.

Stanton: That’s right.

Kerr: So, I’m an Indian comparing my local opportunities vs. bidding for US-dollar-based work in America.

Stanton: That’s right. You might take a local freelance job that is in rupees, or you might take a freelance job that is in dollars.

Kerr: So, when we have a big movement in the rupee, does that affect the amount of work that the US provides into India?

Stanton: It turns out that the answer to that question is: Not really. It looks like there are just different types of employers on these platforms, some of whom are comfortable hiring workers abroad, others of whom tend to use these platforms more to source labor from the United States. But you tend not to see people substitute from workers in the US to workers in India when workers in India become even relatively less expensive compared with workers in the US. It looks like there’s not as much substitution across places as you might have expected.

Kerr: You believe this is primarily someone’s comfort zone?

Stanton: Primarily someone’s comfort zone is what I would think the explanation is, given some of the other work that I have done looking at where people tend to hire.

Kerr: Can you tell us a little bit about that?

Stanton: In this work, one of the things that we did was we looked at people’s hiring patterns in different countries. What we found was there was a fairly big difference in terms of the persistent nature of where people hired. To try and get at this, what we did was we know the names of employers in this online labor markets database, and we classified whether those employers are likely to have an ethnic Indian background.

Kerr: Last name is Ghosh or Patel.

Stanton: Or something like it. It’s fairly likely that that person had a family history in India. All of these people, however, live in the United States. Take two potential employers that are hiring. One’s last name is Stanton or Kerr; one’s last name is Patel or Ghosh. It turns out that the employer with the last name of Patel or Ghosh is much more likely to hire workers in India than the Stantons or Kerrs of the world. What’s remarkable about this, however, is that the job applicants that they receive don’t know whether they’re a Stanton or whether they’re a Patel.

Kerr: They only know the employer name.

Stanton: They only see that the employer is located in a certain area, like Mountain View, California, but they don’t know the name of the employer, itself.

Kerr: Or even the company name?

Stanton: Or even the company name.

Kerr: OK. But this has persistence over time; that once I start working with India, I’m more likely to do my second, third, and fourth contract in India.

Stanton: That’s right, but the initial contract placement is highly determined by your own ethnic background and whether you have exposure to a country or not.

Kerr: This is both very interesting and ironic in that these platforms are meant to remove all these frictions toward global exchange of tasks and trade and labor services, and yet many frictions are resurfacing or taking on a new form.

Stanton: That’s right. The traditional frictions that you might have thought about in terms of global trade tend to actually show up online as well, even in contexts where the platform, itself, was meant to overcome some of these frictions. Just the initial conditions or the initial biases—or whatever it is that is driving the Patels to place a contract in India and the Stantons to place a contract in Australia—manifests themselves into widely divergent experiences on the platform they’re after.

Kerr: Chris, we’ve mostly talked about the employer side. Take us to the worker side. What does a job experience look like on the platform, and how do they build up over experiences or over jobs that they work on?

Stanton: You might imagine that if you’re a first-time employee attempting to sell your services in a market—where you don’t know anything about who might be hiring you—that that’s a relatively difficult proposition. You don’t have a resumé, or you don’t have a verifiable resumé. You might have a picture, you might have a text description about your background, but those are all things that might just be copied and pasted from someone else’s website or someone else’s CV, itself. So first-time workers on these platforms have tremendous difficulty breaking into the market. It looks like they end up working for relatively low wages in order to tempt some employer to take a chance on them in order to give them the first shot at developing their resumé. Then, after that, that employer might leave them feedback, that employer might leave a record for others about the quality of the transaction. And in response to that information, career trajectories are determined. So people who have bad first-time trials tend not to do much later on. People who are revealed to have provided a good result on that first job tend to have their careers take off.

Kerr: So, Chris, as the employer, I can give feedback both in the form of number of stars and I could also write text-based comments. Does one matter more or less for the future career opportunities for the employee?

Stanton: Our colleague who is in the Economics Department here at Harvard, Amanda Pallais, did an experiment to attempt to disentangle this. In her experiment, she randomly hired different workers who had never been hired on the platform before, gave them a task, and evaluated the quality of that task. And for workers in the treated group, she hired them and gave them stars. And for workers in another treated group, she hired them, gave them stars and a detailed comment. And then the control group was workers who weren’t hired into her experiment. She was able to compare workers who were randomly hired for a job and given more or less detailed feedback against this control that wasn’t hired. It turns out just hiring someone and giving them this feedback appears to …

Kerr: … giving them just the stars …

Stanton: … just the stars … appears to get their foot in the door and allows them to go on as long as the star is a reasonably good rating. Just the appearance of that star moves the needle quite a bit. My interpretation of this is that, if you, Bill, as an employer, know that you can’t interview everyone, but if you have seen that Amanda hired this person and left them feedback, you know that that person was probably vetted by Amanda. As a result, you might follow Amanda’s lead and be willing to hire this person, because you expect that she also vetted that person in the past.

Kerr: So, Chris, while most of us would think about hiring a worker directly, your research has also highlighted the rise of intermediaries on these markets. Tell us about an intermediary and what is that doing?

Stanton: Let me tell you about an intermediary by way of an example. One of the first intermediaries on the oDesk or on the Upwork market was an organization called CUE Code. And CUE Code is headed by a guy who lives in Siberia named Iv Guinnea. And Iv Guinnea went to the Technical University in Omsk and is a programmer. And Iv Guinnea decided to bring on his friends onto the platform. And everyone else who is in his organization, CUE Code, also went to the Technical University in Omsk, lives in Siberia, and now works on Upwork. And the way that this works is Iv Guinnea has worked on Upwork before. And he has a reputation, and his friends have not. And so, in order to enable his friends to break into the market, if his friends say, “I know Iv Guinnea,” they get to put Iv Guinnea’s reputation—or the collective reputation of the organization—in their own profiles.

Kerr: So Iv Guinnea is now taking on the vetting role.

Stanton: Iv Guinnea takes on the vetting role, and in exchange for taking on that vetting role, Iv Guinnea takes just a small fraction of his friends’ wages or takes some other type of cut—that maybe his friend brings him a fruit basket. But it looks like this collective reputation is enabling his friends to break into the market as a result of being a member of one of these intermediaries.

Kerr: OK, I can imagine if you’re in remote Siberia, it’s hard to escape Iv Guinnea if you were to later bolt and go off on your own. Does the platform generally allow people to be a part of one of these intermediaries for a while and start freelancing on the side?

Stanton: The way that the early oDesk founders conceived this setup was actually ingenious, because they worried about this problem as well. And so, if you come onto the platform as one of Iv Guinnea’s affiliates and you decide to leave, they have given Iv Guinnea the right to obscure all of the feedback or all of the reputation that you have earned individually. So this gives him a long-term incentive to attempt to bring good-quality people on, because he knows he can capture some of the long-term benefits of that leg work to certify and vet new freelancers who might be coming into the market.

Kerr: So it sounds like the platforms are encouraging of the intermediaries.

Stanton: That’s right.

Kerr: But it wasn’t something that they anticipated, either.

Stanton: It wasn’t something that they anticipated. In fact, a Russian coder went to the CEO of the nascent company that was called oDesk at the time and said, “Well, I have all of these other programmers whom I would bring onto the platform that would be beneficial to you, but what’s in it for me?” And the CEO and founder chatted about it and said, “Well, maybe we could make this work.”

Kerr: If you were a first-time employer or young employer, would you recommend using one of these intermediaries?

Stanton: That’s a good question. So experienced workers have been vetted by others, and, as a result, you probably don’t need to rely on the intermediary for someone who has been hired by past people who don’t work through an intermediary. For new people, the question is: Is the intermediary doing much for you? And the answer is: Yes, but you’re going to pay for it. So people who are certified by one of these intermediaries tend to earn higher wages, and, as a result, you’re going to get what you pay for. So you’re probably going to get a higher-quality person, but you’re going to pay for that higher quality.

Kerr: OK, and so Iv Guinnea is both able to do his vetting, increase the workers’ wage who’s joined his company, and also get a cut off of that wage. So he’s not just taking a cut of what the wage would have been otherwise, but a double boost?

Stanton: That’s right. The double boost is that it’s more likely that you’re going to hire that person, because that person is vetted, and you’re going to hire that person at a higher wage rate than you otherwise would.

Kerr: So, Chris, this whole digital labor world is fascinating, and it’s one that we, as researchers, are learning a lot about how labor markets operate through study of them. And yet, when you mention the size a little bit earlier, if you sort of add up all of this, this is probably still not the equivalent of a Wipro or a Manpower or a temporary help agency. Like, the net economic size of $20 billion, $30 billion is not enormous. So what are the largest barriers for the use of these platforms? What is keeping them from growing like an Uber or growing like an Airbnb in terms of much more rapid penetration?

Stanton: So in 2007 or 2008 when I started working on these topics, I said to myself, “This has the potential to radically change the way that we will see work done in the future.” And if you do a thought experiment on your own, and you said, “What is likely to be the firm that has the bigger impact: one that allows you to summon a taxi from your phone or one that allows you to hire anyone to do basically anything that can be digitally delivered?” you probably would have said, “Upwork, which allows you to hire anyone to do anything.” And it turns out the taxi model is quite a bit bigger. And it’s an interesting question as to why, and I can speculate. My mental model of this is kind of twofold. One is that established firms have to learn how to integrate people who are hired through this channel into their processes and systems, and that takes people’s time. And so taking people’s time is expensive and may not be worth the management time to learn how to use these channels for very established companies. The other aspect that I suspect is hindering some of the adoption of this is that, really, these platforms are targeted to people who are looking for services or capabilities that they don’t have in-house. And it’s quite difficult, actually, to hire and vet someone when you, yourself, don’t have the knowledge needed to spec a task and to communicate with someone to do that task. And so integrated consultants like Wipro or Tata or others do a lot of that intermediate work for you in order to just get from point A to point B without having to have you, Bill, learn everything about building an app or figuring out JavaScript.

Kerr: It’s a very classic question for our MBA students. They’re starting a new company and the number-one idea is, “Well, why don’t I get some tech development work done abroad?” But if I’m a non-tech founder, how am I able to actually spec out what that task should be and actually monitor the outcomes?

Stanton: That’s right, and I suspect that that is at least hindering much of what you might think of as the potential for these markets—to tap into technical talent that is distributed globally but might be somewhat hard to manage for people who don’t have the background to actually spec a task and vet people who would do the task.

Kerr: So, Chris, if we also generalize from the specifics of these markets, what are other things that we can take for the broader labor market—or for the way that firms are thinking about their personnel decisions—that digital labor is uncovering for us?

Stanton: One of the things that I have found interesting working in this area is to try and take some of the lessons that I have uncovered and to apply them within organizations. And there are a couple of things that are beneficial about markets. And so you may not need to think about hiring people who are located abroad when you’re thinking about trying to take some lessons out of this work, but to think about maybe creating markets within your own organization as a substitute for using some of these platforms. So the two benefits for markets that I tend to see are 1) self-selection and 2) competition. So what do I mean by self-selection? Any teacher knows that if you have students who are interested in the material, that you can just do a lot more with those students than when you’re teaching a class that is, say, the first-year required freshman calculus course. Half of the people in there aren’t going to really want to do it. And so if you set up a market …

Kerr: Half?

Stanton: Maybe more [laughter]. Three quarters to 90 percent.

Kerr: OK.

Stanton: So if you set up a market where you get people who want to do the work or self-select into doing the work, you can just do a lot more with those people. And so you get people who have better skills, better training, and better motivation if you assign work through a market. Most firms, however, don’t do this. So if you think about most firms and their processes, a task comes in, a manager might evaluate that task, and then they’ll say, “Hey, Joe, you can probably do this. Why don’t you give it a crack?” rather than farming something out to say, “Here’s the task, who do you think is a good fit for this task based on your ability to see the requirements?” and select into doing that task. And so the power of markets to get self-selection is potentially really powerful, because one of the things that platforms like Upwork allow is for the best person in the world to be working on your task. And the power of markets, internally, means that the best person at your organization, if you’re in a large organization, can be adopted to the best use of their time. The second nice implication of using markets is that it aligns incentives. And if you have a way, as a project manager, to slice and dice work or to make work more like a security—where there’s a measurable outcome and you can compensate someone for that measurable outcome—then you can actually provide incentives to get people to do individual tasks in a way that you might not have thought about in your classical production stream before. And so the selection aspect and the incentive aspect are potentially very powerful for managers who are thinking about taking some of these insights and applying them internally.

Kerr: And these incentives that you’re describing, are they holding contests for work to be accomplished?

Stanton: It could be contests, or it could just be that you have a price for a given task if it’s completed by a certain date. But contests are very powerful motivators to get people to supply effort and to apply solutions that may not be obvious. So one model is a model that is by Topcoder, where Topcoder will run contests between different developers, and they just get phenomenal solutions as a result of this model.

Kerr: OK, ways of finding the true fit for the role, and then also have them highly incentivized to deliver on those results.

Stanton: That’s right.

Kerr: And as you highlighted on the global nature of this, the Topcoder incentives and others have often had people from all around the world bidding on and competing in these projects that are locally sourced.

Stanton: That’s right, and not just bidding on but winning some of the contests that might result. So, if you think about NASA and data science challenges to put rockets into space, it’s often people in remote parts of the world who might win or do very well in some of these challenges to help NASA or other innovative organizations with problems that you, otherwise, might not have found a solution for.

Kerr: And you have an experiment underway to try to implement some of this inside a company. Can you tell us a little bit about that?

Stanton: In this experiment, what we’re actually trying to do is take some of these lessons and to apply these lessons to their work stream. This is at a large outsourcing firm. And what we’re trying to do is to run, in parallel, work that is traditionally assigned by a manager who says, “Worker A, you do this; worker B, you do this.” And, instead, what we’re going to do in parallel is to say, “Well, here are the exact same tasks, we’re going to set up a market for those tasks. Let’s see who does better—the market or the manager who has traditionally been using this work stream.” My bet is on the market.

Kerr: OK, but while your bet is on the market, what are the downsides for the market?

Stanton: Well, there are a couple of downsides, or at least a couple of things that you need to think about. The first is writing proposals or writing tasks that are clear enough for someone to understand what it means to get from point A to point B to deliver. And the ability to write a clear specification or to write what looks like a contract with a worker is a transaction’s cost that needs to be accounted for. So in very large organizations, these specs tend to exist already, and so it’s somewhat easier to move to a market. The other thing that I have found empirically is that the market types of work need to be sliced and diced kind of narrowly. So, you can’t tell someone that they’re going to be given a task that takes a year and expect them to deliver if you’re not giving them an outsized reward. So giving people small-ish tasks that are modular and can be accomplished relatively easily over the span of a few days or a few weeks is something that takes some management time and talent. And so, what determines the ultimate success of this is actually the management of it, or the input into who’s actually figuring out what the appropriate unit of work is to be placed on the market.

Kerr: I think, which you’re also highlighting—to go all the way back up to where we began with Upwork—is that the technology has progressed to the point now where markets can become much more viable inside of companies, in organizations where we might not have traditionally had that capability.

Stanton: That’s right. It’s just a question of knowing how to use those tools at one’s disposal.

Kerr: So, if you think about the future of work, Chris—summing all this kind of stuff up—do you think digital labor is going to become a much more prominent part of the future?

Stanton: That’s a great question. It’s a question that comes up in my teaching, and it’s a question that leaves students with opinions that vary from “Yes, certainly,” to “No way, you’re crazy!” I think—if I could identify the one factor that drives people’s differences in opinion here—it is the evolution of complimentary technologies that will facilitate work that looks more like the traditional model: so, augmented reality or better video conferencing technology or better technology that makes the digital work stream look more like a naturally occurring work stream that most people are used to.

Kerr: OK, so if we all have virtual reality headsets, where we think we’re sitting around the same conference table, you envision that as being a place where digital labor just becomes very viable.

Stanton: It removes some of the aspects of this model that kind of look unnatural now, other than maybe coordination across time zones or idiosyncratic schedules. But, otherwise, if you can put on a headset and make it seem like everyone on your remote team is in the same place, that’s a potentially pretty powerful innovation.

Kerr: Ironically, they describe that we’ll know that virtual reality has truly hit when people who are sitting around the same conference table decide to put on the headsets so that they can see the extra, sort of, possibilities through the technology.

Stanton: Yeah.

Kerr: Chris, thank you for joining us today to share your views on digital labor in the gig economy.

Stanton: Thank you, Bill. This has been great.

Kerr: And thanks to all of you for listening in.