The Problem with Gig Work

[Music] Heidi was an insurance agent for 16 years in Florida in a region called the Treasure Coast, about 125 miles north of Miami along the Atlantic shore. It's a little bit of paradise with rivers and estuaries and palm trees. Today's story starts with her journey from full-time employee to gig worker. Really what kind of changed was motherhood for the third time. Basically I drove an hour away to North Palm Beach to my office. When I had my daughter, I don't know something inside of me changed and you know you're like I can't be this far away. Heidi isn't her real name by the way. She decided to use a pseudonym for reasons that will become clear later. Meanwhile around the same time in Denton, Texas, a suburb outside of Dallas, Willie Solis had slightly different reasons for starting gig work. I was working as a construction worker and I ran my own business since 2008. I was in the process of transitioning to a different state of Florida and moving my business over there. I remember it required some licensing and so there was going to be some downtime. Heidi and Willie didn't know each other. At least not at that point in late 2019 and early 2020. What they had in common was that they both chose shopping and delivery of groceries and supplies as the gig of their choice. And they were both providers for their families, not part-timers looking for extra money. The other thing they had in common was they both stumbled upon the same shopping app, SHIPT, which at the time primarily served target stores. It's since expanded to a lot of other big chains. The primary reason Heidi and Willie were attracted to SHIPT was their pay model. Seven and a half percent of the expected order total plus a flat $5. That was a very clear transparent pay model where I knew that if I had a $100 order I would make $12.50. If I continued to upsell the customer or make that purchase higher, then I would get more money. Upselling means that if Willie can't find an item in store at Target, he was allowed with SHIPT to contact the customer to see if he could substitute it for an item that cost more. Even with a 7.5 percent commission, that meant he could make just a little more on the order. Instacart, the main rival to SHIPT, didn't let you do that. And for both Heidi and Willie, going with SHIPT initially paid off. I really enjoyed it. The pay was equally as good. I was making roughly between $18 to $25 an hour depending on the day. I needed to make $6.50 every week and that was my bottom. That was just surviving. Golden days, I would do 40 orders-ish and I was really making anywhere between $11 and $1300 a week. And then over the course of a few months, something happened. At first, I was really noticing $200 to $300 in difference to the negative for me. I was then doing $50 to $55 in a week. So I was really working a lot harder for a lot less. So in order that was $100 an order, the shopper would be getting like $9.50. While it may not sound like a lot of a decrease, it is a significant amount when you start multiplying it times the number of orders that we do. The Pay Change came all of a sudden without any notice and it was really affecting people. The way that we are paid makes no sense. Why is this paying this and this is paying this? I mean, you could sit there and rack your brain for hours, days, weeks and you're not going to make sense of it. I had a conversation with one lady. She didn't know how she was going to make her mortgage payment or her car payment and feed her children. That conversation with that lady really struck a nerve with me and that's when I decided somebody has to do something. Heidi and Willie did do something. From sleep, this is High Fine Nation, Philosophy in Storyform. Reporting from Princeton University, here's Barry Lamb. The pandemic fueled an exponential surge in gig work, particularly in food service and grocery delivery. Many of us were able to make it through the pandemic because of gig work, as customers, as workers and as retail businesses. At the heart of the gig economy is data collection and algorithms. Every one on every side making any money from the gig economy is beholden to them. But these companies, like DoorDash, Instacard, Uber Reads and Shipped, have made their algorithms a trade secret. Today on the show, I bring you the story of how one rag tag group of workers tried to figure out what was happening inside of the black box. And what it shows about the future of gig work. In this highly unregulated space, the gap between what is legal and what is moral is as wide as anywhere else in big tech today. I originally did a version of this story for the WNYC Studio Show Radio Lab. That story appears in their gigaverse episode. This is the long form, High FI Nation version. The official story from Shipped during this period was that there was going to be a city by city rollout of a payment system where shoppers are paid based on effort made during a shop rather than on a flat commission. The company said that this was done for fairness reasons, but the calculation of effort was done by its proprietary algorithm. Shipped loves to say that we get paid on effort, but we don't know exactly how that number is being calculated. There's no way that the system accounts for things like how much an item weighs or how many steps you have to take from the bottom floor to the third floor. So we're always chasing our tails thinking that we're going to make more money if we just work ourselves even harder. You might be wondering why gig work companies are allowed to do this. The answer is intellectual property. Anything a company pays to design, even a pay algorithm, can be owned by it so that no one outside the company has a right to see it. Heidi, what bothers you more at this point? Is it how opaque the pay model is or the fact that it's consistently lower than it was before? I don't think I can put one in front of the other very honestly. They're both pretty equal to me because it all affects my bottom line and it affects me mentally, physically, emotionally. We don't have control or agency over the way that we're being paid. We don't have a clear understanding of how much we're going to make. They're trying to be an employer in certain respects when it suits them and they're trying to keep the work as there's free-lances in other respects when it suits them. Daniel Halliday is a political philosopher at the University of Melbourne. He thinks there's something wrong happening in the modern day gig economy and it's wrong in exactly the way a lot of workers say it's wrong. We have a very good sense as people in the workforce, what the difference is between being an employee and being a freelancer. And what is fair to do to one but not the other? Being an employee is a matter of giving up freedoms in exchange for the security of a fixed paycheck. On the other hand, giving up security to become a freelancer comes with it power that employees don't have. One of those powers says Dan is the ability to set prices. If you're selling your labour, it's your labour, you choose what to sell it for. So if I'm the dishwasher repair person or someone who comes around to your house to fix something, I am in a position to set my own prices and you the customer in a position to say hang on a minute I'd rather not pay that much can we sort of haggle about this. A company that takes away the right to set the price of labour from a worker is treating them not as a freelancer but as an employee. Other rights that come with being a freelancer is the ability to set your hours and the ability to where what you want to where. One of these freedoms are entitlements for employees who can be made to show up to work on a schedule and can be forced to wear uniforms. Another freedom that some of these apps are taking away from freelancers is the freedom of capital investment. This is the freedom to choose how much money you put back into your business and can be anything from buying new cars or clothes to new phones or computers. If you're an employee, you're an employer. is responsible for the expenses. And if you're a freelancer, you're supposed to have control over them. But in the gig economy, it's one more freedom a company can take from its workers. I once had a driver who was absolutely livid about the fact that Uber required that your car had to be maximum five years old. And he was driving his Toyota SUV, which was three or four years old. And he was telling me he had this wonderful BMW SUV which was a whole lot nicer. But it was now six years old. So it wasn't allowed to drive it anymore. And that's an example of, you know, the capital that Uber drivers use is the vehicle. And they can't actually choose what kind of car to drive. They have to abide by Uber's requirements. Vehicles are very expensive for us. The requirement to have a car within five years old will exclude a number of drivers, including people like me. I actually thought of doing Uber once during the University of Education. Really? Well, I thought about it for a few minutes. As a research project or because you were interested. You needed the money. I mean, come on. Well, maybe a bit of both. I don't know, or just to pass the time. It's before I had children. But I quickly found out that the car I've got was way too old. Are there examples of appearances? So how you dress? Yeah. Uber has this thing, Uber Black, where you've got a fancier car and the driver would often be wearing a tie. I looked it up. Uber Black doesn't have a formal dress code for drivers. But its requirements on vehicles is strict enough. You have to have a black exterior and black leather interior. And only a set number of models can be Uber Black vehicles. You have to have a 4.85 rating as a driver. And you have to have commercial, not personal auto insurance. Vehicles can't have dense, paint scratches or bumper stickers. And you're expected to dress accordingly. Another crucial freedom for Daniel is whether you have the freedom to access the market for your labor. Many gig-working apps can and very often do suspend or ban gig-workers who are underperforming, getting four stars rather than five. It's not obvious why a platform should have the authority to kick someone out of the workforce effectively. Why can't we just have variation in quality? Maybe you get what you pay for like you do in other kinds of sectors where freelance work is common. The freelancer can't be fired. They might well perform poorly and their customers may well be unhappy and that might affect their reputation and how much money they can charge. But they can just sort of keep going. I was taken in by this really insightful point because I was one of those people who reacted with, well of course they get to kick people off. If they're slow or whatever they have a reputation, but you convinced me. You convinced me by this example where if you have a plumber who's a shitty plumber, there's no force outside of just the market that prevents them from working. They can advertise and they have a reputation as that shitty. But they can still advertise being a plumber and there could somebody who can hire them and knowing that they're going to hire a shitty plumber. We allow that. Yes, and we might welcome it. Again, the contrast to Airbnb is quite instructive, where there is actually a given neighborhood quite a bit of variation in price. And that's got to be because of some kind of variation in quality, right? Right. And just to be clear again, you're not arguing that it's like wrong for the apps to ban certain drivers. You're just saying that they're treating people more like employees than freelancers and they're trying to have it both ways. Yes, yes. During Thanksgiving, there was a lot of lines at the grocery store. Willis O'Lease, gig worker. And the app gave you a certain amount of time to go through the checkout process and deliver. And if you didn't do it within that period of time, then you were late. And if you were late, then that counted against you. And it could lead to deactivation or you basically being fired. And that's what happened to Willi over Thanksgiving. He was suspended and had to go through a retraining process, leading him to lose wages over the course of the days he was prevented from shopping. It's also the reason Heidi is going by a pseudonym. She's afraid she'll get deactivated from SHIPT in retaliation for speaking out. SHIPT can deactivate for many reasons, or any reasons they like. Interactions like this, and being prevented from actually voicing them on SHIPT's Facebook page, led Willi to create his own Facebook group, the SHIPT list, where SHIPT shoppers could air any grievances they had about customers or the app without the fear of censorship. That page would prove useful as the rollout of the new payment algorithm happened. First, the stories came from Kalamazoo, Michigan, and then San Antonio, Texas, the first two cities that were hit with the rollout. Within about a two week period of time, I talked to over 650 people. This is to direct messages, and everybody was complaining about how aggressively they had lost income. So Willi had an idea. He would try and figure out what shoppers were getting paid for each order under the new model, and compare it to what they would have been paid under the old model. That way he might find out just what the change was between version 1 and version 2 of the pay algorithm. People were putting screenshots of what it was that they were making on the orders, and they would either send it to me to messenger if they were afraid that they were going to get deactivated, or they would post it up on the Facebook group on these screenshots. They would have information as to what the order pay was like, and how much they were making. Automatically, I was like, "Yes, I'm in. Let's do it." Heidi was already collecting her own data. Trying to figure out some way to predict her pay under V2. It was really a fan excitement, really. I don't have to do as much work for myself anymore. That was kind of the kicker for me. For a lot of gig workers, almost every single shop decision comes down to whether it's worth their time, effort, and gas money. Not having any idea as to how the algorithm was calculating their pay, amounted to a gamble that they would earn money rather than lose money on any particular gig. This wasn't a problem with the old pay model, which told you exactly how you were going to get paid before you accepted any order. And it isn't a problem with traditional gig work, when you could haggle with the customer about what exactly you're supposed to do for how much pay. The new pay model took all of this power away from shoppers. Now, banding together through this little Facebook group, shoppers from across the country could pull their pay data across hundreds and hundreds of shops. All of the screenshots went to Willie, who entered the numbers into an Excel spreadsheet by hand, comparing V1 to V2 pay. And the screenshots, they kept coming. That was a clear moment where I realized I couldn't do this on my own. It was just taking away too much of my own time, and I wasn't able to really work. I was basically consumed by just gathering all the data. I was spending very little time working and more time advocating. I really hurt myself, my daughters, and ended up having to help with some of our bills. My name is Drew Embrojie. I'm the digital director at co-worker.org. Co-worker.org is, among other things, an online platform for gig workers to start petitions to the app companies. The site was booming in the early days of the pandemic, when shoppers and food delivery drivers were all of a sudden essential workers in a hazardous environment. They did things like start petitions for protective equipment or hand sanitizer. Willie had some previous experience with Drew and the site, and looked to them for help with this new project. Willie and I talked and we said, "Actually, if we get enough data, can we prove that this is bad for shoppers? Is it bad for shoppers? Are there things we can learn about it that can help make it more predictable?" The goal of this was never, "Let's figure out the new algorithm." Because one, that's a highly protected trade secret that we would be in big legal trouble if we tried to replicate. The goal was always to say, "How can we help workers adapt to the new algorithm?" So at the end of a job, you get just a summary of what you did and how much you got paid, kind of like a receipt or a pay stub. And so workers had albums of screenshots on their phones. So I was like, "Will you send me everything you have?" And I'll try and reconcile it. An initial glance at the incoming data convinced Drew that he was in over his head. It would have been easy if pay was universally lower under V2 than V1. But it wasn't. It looked all over the place. Some people got paid higher on similar shops, some lower. You couldn't just tell looking at a spreadsheet, what the pay depended on. There's a difference between these algorithms. It's not linear. It wasn't something I could even begin to suss out. Then I said, "You know, I need help. I'm not a data scientist." And that's where I kind of talked to my coworkers. 'Cause anyone know anyone who's really good at working with data who can help me process some of this. My name is Dan Kalachi. I'm a PhD candidate at MIT at the Media Lab. Enter our next character. Dan at MIT. A young data scientist. And a former gig worker himself. If you thought the point of an MIT data science degree is to make a ton of money helping big tech companies turn data into profit, you'd be right. But occasionally someone like Dan slips through, someone interested in labor and labor history, and someone who is looking for ways to help workers benefit from the data that's already being collected about them on their phones. Someone who reaches out to co-worker.org to see if his skills could be useful. I met with Drew one afternoon just on Zoom. He showed me the spreadsheet and I was looking at the screenshots that they had collected immediately. I was like, we could do this automatically. We can take the screenshots and turn it into this spreadsheet line in an instant with a handful of code that I could write in an afternoon. At that point, the sky was the limit in terms of data collection. Any shipped shopper can send as many screenshots as they wanted, and no one would have to be standing by entering the data by hand. The next problem was, where would they send it? Not to Willie. The whole point was that he couldn't manage it. Maybe to Drew? And then I was like, well, I have experienced making apps that people use and thinking about usability for folks. Dan's idea was to make a website or an app, a kind of destination where shipped shoppers would go to upload their screenshots. He was thinking like an app developer and a researcher, not a gig worker. Ultimately, it was Willie and other shoppers that ended up with the winning design. Texting is the way to go. Nobody's going to be on a laptop. After every shop or at any shoppers convenience, they would take a screenshot and text it to a number, a bot. That bot would automatically put the data on the central spreadsheet. The team also gave shoppers an incentive to text the bot. Helpful information. Once you sent over 10 screenshots, we would tell you your average pay, the average percentage of your pay that was based on tips, and whether or not you were getting paid with V1 or V2 and how often, and your average pay difference between V1 and V2. And these were all commands that they could send to the bot. We started rolling out the full system in June of 2020. This is when we really felt like the texting bot is working. It's giving people info about their pay, it makes sense, people understand it because we rolled it out to a few workers and got feedback. And then it was Willie doing the hard work of organizing. Between June and October of 2020, the team made an online drive. To get as much data as possible from volunteers shipped shoppers. This happened at the same time that shipped was rolling out version 2 of the pay model city by city across the country. They got hundreds of people across thousands of shops, but since it was volunteer, they had little way of knowing whether this was a representative sample. Shipped, of course, had all the data from all shoppers, but that's proprietary information. Nonetheless, the team now had a data scientist on their side who could do as many different types of analyses as would be informative. What they found after the break. High-fi nation will return after these messages. Dan Kalachi finally ran the numbers after the end of the campaign in October of 2020. Remember the context. It's phase one of the pandemic. No vaccines, no N95 masks, no end in sight. Large numbers of people are suddenly wanting someone else to do all of their shopping and delivering. And a large number of newly unemployed people are looking to make ends meet. We found from about 200 people who had submitted long histories of orders that 41% of people were getting paid less under V2 than they were under V1. This meant that 59% of people who chose to submit screenshots were not doing worse under V2 than V1. There are some workers who were paid more under the new algorithm shift. Or as much. Yeah, or as much. A lot of people were getting paid around the same, minus or plus 5%. It's fair to say that shipped was paying a little bit more for labor on their bottom line under V2. Is that correct? Yeah, I would say that. Okay, let's look at all the orders that were paid under V1 and all the orders that were paid under V2 and just look at the pay difference. And when you do that, V2 pays out more. That's great. That's good. Things aren't looking that bad for shipped. The people like Heidi and Willie and all of the people who suddenly found themselves doing more work for less. They were actually in the minority. But it's when you disaggregate and you look at by person, how much a person is getting paid for the same work. Then you see that there's this whole group of people who are getting paid less. And not just like on average across weeks of shops. It's week to week to week. They're getting paid less under this new algorithm. The company said, "Drew and Broggy? Co-worker.org." Under the new algorithm, everyone's going to make the same amount if not more as they were making before. The shoppers that are making less are making 11% less on average. If you give 40% of your workforce a pay cut in the middle of a global pandemic, in the early months of a global pandemic, that's not a good labor practice. What are the 41% doing under that algorithm that makes them underpaid comparing V2 to V1? This was like the big clue that I found. The answer to my question actually requires Dan to do something he's just not legally able to do, which is to figure out how the V2 algorithm works. But in lieu of that, Dan ended up monitoring a technical blog written by Shipped Engineers, where they post little bits of algorithm design. It gave him a big enough clue to speculate as to what the new algorithm was doing. One of the problems they posted about was about estimating the amount of time that an order would take. So the way they do this is they obviously look at driving times, the distance someone's traveling, but the sort of innovation, quote-unquote, that they do here, is workers have to physically go in a store and pick out items. They did their best to estimate how long that would take the average worker. They did this by looking at the square footage of a store, people's walking speed, you know, number of items, how they're distributed physically across the store, take all those pieces, and turn it into the amount of time it would take someone to pick everything up, check out, get to their car. I think that the V2 algorithm was taking this thing that they had developed, this shopping time estimator, plus driving time delivery drop-off time averages, and then attempting to pay people $15 an hour. That's my guess. Why $15? I think $15 for a couple of reasons. One is, that's what workers speculate. They're more experts on this sort of thing than I am. They are living under the algorithm all the time and sort of are guessing constantly. And the second reason is, one thing I was able to do was estimate the average order length. So how long it takes someone to complete an order. I estimated it basically around an hour, and the average pay for V2 is somewhere slightly south of $15 an hour. If the algorithm is predicting how long it's going to take you to shop, the biggest point of variation between workers is probably just how long you're spending in the store. It could be that you can't find an item and that sits you back like 20 minutes. Or it could be that there's a long line that sits you back another 10. It could be that you have a bad leg and that sits you back on average like half an hour for every order. Like it could be that the 41% I'm measuring is just a 41% of people just shop slower than average. Dan speculations are probably not the full story. It might be that some of the 41% were consistently taking more expensive orders during the era of commissions. And all those earnings were erased under V2. It could be that the 41% were veteran shoppers who didn't have access to little bonuses and promo pay that shipped and a lot of gig companies use to entice workers when they start. For their part ship responded to the findings with a denial. They said that this was not a representative sample. They had all the data. They knew the full story and their data told us that they were going to be a real deal. the different story, but they didn't tell anyone what that story was. Shipped declined my request for a comment, pointing to their press releases from this period. It's indisputable though that many individual shoppers were making about 10% less under V2, despite Shipped initially claiming that no one would be. They've since backed off that claim, but maintain their new pay model is fairer. Dan Kalachi disagrees. So for one, I don't think $15 an hour flat is the solution. I think to be truly fair, you need to pay people, especially if they're full-time workers, which a lot of these workers are, enough that they can live a reasonable life in the place that they're working and living. $15 an hour doesn't do that everywhere. But it isn't just about the amount of compensation that matters for issues of fairness. There's a complicated argument we can have about fairness and what it means in the era of the app-based gig economy. Shipped for its part can argue that V2 is fairer than V1 because shoppers under V1 can earn high commissions on expensive but not particularly effortful items like watches or SD cards, while shoppers who have to buy cheap but heavy items like cases of water do more work for less pay. Instead, Shipped would rather pay everyone blindly the same amount. For how long they think the ideal shopper should be spending in a store on any given shop. Isn't treating everyone the same without regard for individual differences in circumstance the definition of fairness. For Dan Kalachi, it isn't far from it. It's reminiscent of old time studies and factory work. Back in the day, scientific management techniques were proffering that management should strictly measure and understand how long it took their workers to complete a task. If I had a prototypical worker, how long would this task take? And then based on that, they would set an hourly wage. And they would do the time studies repeatedly over and over again to make sure they were basically paying people as little as possible. And if you're too slow, we'll too bad. We're at the point with these jobs where you have enough data as a company to pay some in a fair wage without just relying on the average. These companies have so much information about these workers. They know how often they stop at gas stations. They know their car models. They know the mileage that they get. They know the tax credit they get for per mile travel on their vehicle. Some of these companies, if I'm expanding beyond a shift, know the monthly payment that they pay on their car. And so if you know all of these things, you have every tool possible to calculate a fair wage to support a worker. If you're trying to make an algorithm to pay someone a wage, it should take into account all of these different pieces. I mean, what you're describing here is, for management's perspective, it's better to do the average because then I could say that people who are slower are slacking. And I don't want to have to pay for the time that they're slacking off. That's the management's perspective, right? Yeah, exactly. And you're saying that this particular kind of algorithmic wage calculation is the modern version of that. Yeah, precisely. So you want the algorithms to compute individual fairness. You want it to go by, well, this person drives a car that is old and you know, only gets 10 miles per gallon. So we need to take that into account. I mean, that's what you want. I think it's tricky, right? Because doing that necessitates collecting a lot more data about people. It necessitates workplace surveillance. Well, I think if you're doing all that surveillance, like if you are collecting a lot of information from workers, I feel like that's probably the only ethical step. But the other direction is really just, if you're going to pay an average wage of $15 an hour, you know how long people are working for. You see when they accept a job when they drop it off. Why don't you just pay people $15 an hour? Why did ship change from version one to version two? Was it for precisely the reasons they said? Fairness. Only they had a very different conception of fairness. No matter how I look at it, even the interpretation of Dan's results that is most generous to ship, raises more questions than it answers. You found that on average, shipped is paying more to workers, even if it's paying 41% of workers less. Why wouldn't they just go back to V1? What's their interest in staying in V2? I think companies like shipped are willing to pay a kind of premium in order to reduce transparency. I do think the transparency is an enormous problem, or it's reason why you might shift to a new algorithm. It allows you as a company to adjust and change pay completely opaquely. This was the big shift for them from V1 to V2, but now it's V something. One of the most interesting things we found is that over time, the pay was going down for everybody. When we looked at the data set static, 40% of workers are making less under the new algorithm. But over time, if we looked at just the end of the data set, that number went up to 60%. If you looked at just the last week of data. And then we also saw that the average payout that everyone was receiving fell from the beginning of the data set in April to the end of the data set in October 15% over time. It really was over time, the app driving down the cost of labor. Right, and I think it comes back to what are these algorithms designed to do? They're designed to calculate the exact lowest wage that they can pay the workers to get them to keep doing the work. And so, of course, they're trying to soften the blow by sweetening it up at the beginning, and then really seeing how low they can get those payouts to be over time. Here's one data point from Heidi. She's had a customer over this period who's ordered the exact same things every time. Under V1 in 2020, her payout was $20. Then in the early stages of V2, it was 15. Now, even with 2022, 23 inflation, her payout is $9. Same customer, same work, same order. And you know, two years since we did that campaign, their algorithm could have changed dramatically in that time. Be paying people much less, much more, be paying different kinds of people, different amounts. They could have changed it so it had just, you know, based on your metro area or your demographic or something inferred about you. But there's no way of knowing. They get out of that by just shifting to a trade secret algorithm and plotting along. Hi, Phi Nation. We'll return after these messages. In one sense, Black box pay algorithms in the gig economy are simply an extension of what all employers are trying to do. They're trying to find the highest the customer is willing to pay for something and the lowest amount of workers willing to accept for a job so that they can make the most profit from the difference. At-based gig work is different in being so good at this that workers don't stand a chance when they try to be the ones who profit from their work to make a little more than the bare minimum they would like from a job. But the political philosopher Daniel Halliday believes that at-based gig work is more than just highly efficient capitalism. He thinks that withholding information from freelancers is another way these companies are treating gig workers immorally by unjustly depriving them of a freedom to which they're morally entitled. To put it slightly more specifically, why would some other party have the authority to withhold that kind of information, particularly if the consumer hasn't got a reason for that information to be withheld. If I have called a newbie on my phone, it's not obvious why either consumer has any interest in the driver not knowing where I'm going to go until they accept. So why would some other party have the authority to do it instead? Labour laws in different countries and in different states of the same country have inconsistent formulations of what makes for an employee and what makes for a freelancer. They all give a list of criteria that makes someone either one or the other. In some jurisdictions, workers are winning more protection and others app companies are winning. It is worth remembering that digital platforms or smartphone apps. And what they're doing, they're not. is something that platforms have been doing for a long time, which is putting labor into contact with consumers, and they just do it in a much more sophisticated way. Where there's a problem is how they've been able to operate, but they seem to have done, is created this situation, where workers are in this position of being expected to be like freelancers insofar as they're absorbing the cost of being a freelancer, but they don't get the benefits of the freedoms that freelancers get. - Dan Halliday doesn't think law has caught up to morality. At-based gig work wants to take away as many freedoms from freelancers as possible without giving workers any benefits of employment, and the laws allow them to do that. It doesn't have to be that way. Since I first reported this story last summer with Radio Lab, Washington, DC, and the state of Minnesota have both sued-shipped for evading employee protection laws by misclassifying ship shoppers as freelancers rather than as employees. Ship rival Instacart agreed to pay $46 million to the city of San Diego for misclassifying shoppers in the same way. Cities and states seem to be catching on to the idea that the injustice involved in the gig economy is a matter of classification. But that isn't the only way to see the problem. There was this question that was raised in your paper, which was, "Can higher compensation make up for the injustice of depriving freelancers of certain rights under the gig economy?" What do you think is the answer to that? - It's a very good question in general, whether just being paid more means you can't have as much of a complaint about what's expected of you compared to another worker. If my employer is paying me an enormous amount, can it make demands of me that it couldn't otherwise make? Offhand, it's things like the answer's yes. If it turned out that Uber drivers were making enormous amounts of money, would we be so bothered that the platform was in a sense having it both ways? The more I think about this, it's fairly obvious that the problem is low-paid precarious work. - Ultimately, it might just be that the freelancer, employee categories are just too rough. Too out of date for what employment has become. Instead, the more freedoms you give up as a freelancer to secure a certain kind of gig work, the more protections you're entitled to, or the higher your pay ought to be. The less you give up, the more power you have to price your own labor, to know transparently what work is required of you before you accept or decline a gig, then the more you're subject to the precariousness of the market. We should be able to be more or less employee-like in the eyes of the law. What can't be sustained is the status quo. With app-based companies operating as employers, with enormous workplace surveillance capabilities, extracting every last bit of value from workers, with their teams of PhD programmers, lawyers, and public relations professionals, while workers have Facebook groups. Drew, how does the story of "Shipped V2 End"? Yeah, it's kind of sad. We came out with this report. We hit them in the press. We had workers who were in Birmingham, and in Minneapolis, at Shipped and Target headquarters. And still, that wasn't enough, and they were able to be dismissed. That was demoralizing for a lot of workers. They continued to roll out the pay structure nationwide, and it's been like that ever since. Almost immediately, you noticed a morale change among shoppers. I was discouraged. I was upset. I was outraged. I was angry. You put in a lot of effort to traffic change, but at the end, you don't get the change that you ultimately need or want. Are there days I'm like, "How am I going to get milk and dog food and dinner for my family?" Yep, absolutely. I've started looking to go back into the professional world because at this point, I'm like, "Man, I can't." It kind of puts you between a rock and a hard place, because I still love what I do. I still love my customers. A lot of my customers actually depend on me. And I would hate to leave them. Be a lot of sadness the day that I have to tell people that I'm done. We are the ones who are treated as expendable. Well, we can just pay them less, because they're going to take it anyway. (upbeat music) Not once that I ever feel as devalued as I do now. (upbeat music) - Willie, are you still able to make a living? - Not through the gig economy. I have to actually take on another job, a regular W2 job. I'm basically working between the gig economy and a regular W2 job. - You're doing both part time? - It's full time. I'm working two full time jobs, basically. - What's your W2 job? - I work for a retailer, basically, like a convenience store, and I work overnight. - Wait, you're working for a convenience store, eight hours overnight, and then you're shopping eight hours a day? (laughs) - Yes. When do you sleep? - I have to find time. Whenever I can get a quick minute or so to break for myself, when I can take a break, but through the advocacy work and doing all these two full time jobs, it's a lot of work. And I don't get a lot of sleep. It is what it is, and it's something that I have to do right now. (upbeat music) - Which brings us to the future of gig work. The services these apps provide are indisputably good. Think about how many elderly disabled or at-home caretakers have come to depend on them. And most of the workers I spoke to actually like doing the work. You'd think that this would make gig work platforms the easiest business possible. Lots of customers paired with a willing workforce. But DoorDash lost more than a billion dollars in 2022. The value of Instacart dropped 75% in the last year. Uber reported $9 billion in losses in 2022 and has never been profitable. And all of this is happening with lawsuits piling up against these companies. The middlemen between the hidey's and willies of the world and their customers aren't doing too well. Okay, well Daniel, well you tell me, how would you like things regulated? - It's the pro market view. And then there's the sort of the pro intervention view or the pro state view. I'd like to see some kind of outcome in which there's just more diversification whereby some apps may stick to the existing model. Some apps may give workers more freedoms. And then there's apps that may be going the other direction where they continue to boss you around, but they give you a bit more by way of the protections of employees. - We are developing-- - Dan Kalachi, MIT. - Additional tools that work for other apps. I developed an app called Gigbox that basically serves as a work tracker for gig workers. Weeklock is an app. It basically turns your phone into a sensor that you can use to collect all sorts of data about your working life. Ranging from the ambient temperature and the warehouse you're working in to your total standing time while in a recording studio. And the key thing about it is that all the data that it collects, all of that is saved to your phone and not shared with anyone. And so you choose who you share it with, whether it's an organizer, an advocate, a researcher, like me or your union. - Wouldn't the sea change or the revolution come from for every ship, for every Uber, for every postmates? There is a fair wage version of it designed by someone like you that takes the same kind of data that these other apps take, but computes the fair wage version, the worker generated fair wage version. And then that app is put up next to this other app and it out competes these other apps for workers. Wouldn't that be the revolution? It seems like all the data is already being collected, as you say. It just takes someone to design an app not from a managerial perspective, but from a worker perspective. - I think that would be great. But I wouldn't change to the story that you just told, which is that I wouldn't design it. It would be co-designed or participatory designed with workers. (soft music) I think one of the problems with these algorithms, like I think the engineers at SHIPT were trying to make something that's fair. I think they were, but it wasn't designed democratically with workers, the algorithm. And we're starting to have tools that allow you to design algorithms in democratic ways. You get people to vote, to give their opinion, and it changes the parameters of the algorithm. I do think that maybe a fair platform work sort of from the worker perspective is in the future. And some of that's already happening. There is one in Colorado that is a democratically owned, so a worker owned cooperative that is developing a delivery. platform. It's working now. You can order food from it in Colorado. And they take into account every stakeholder involved from the restaurants to the workers to the customers in designing how the platform is working from the algorithm to the pay to the user experience. And I think more things like that are kind of the future of this sort of work. We have more tools, data like this and tools like this give us more opportunity to create participatory systems. We can do this. High-fine nation is produced written and edited by Barry Lamb. Story editor for this season is Eleanor Gordon Smith. Additional producer on this episode courtesy of Radio Lab is Becca Brustler. Production assistance provided by Maddie Hettler. For Slate Podcasts, Alicia Montgomery is VP of audio. Derek John is executive producer of narrative podcasts and Ben Richmond is senior director of operations. Follow High-Fine Nation on Facebook, Twitter and Instagram @HighFineNation. That's HIPHI Nation. Complete transcripts, show notes and reading suggestions for every episode is available at highfineation.org.