[01:04.980 --> 01:05.720]  Welcome back.
[01:06.100 --> 01:07.480]  How's the Sunday coming along?
[01:07.720 --> 01:08.240]  All right?
[01:09.320 --> 01:09.940]  All right.
[01:11.440 --> 01:12.020]  All right.
[01:12.240 --> 01:12.820]  This is it.
[01:12.960 --> 01:18.000]  You didn't quite make it, but you almost made it through the endurance test, which is a new hope.
[01:18.180 --> 01:19.900]  There's so much to do, so much to see.
[01:20.160 --> 01:21.060]  Fantastic program.
[01:21.060 --> 01:22.200]  Thanks to all the speakers.
[01:23.860 --> 01:26.580]  Before we get to this next talk, I'll make a couple of brief announcements.
[01:27.040 --> 01:30.100]  First is that closing ceremonies at 6.
[01:30.360 --> 01:33.140]  We're going to be simulcasting that to room 206.
[01:33.140 --> 01:37.180]  So if this room gets a little full, 206 is a place to go rather than the little theater.
[01:37.300 --> 01:39.920]  That just makes logistics a little bit easier for everybody.
[01:40.580 --> 01:47.980]  Also, with closing ceremonies, which is a fun event, we just sort of thank everyone, summarize how things went.
[01:48.520 --> 01:49.740]  Afterwards, we have a band.
[01:49.900 --> 01:50.860]  They're going to start out there.
[01:50.980 --> 02:00.180]  They're going to march down to the outside area, and you're welcome to hang out, enjoy some salsa music after the closing ceremony is done.
[02:00.180 --> 02:05.000]  And after that, or maybe even during that, we do need a lot of help with packing up.
[02:05.140 --> 02:08.490]  A lot of relatively heavy labor, but unskilled labor.
[02:08.680 --> 02:13.320]  So you're welcome and encouraged to stick around if you have time after closing ceremonies.
[02:13.320 --> 02:14.500]  Help out with the packing up.
[02:14.600 --> 02:21.680]  That also includes packing up some outbound rental trucks tomorrow morning, starting at around 9 in the loading dock in D'Angelo.
[02:21.820 --> 02:23.340]  So we do put out a call for volunteers.
[02:23.340 --> 02:26.190]  You'll hear this a few other times during the day.
[02:26.180 --> 02:26.840]  During the day.
[02:27.050 --> 02:29.050]  Don't forget to stay hydrated.
[02:29.560 --> 02:31.990]  It's still a hot day, hot, sticky day out.
[02:32.100 --> 02:34.400]  It's pretty nice, air-conditioned inside.
[02:34.860 --> 02:40.880]  And I don't need to remind you as much anymore to be excellent to one another, because we have all been excellent to one another, haven't we?
[02:41.140 --> 02:42.380]  And that's been, yeah, thank you.
[02:42.440 --> 02:43.460]  And that's been really excellent.
[02:43.660 --> 02:45.780]  So, and we'll talk about that more in closing ceremonies.
[02:45.940 --> 02:55.640]  So if you're thinking, oh, I got a bus to catch, I got a plane to catch, I got to wash my hair this afternoon, maybe stick around, because it's going to be fun as we get towards the end of our three-day event here.
[02:55.640 --> 03:01.120]  So up next is Dan Calacci, and we're going to hear all about how to bargain with a black box.
[03:01.280 --> 03:01.880]  Please welcome Dan.
[03:08.190 --> 03:08.490]  Thanks.
[03:08.750 --> 03:08.970]  Hi.
[03:09.830 --> 03:10.570]  I'm Dan.
[03:10.830 --> 03:12.910]  I'm a PhD student at MIT.
[03:14.290 --> 03:22.030]  And today, I really want to talk to you about why I think the future of the labor movement needs hackers.
[03:22.030 --> 03:37.710]  So I'm going to talk about one project that I've worked on with a group of platform workers, worker organizers, who sort of led basically a project to audit an algorithmic pay change in a delivery platform.
[03:37.750 --> 03:49.810]  And then a bunch of other little pieces of things that are moving around in the labor movement related to building technology, making technology accessible, and increasing digital literacy for workers around the globe.
[03:50.610 --> 03:58.810]  And I'm going to try and convince you that working on this stuff, if you have the technical skills, is worthwhile and needed.
[04:00.310 --> 04:04.950]  So this is sort of the thesis, the argument here, that the future of the labor movement needs hackers.
[04:05.230 --> 04:11.710]  I'm going to quickly just acknowledge the colleagues of mine that work on these projects with me.
[04:11.710 --> 04:15.230]  Johnny Penn, Christina Kochloff, and Nathan Freitas.
[04:15.770 --> 04:23.390]  We all sort of work on different related labor and technology pieces together in tandem and volunteer in other capacities.
[04:23.850 --> 04:29.890]  If you're interested in this stuff, I highly recommend you go check out their Twitter profiles, websites, etc.
[04:31.930 --> 04:34.290]  So, on to the convincing.
[04:34.290 --> 04:44.490]  So, to start, I'm going to give you a motivating example of why building technology for the labor movement is crucial now and will be increasingly crucial in the future.
[04:45.010 --> 04:50.810]  So, in the summer of 2020, SHIPS, which is Target's delivery platform.
[04:50.810 --> 04:53.050]  It's like Instacart, but just for Target.
[04:53.390 --> 04:56.290]  They changed their payment algorithm for workers.
[04:56.550 --> 05:02.210]  They went from a really clear pay scheme to basically a black box algorithm.
[05:02.210 --> 05:04.710]  They used to pay 7.5% of an order total.
[05:04.850 --> 05:12.530]  So, if you paid 100 bucks for a bunch of items from Target, your delivery worker would get $7.50 plus a $5 base fee.
[05:12.730 --> 05:17.350]  This was good for workers because they could see exactly what they were going to make for a given job.
[05:17.750 --> 05:22.070]  The contract workers, they get on certain jobs, they get delivered them by an algorithmic system.
[05:22.530 --> 05:26.970]  One that nudges them, pushes them in different directions, controls them in some capacities.
[05:28.330 --> 05:34.450]  And this is a piece of clarity in an otherwise very murky system for these workers.
[05:35.150 --> 05:40.930]  But in the spring of 2020, SHIPS decided to move to a black box algorithm.
[05:42.610 --> 05:55.010]  This algorithm, as it started to roll out to different workers, who randomly essentially got the new version of the software on their phones, found that their pay was dipping.
[05:56.190 --> 05:57.410]  Workers started protesting.
[05:57.670 --> 05:59.550]  They started sharing their stories online.
[05:59.850 --> 06:09.690]  So, online Facebook groups and organizing forums were full of people just sharing screenshots of pay that, you know, here's an order from three months ago where I made $50.
[06:10.290 --> 06:13.090]  But here's, you know, a really similar order today where I made $30.
[06:13.470 --> 06:15.990]  This is a big pay cut for a lot of workers.
[06:16.110 --> 06:19.170]  And these workers are low wage folks doing a lot of very critical work.
[06:19.170 --> 06:20.490]  And this was in the middle of the pandemic.
[06:20.810 --> 06:32.410]  If you remember the whole narrative around essential workers, around people delivering these goods to folks who are staying at home, trying to stay safe and healthy, this was a huge affront.
[06:34.850 --> 06:40.710]  So, the original pay structure, like I said, was 7.5% of that order total plus $5.
[06:41.150 --> 06:47.870]  And this new algorithm, people started basically going online and speculating about what it was doing, how it was affecting people.
[06:48.850 --> 06:51.690]  You would get a change and you would think it would work one way.
[06:51.830 --> 06:55.930]  You would speculate, oh, maybe they're trying to give me an hourly wage based on my distance or miles.
[06:55.930 --> 07:00.150]  And you get a new shop and it would completely, you know, change your hypothesis.
[07:00.750 --> 07:03.890]  So, I want to introduce you next to Willie.
[07:04.210 --> 07:05.230]  This is Willie Solis.
[07:05.370 --> 07:07.090]  Willie Solis is a worker organizer.
[07:07.290 --> 07:08.330]  He's a gig worker.
[07:09.010 --> 07:16.430]  And during that spring and summer of 2020, Willie started seeing these posts on forums online, on Facebook groups.
[07:18.170 --> 07:24.210]  And he started collecting these screenshots, the ones I mentioned before, that show worker pay.
[07:25.090 --> 07:31.430]  And he started asking people to send them to him and they started pouring in.
[07:31.730 --> 07:38.070]  So, these screenshots are screenshots of people's record of their pay for specific shops over time.
[07:38.150 --> 07:41.910]  So, they include, you know, how much someone paid for their order.
[07:42.050 --> 07:42.950]  It includes the tip amount.
[07:43.130 --> 07:45.130]  It includes the amount of time that it took to deliver the order.
[07:45.350 --> 07:47.030]  It includes the date it happened.
[07:47.310 --> 07:48.290]  All this information.
[07:48.290 --> 07:58.710]  And so, the plan was, or I guess, Willie had this idea that he could collect these screenshots and look at how people's pay changed over time.
[07:59.150 --> 08:03.310]  But once he started getting the screenshots, it became really overwhelming.
[08:03.670 --> 08:06.010]  He would take them and put them into a spreadsheet.
[08:06.390 --> 08:09.950]  And also, once he got it into the spreadsheet, he didn't really know what to do.
[08:09.950 --> 08:11.330]  He's not trained using data.
[08:11.490 --> 08:13.670]  He doesn't know how to build tools.
[08:13.810 --> 08:14.350]  Oh, I'm sorry.
[08:14.530 --> 08:16.810]  I think I'm screwing up my slides.
[08:18.950 --> 08:29.890]  He didn't know how to build tools to look through this information in ways that could really help him and other workers make sense of what was going on.
[08:31.850 --> 08:34.350]  And that's when I got connected with Willie.
[08:35.770 --> 08:39.230]  I got connected with Willie through an advocacy organization called Coworker.
[08:40.050 --> 08:40.970]  You can look them up.
[08:41.050 --> 08:41.370]  They're great.
[08:41.450 --> 08:45.970]  If you have any interest in organizing your workplace, they're a good place to start for resources.
[08:46.750 --> 08:49.210]  This is 2020, so we're meeting virtually over Zoom.
[08:49.430 --> 08:53.070]  Willie with his phone in his car and me with my computer in my office.
[08:53.910 --> 08:57.530]  And we talked for a couple of weeks.
[08:57.530 --> 09:11.090]  And basically as soon as I saw this system of those screenshots plus this back end, I knew that we could do something way more effective than someone manually coding a bunch of screenshots into a spreadsheet.
[09:12.210 --> 09:16.810]  So what we did was we made a texting bot.
[09:17.890 --> 09:26.630]  This texting bot would take screenshots from workers and would parse out people's pay, all the details that I mentioned before from the screenshot.
[09:27.070 --> 09:30.110]  And at the first pass, upload it to a Google Sheet.
[09:30.430 --> 09:38.910]  We would do a Google Sheet because it would allow organizers like Willie to look at workers' data and be able to sort of track the process as we got submissions in.
[09:39.310 --> 09:46.890]  And I want to sort of make note as I go along of these different really common tools that are useful for these projects.
[09:46.890 --> 09:54.490]  Because there's a lot of organizers who want to be able to use technology to expand their organizing efforts, collect data like this.
[09:54.830 --> 10:01.110]  But using like a database for an organizer who's untrained is an impossible ask.
[10:01.110 --> 10:13.790]  It's people are not, you know, trained as computer scientists or as data scientists in when you're driving delivery work, or you're doing low wage like contract labor.
[10:14.050 --> 10:15.250]  It just doesn't exist.
[10:15.490 --> 10:24.950]  Even within large organizations like groups that, you know, do worker advocacy, that digital literacy and maturity isn't really there at the moment.
[10:25.290 --> 10:26.210]  It's not there yet.
[10:26.210 --> 10:42.510]  And this is a big asymmetry between the workers who need information to start proving things about their work, and employers who are increasingly using data to manage workers and make important workplace decisions for them.
[10:43.810 --> 10:46.850]  So we started collecting these screenshots using this tool.
[10:47.030 --> 10:48.610]  We developed it over a couple of months.
[10:49.230 --> 10:50.770]  There's an academic paper.
[10:50.910 --> 10:53.830]  We have a little documentary video that details the process.
[10:53.830 --> 10:57.990]  It's like a very co-designed piece of work.
[10:58.090 --> 11:00.870]  Because this was something that, you know, Willy started.
[11:01.030 --> 11:01.690]  Workers started.
[11:01.830 --> 11:02.950]  They started asking the questions.
[11:03.090 --> 11:03.850]  They created the method.
[11:04.190 --> 11:10.630]  Really all I did as a technologist was walk in and make it go faster and at scale.
[11:10.870 --> 11:12.490]  But the process was the same.
[11:12.870 --> 11:13.970]  The goals were the same.
[11:14.650 --> 11:23.250]  And as we were doing that, we started to be able to counter a narrative that Shipt had pushed about the new algorithm.
[11:23.250 --> 11:28.050]  They had repeatedly claimed, you know, they see workers online talking about it.
[11:28.150 --> 11:31.990]  They repeatedly claimed that the new algorithm was more fair than the previous one.
[11:32.370 --> 11:39.590]  And their sort of accounting of how the algorithm worked is that it would pay you for your effort.
[11:40.050 --> 11:42.370]  Now what effort means was never defined.
[11:44.210 --> 11:46.550]  And people obviously got upset.
[11:47.850 --> 11:58.630]  But as we created, as we started to collect data, people started to feel a little more empowered about speaking up and speaking back against that narrative from Shipt.
[11:59.170 --> 12:05.730]  It got to the point where we collected hundreds, thousands of screenshots from hundreds of people over the course of just a few weeks.
[12:05.730 --> 12:07.930]  Now I don't know if anyone here has ever done like a...
[12:07.930 --> 12:09.610]  For me, I'm a researcher.
[12:09.890 --> 12:11.570]  So this is sort of like a field experiment.
[12:11.950 --> 12:17.730]  I don't know if anyone's done a thing where they try and get several hundred people to participate in a like...
[12:18.810 --> 12:22.990]  Even a collective action sort of thing or something they think is good for them as benefiting them.
[12:23.130 --> 12:24.030]  But it's really hard.
[12:24.590 --> 12:29.370]  We had 200 people sign up and start using this bot in like the first two weeks.
[12:29.370 --> 12:30.390]  It was amazing.
[12:30.770 --> 12:39.210]  And a big part of this is that the bot would actually give people feedback about their pay to them specifically.
[12:39.450 --> 12:40.390]  Which I'll talk about in a second.
[12:41.350 --> 12:42.010]  These are...
[12:42.010 --> 12:46.930]  This is sort of the upshot, I would say, of the study of the...
[12:46.930 --> 12:52.090]  So we did a blog post after I analyzed a bunch of the data and talked with workers about what they wanted to see.
[12:52.710 --> 12:54.870]  And these are two different things.
[12:55.050 --> 12:58.330]  The plot on the left, each square represents a worker.
[12:58.910 --> 13:06.810]  And it's colored by how much they were getting paid, their pay difference between the first algorithm, which we call V1.
[13:07.130 --> 13:10.650]  That was like the colloquial name for it in the organizing circles.
[13:10.790 --> 13:13.150]  And V2, the new black box algorithm.
[13:13.530 --> 13:18.610]  So everyone in the green is actually making more on average with the new algorithm.
[13:18.830 --> 13:24.310]  There are a group of people who are making more money per shop than they were before.
[13:24.310 --> 13:35.550]  But then, if you go down, you see all these folks in the yellow, in the orange, in the red, who effectively got between, you know, a small to a very big pay cut.
[13:35.550 --> 13:42.710]  Now, this is effectively, you know, for some folks, it was 20%, 30% of their take-home pay suddenly like that is gone.
[13:43.350 --> 13:45.010]  They have to work more hours.
[13:45.230 --> 13:48.070]  They are less certain about how much they're going to take in per week.
[13:48.250 --> 13:50.250]  They don't know if the algorithm is going to change.
[13:51.350 --> 13:55.330]  And so that was the working reality for folks who experienced this difference.
[13:55.950 --> 14:03.110]  And interestingly, the plot on the right, one of the things that I tried to look at was, well, you know, did they just get a bad hand?
[14:03.270 --> 14:05.810]  Like, did they get bad orders during the time we're looking at this?
[14:06.110 --> 14:09.330]  We did get, when people submitted these screenshots, it was like six months of their pay.
[14:09.570 --> 14:12.530]  So we could see really over time how much that they were making.
[14:12.530 --> 14:28.010]  And on the right, over a course of a few weeks after the algorithm, the new payment system was instituted, we looked at the difference between folks in the green line who were making more under the new system and folks in the red line who were making less.
[14:28.610 --> 14:33.450]  And what this shows is that people who are making less, making less consistently.
[14:33.770 --> 14:38.490]  There's something about those workers that the algorithm doesn't like.
[14:38.490 --> 14:41.470]  The way they shop, the amount of time they spend, etc.
[14:41.870 --> 14:46.830]  This is something that we've tried to investigate further, but we don't have enough data.
[14:47.250 --> 14:51.750]  And SHIP did not respond, I think, really adequately to these concerns.
[14:51.890 --> 14:55.490]  We created a report, they responded saying that the data was not representative.
[14:56.530 --> 14:58.230]  Which, you know, to be fair, it's true.
[14:58.490 --> 15:03.410]  We got a bunch of people who were already interested in organizing, wanted to donate their screenshots.
[15:03.970 --> 15:07.190]  But that's still, that was 40% of workers who got a pay cut.
[15:07.190 --> 15:09.310]  And they didn't acknowledge that.
[15:10.270 --> 15:19.390]  But this was, beyond trying to get a response from SHIPs, really useful in drumming up support within the organizing community for folks like Willie.
[15:19.610 --> 15:24.590]  Because they could show that their narrative, their experience, was actually happening.
[15:27.650 --> 15:32.970]  And basically, after the report dropped, another series of protests happened.
[15:33.870 --> 15:36.730]  These were happening, you know, not because of the report.
[15:36.910 --> 15:38.230]  I want to just make that really clear.
[15:38.610 --> 15:44.950]  But I don't know if you can see, there's one person in there with a sign that says, MIT says, SHIP keeps cutting pay.
[15:47.090 --> 15:56.350]  So we got to basically support all of these movements around different parts of the country with this data-driven report.
[15:56.350 --> 15:59.850]  And the tool that we built, like, to remind you, it's a texting bot.
[16:00.290 --> 16:02.550]  It, like, does some OCR and some screenshots.
[16:02.850 --> 16:07.010]  I hacked together the script in an afternoon and sort of tweaked it a little bit over time.
[16:07.670 --> 16:08.810]  That's not that bad.
[16:09.070 --> 16:12.070]  I'm sure a lot of people in this room could build that in a weekend.
[16:13.210 --> 16:14.830]  But it had this kind of impact.
[16:15.790 --> 16:17.450]  So I want you to sit with that a little bit.
[16:19.510 --> 16:27.070]  So beyond platform workers, which are this big, interesting, important category.
[16:27.250 --> 16:30.270]  A lot of my work focuses on them because they're managed by algorithms.
[16:31.770 --> 16:36.590]  And in the future, data will be more and more important to sort of proving a counter-narrative to those algorithms.
[16:37.330 --> 16:42.570]  Data and technological tools can also really help just everyday workers, normal folks.
[16:42.710 --> 16:47.250]  Folks in office jobs, folks who do manual labor, folks who work in warehouses.
[16:48.050 --> 16:55.470]  And so I wanted to share with you some of our work building tools for more general organizing.
[16:55.810 --> 16:59.490]  And, again, one of the things I just want to stress is how simple a lot of these tools are.
[16:59.750 --> 17:01.050]  They're not super complicated.
[17:02.130 --> 17:03.390]  Folks here can work on these.
[17:03.390 --> 17:08.330]  So I'm going to motivate this also with another big problem.
[17:09.370 --> 17:10.290]  Wage theft.
[17:11.030 --> 17:17.230]  So wage theft is an enormous problem globally and especially in the United States.
[17:17.570 --> 17:21.610]  Wage theft happens when you work some hours and you don't get paid for them.
[17:21.730 --> 17:24.610]  Or you're supposed to get benefits, but you don't.
[17:25.050 --> 17:33.370]  Some folks argue that many platform workers are experiencing wage theft because they are misclassified as contract workers.
[17:33.370 --> 17:34.210]  Instead of employees.
[17:34.710 --> 17:37.310]  And so if they're misclassified, they don't get benefits they're entitled to.
[17:37.770 --> 17:39.010]  That's one form of wage theft.
[17:39.510 --> 17:42.770]  Another might be you work overtime, but don't get paid for it.
[17:43.290 --> 17:48.250]  Or you are given sick days as part of your contract, but you're not allowed to take them.
[17:49.010 --> 18:04.670]  And all of that value, all of that money and benefits that people lose every year, basically costs low wage workers almost $3,000 a year in costs.
[18:04.910 --> 18:11.590]  And when you scale that up to every low wage worker, lots and lots of people in the United States, this is an enormous problem.
[18:11.590 --> 18:20.010]  One study looking at a few major cities in the United States estimates that about 70% of workers have experienced some kind of wage theft.
[18:21.430 --> 18:32.410]  And this study, the 70%, doesn't even count not getting a break, those overtime violations, and other problems that we do consider wage theft, which are close to impossible to measure.
[18:32.410 --> 18:32.990]  Right?
[18:33.650 --> 18:36.430]  It's really hard to report, it's hard to get information about it.
[18:37.230 --> 18:44.150]  But data-driven tools, where people contribute information or share it and then aggregate it, can help with problems like this.
[18:44.790 --> 18:51.270]  I'm going to put it, talk about wage theft sort of in scope a little bit more, to give you a sense of the scale of the problem.
[18:53.010 --> 18:58.930]  This circle represents the amount of money that gets stolen in all robberies every year in the United States.
[18:58.930 --> 19:05.090]  This is people, you know, robbing homes, burglary, violent robbery on the street.
[19:06.510 --> 19:10.010]  I don't remember actually if this includes white collar crime, probably not.
[19:10.870 --> 19:12.930]  But $438 million, that's a lot.
[19:13.290 --> 19:15.930]  I mean, you can see how tiny the circle is, we're about to get bigger.
[19:17.950 --> 19:22.890]  This is how much money is stolen from workers through just hour violations.
[19:22.890 --> 19:30.370]  So you work hours, you get your paycheck, it pays you for less, and you're left sort of wondering what to do.
[19:31.950 --> 19:33.870]  That's, I mean, this is not to scale.
[19:34.690 --> 19:41.170]  That's $15 billion, with a B, dollars every year that gets stolen from workers.
[19:41.390 --> 19:46.330]  And that just goes from low wage workers to management to owners of companies.
[19:46.330 --> 19:49.050]  It's just this massive wealth transfer.
[19:50.650 --> 19:59.950]  And people estimate that all wage theft, which includes those overtime violations, other benefits, stuff like that, might be up to around $30 billion a year.
[20:00.450 --> 20:01.390]  This is huge.
[20:01.530 --> 20:06.350]  This is like a really important problem, and this is a big issue if you think about income inequality.
[20:07.730 --> 20:25.070]  And one thing that we're trying to do is build tools that let people record their work, share that information with researchers, advocates, and organizers, and use it to prove working time, prove working conditions, and support workers' narratives.
[20:26.490 --> 20:30.730]  There's a whole host of, like, different projects around this kind of work.
[20:30.970 --> 20:37.850]  But the thing I want to really talk about here is this tool that we're building called WeClock.
[20:39.030 --> 20:42.430]  So WeClock, which is this app on the left, you can download it.
[20:42.510 --> 20:43.430]  It's sort of, like, weird.
[20:43.570 --> 20:45.690]  It's like a research app, basically, at the moment.
[20:45.850 --> 20:48.270]  But it basically operates like Strava for work.
[20:48.570 --> 20:52.390]  It basically takes all the sensors on your phone, location, movement.
[20:52.390 --> 20:58.190]  It connects to, like, HealthKit and Samsung Health, environmental noise, so it uses your microphone.
[20:58.330 --> 21:03.670]  If your phone has one, it'll measure temperature, because for workers in places like warehouses, that can be really important.
[21:05.390 --> 21:11.070]  And all of these different pieces of just, like, raw data get stored privately on people's devices.
[21:11.430 --> 21:20.070]  So this is not an extractive tool where I'm a researcher or a company, and we're storing a big data set of people's information at work.
[21:20.070 --> 21:25.430]  This is, you use this, and you get the data, and you get to choose what to do with it.
[21:27.010 --> 21:30.210]  Which is crucial, I think, in the worker movement.
[21:30.670 --> 21:34.770]  Workers are already sort of subject to enough extraction.
[21:35.650 --> 21:40.590]  All of these tools, we try to design them so that they're empowering rather than extractive.
[21:42.450 --> 21:54.790]  And the idea here is to automatically track basically anything you could think of that would be relevant to proving your experience as a worker.
[21:57.050 --> 22:02.510]  And one of the issues here is that even with you build an app, it creates a data set.
[22:03.270 --> 22:04.870]  Unionists are still not technologists.
[22:05.130 --> 22:10.670]  So we have to do work to build up a digital literacy and a capacity inside of the movement.
[22:10.670 --> 22:27.270]  So another thing that we do and that folks could participate in is doing things like running workshops, sharing how to use tools like WeClock or even just like even spreadsheet tools, honestly, with some worker organizers to track time, track hours, and create a working record.
[22:28.070 --> 22:36.550]  So we've done this over the past year where we run digital workshops over Zoom with trade unionists from around the globe.
[22:37.090 --> 22:55.190]  And some of these folks, these are folks who don't have laptops, who zoom in on a phone with inside of like a barn with a bunch of chickens, and they're trying to figure out how to use their phone to make it easier for their workers that they represent to prove that they worked some hours.
[22:57.530 --> 23:05.890]  And that sort of practice of teaching people how to use data, how to think about it, is also a crucial, important part of the movement.
[23:05.890 --> 23:17.210]  I want to suggest that you can think about it as sharing information with folks, making it more accessible to people who don't have access to training about how to use these tools.
[23:19.390 --> 23:22.370]  But we know that building tools like that are just the start.
[23:22.610 --> 23:24.570]  So we're trying to create something.
[23:25.510 --> 23:31.810]  It would be lovely if some of you became involved in the future as we build it out, called the Workers Tech Hub.
[23:32.450 --> 23:46.450]  This aims to be a kind of community of practice for technologists, public interest folks, workers, worker organizers, advocates, researchers, to help build a technology practice around labor organizing.
[23:46.450 --> 23:49.910]  And collecting workplace data, but from the worker perspective.
[23:52.410 --> 23:57.330]  These are sort of the main things that we think it entails for the most part, right?
[23:57.490 --> 24:00.130]  So engaging workers, engaging labor unions.
[24:01.290 --> 24:03.910]  Not all workers, especially in the United States, are in labor unions.
[24:04.090 --> 24:07.710]  A lot of them are in informal or other kinds of worker groups, like worker centers.
[24:07.710 --> 24:12.850]  But they can also help collect data and aggregate it.
[24:13.330 --> 24:18.190]  But when data gets aggregated, understanding data stewardship and governance.
[24:18.390 --> 24:19.530]  So who gets control of the data?
[24:19.670 --> 24:20.750]  Who gets access to it?
[24:20.830 --> 24:22.030]  How do you mediate that access?
[24:22.150 --> 24:26.410]  And how do workers maintain consent over who can actually see their working time?
[24:26.530 --> 24:27.890]  Because some of that stuff is very sensitive.
[24:29.010 --> 24:30.090]  That needs work.
[24:31.030 --> 24:34.150]  Like doing stuff like we did in that first project I mentioned.
[24:34.590 --> 24:38.630]  Where you audit new algorithmic management technologies.
[24:38.930 --> 24:43.790]  Is a huge interesting future and now area of work.
[24:44.450 --> 24:49.030]  Doing these kinds of community audits of management tech.
[24:50.370 --> 24:55.270]  These kinds of technologies like platforms that give contract workers a job.
[24:55.470 --> 24:56.930]  Based on some criteria.
[24:58.110 --> 24:58.930]  Are going...
[24:58.930 --> 25:02.130]  It's going to spread to other areas of work in the near future.
[25:03.110 --> 25:08.370]  Already there's plenty of new platforms that do things like try to gigify nursing.
[25:09.170 --> 25:11.910]  Try and do this often for freelancing work.
[25:12.170 --> 25:12.970]  Creative work.
[25:13.450 --> 25:13.850]  Writing.
[25:14.570 --> 25:18.670]  And as people's jobs are more and more algorithmically mediated.
[25:18.670 --> 25:28.350]  We're going to need access to data and tools to try and counter, test, and inquire about how those systems work.
[25:33.280 --> 25:37.540]  So these are two examples of tools we built as part of...
[25:37.540 --> 25:38.540]  So WeClock's the app.
[25:39.820 --> 25:42.020]  But you generate all this data.
[25:43.120 --> 25:44.160]  And it's...
[25:44.160 --> 25:46.440]  You know, as far as data goes, it's clean.
[25:46.600 --> 25:47.460]  It's generated from a phone.
[25:48.180 --> 25:49.360]  It sort of looks like that.
[25:49.360 --> 25:50.880]  That's basically the raw data on the right.
[25:52.280 --> 25:52.680]  And...
[25:52.680 --> 25:58.280]  But, like, imagine a construction worker opening a Google Sheet to that.
[26:00.520 --> 26:01.840]  Hard to know what to do.
[26:01.980 --> 26:04.400]  So we're building a bunch of different tools.
[26:04.540 --> 26:05.980]  And these are, like, very hacked together.
[26:06.140 --> 26:07.060]  All of it is open source.
[26:07.060 --> 26:12.320]  The one on the left is a little app that I'm building that lets folks...
[26:12.320 --> 26:15.900]  That lets folks basically upload their...
[26:15.900 --> 26:18.880]  A zip file of data they export from WeClock.
[26:20.800 --> 26:23.140]  And this focuses on location data.
[26:23.340 --> 26:35.260]  So for workers who work in a particular location, an office, a work site, often location is all you need to prove that you are working during a certain time of day.
[26:36.500 --> 26:41.800]  One example of this is construction workers who work on a work site, often get paid informally.
[26:41.940 --> 26:43.260]  They get paid in, like, cash, right?
[26:43.400 --> 26:52.000]  And so sometimes some departments of labor have told me that construction workers will get paid in cash months and months and months, then not get paid.
[26:52.620 --> 26:56.620]  How do you prove that you stopped getting paid or that you were even working there?
[26:56.860 --> 26:57.920]  You don't have contracts.
[26:58.020 --> 27:01.880]  You got paid using this informal mechanism for months and months.
[27:02.040 --> 27:04.600]  The employer says, oh, I never even hired that person.
[27:04.960 --> 27:05.760]  What do you do?
[27:06.660 --> 27:08.400]  One way is to...
[27:09.200 --> 27:17.040]  This uses some clustering and location analysis or just geofencing to figure out when people were at work.
[27:17.540 --> 27:21.340]  And signs that information so that you know it wasn't tampered with.
[27:21.900 --> 27:26.240]  And we're building in tools to allow people to do things like take photos of that pay.
[27:26.500 --> 27:34.200]  And to prove using something called a tool called proof mode out of the Guardian project that that wasn't tampered with.
[27:34.400 --> 27:41.480]  So you can take that to a department of labor, make a wage claim, and suddenly the worker has better records than the employer.
[27:44.020 --> 27:44.680]  Done deal.
[27:45.720 --> 27:46.900]  That's the idea, anyway.
[27:49.380 --> 27:55.860]  In this tool on the right, I wanted to do an interactive thing, but it's sort of difficult, to be honest, from the stage.
[27:56.040 --> 27:58.920]  So I'm just going to show you a picture of one of the screens.
[28:00.420 --> 28:04.520]  It's just basically another example of trying to look at pay.
[28:06.180 --> 28:12.240]  So we can do things like look at people's movement.
[28:12.580 --> 28:15.180]  So front of house, back of house workers, for example.
[28:15.600 --> 28:22.080]  In New York, you get paid differently if you're waiting tables or if you're in the back doing scheduling or reservations.
[28:22.420 --> 28:24.260]  Or in the front doing scheduling or reservations.
[28:24.760 --> 28:27.860]  But that work looks different from your phone's perspective.
[28:28.600 --> 28:30.240]  If you're waiting tables, you're moving around really quickly.
[28:30.600 --> 28:32.320]  You're bouncing up and down.
[28:32.620 --> 28:33.440]  You're on your feet.
[28:33.740 --> 28:35.440]  But if you're doing back of house work, you're sitting.
[28:36.220 --> 28:37.700]  Those two things get paid differently.
[28:38.180 --> 28:42.220]  And proving those hours is one form, one big form of wage theft in the service industry.
[28:43.140 --> 28:50.580]  So one thing you can do is really easily, it's rough, but you can just set thresholds for some of that sensor data.
[28:51.760 --> 28:56.100]  Apple Health, Samsung Health make it really easy to measure things like standing time, walking time, et cetera.
[28:56.840 --> 29:06.160]  And workers can use this tool to mark off when they were standing, sitting, have the associated data, and then calculate what their wages should have been.
[29:06.680 --> 29:07.920]  And then look at the difference.
[29:09.560 --> 29:12.900]  This is early, but, you know, you can see it's just hacked together.
[29:13.100 --> 29:14.960]  It's a Google Sheet, like, just basically a template.
[29:15.340 --> 29:17.620]  But that's actually the best version of this for folks.
[29:17.860 --> 29:30.020]  Because anybody who is not trained as a computer scientist or, you know, is a hacker can sit on a spreadsheet at least and, like, poke around and make plots with, like, minimal training.
[29:30.660 --> 29:33.280]  So making these tools accessible is really important, again.
[29:34.500 --> 29:37.600]  As part of our workshops, you know, we focused on wage theft.
[29:37.600 --> 29:41.160]  But some of the workers had really interesting campaigns.
[29:41.540 --> 29:44.100]  They had different ideas of how to use the same tools.
[29:44.840 --> 29:48.800]  So another thing that I thought was really nice is that I sort of...
[29:48.800 --> 29:50.560]  WeClock is this really general purpose thing.
[29:50.700 --> 29:52.900]  If you download it, you'll see that it has no direction.
[29:52.900 --> 29:57.780]  It just sort of, like, collects a lot of information from your phone for you.
[29:58.520 --> 30:05.440]  And I've always wanted it to be more pointed, more specific, maybe geared specifically towards wage theft and calculating those lost wages.
[30:05.440 --> 30:13.040]  But the flexibility of the tool has actually been really beneficial because workers themselves come up with these creative ways of using it.
[30:13.400 --> 30:20.100]  So these are three campaigns that came out of our workshop that workers created and started using WeClock to collect data on.
[30:20.340 --> 30:26.240]  And then these tools, the spreadsheet, the little web app thing to all through on.
[30:26.240 --> 30:33.740]  The one on the left is probably the one that I thought was most creative and different, that we just wouldn't have thought about it.
[30:34.500 --> 30:40.700]  This campaign was about workers who often had to leave the workplace really late at night.
[30:40.700 --> 30:46.760]  So you're, you know, I think this was in London, and people are getting home late at night.
[30:46.880 --> 30:47.480]  They're taking the train.
[30:47.620 --> 30:48.540]  They're taking public transit.
[30:48.960 --> 30:52.700]  And some folks felt really unsafe during their transit, getting home.
[30:53.020 --> 31:00.720]  They wanted the workplace to, A, kind of, like, see them, validate that, like, they felt unsafe in these particular locations.
[31:00.720 --> 31:04.500]  And then, B, provide them with alternate transit options.
[31:04.660 --> 31:06.380]  Like a bus, a shuttle, something.
[31:07.080 --> 31:29.120]  But to actually, like, convince management that folks were feeling unsafe and where and what commutes they should really focus on, this organizer took the WeClock app, gave it to workers, and had them record their location, which happens by default, and then answer little survey questions about,
[31:29.480 --> 31:32.940]  you know, when did you feel most feel unsafe during your commute?
[31:33.340 --> 31:39.480]  What location did you, like, do you wish that you had maybe been inside or something like this?
[31:40.400 --> 31:43.000]  And this is a really creative use of the tool.
[31:43.280 --> 31:48.200]  This campaign is still ongoing, so I don't know exactly how it's rolling forward now.
[31:49.840 --> 32:03.320]  But one of the design lessons here is when you are building tools for workers, make them extensible, make them things that folks can play with and mold to their own needs because workers know sort of the problems they have.
[32:03.680 --> 32:11.000]  This is a core problem in designing technology, right, is you have a group of folks, you're the technologist, you're the designer, you have an idea of what their problem is.
[32:11.320 --> 32:25.980]  And in the worker context, it's so diverse, it's so varied, but they have such a domain expertise in the issues that they're facing that working with them, not just like designing for them, but working with them and having them be a part of that process is really crucial.
[32:25.980 --> 32:42.180]  So if you do engage in any of this work, build any technology with workers, try to have them just at the table, like we did with the black box thing, like we're doing here where folks design their campaign, and we'll use this to design the tech in the future to make this workable.
[32:45.420 --> 32:46.660]  I'm almost finished.
[32:46.660 --> 32:51.120]  I want to talk about some general obstacles to these projects.
[32:51.960 --> 32:55.180]  Because it's not... I keep saying how easy it is.
[32:56.180 --> 32:58.160]  But it's not super easy.
[32:59.860 --> 33:11.920]  There's a few different pieces here that are moving, that are part of like larger conversations around technology as well, that are big hurdles for technologists and workers alike.
[33:12.880 --> 33:15.820]  So I'll use an example to talk about some of them.
[33:16.460 --> 33:20.300]  This is something called the module platform.
[33:21.340 --> 33:26.260]  It's basically a belt that you wear.
[33:26.480 --> 33:29.040]  And the belt buckle is packed with sensors.
[33:29.360 --> 33:30.840]  So it tracks your location.
[33:31.260 --> 33:33.980]  It tracks your... actually like blue...
[33:33.980 --> 33:41.400]  I think it uses Bluetooth signal strength to communicate with other belts to figure out how close you are to other warehouse workers.
[33:41.660 --> 33:44.920]  I think that's how they do the contract tracing that's labeled up there.
[33:46.040 --> 33:47.700]  And it does temperature detection.
[33:48.100 --> 33:54.640]  So like, you know, a lot of stuff actually that our app does, but on behalf of the employer.
[33:56.260 --> 34:05.140]  And their killer feature is really measuring when people bend down in an unsafe way in a warehouse to reduce injuries.
[34:05.140 --> 34:13.720]  So this plot over here on the right is their analysis of injury rates after implementing using this technology.
[34:15.620 --> 34:17.060]  But there's a problem here.
[34:17.240 --> 34:22.300]  Just like there was a problem with Schipt's narrative about the new algorithm.
[34:22.680 --> 34:23.940]  How do you verify that?
[34:24.540 --> 34:26.540]  And how do you know how they're measuring injuries?
[34:26.880 --> 34:32.700]  Or what the sensor data they're using looks like to actually measure when people are bending and picking stuff up?
[34:33.260 --> 34:39.360]  If you're a worker and you're organized and you have a team of folks who is critical of that, what do you do?
[34:41.520 --> 34:44.940]  Right now, you have really no request.
[34:45.280 --> 34:48.520]  You can try and collect that data yourself using tools like WeClock.
[34:48.680 --> 34:49.820]  That's part of why we're building them.
[34:50.280 --> 35:05.280]  But if you wanted access to that data, even if you're in California, under the CCPA, the new privacy law that allows you to submit access requests to companies like the GDPR for folks who are familiar with privacy law, you can't get access to that information because you're a worker.
[35:05.800 --> 35:09.300]  Everything you generate is the property of your employer for the most part.
[35:10.780 --> 35:23.960]  A lot of that information is covered under trade secret, IP law, and this directly butts heads against our notions of privacy and against what's useful and really needed in the worker movement in the near future.
[35:24.660 --> 35:27.420]  We don't have... there are no good answers to this right now.
[35:28.340 --> 35:30.800]  Not in the EU, really, and not here.
[35:31.020 --> 35:38.520]  These are like really big open questions about how legal institutions should treat data for folks who are working.
[35:40.200 --> 35:41.560]  I actually have one of these belts.
[35:41.580 --> 35:42.080]  I bought one.
[35:43.980 --> 35:47.680]  If anyone's interested in poking around and figuring out how they do what they do.
[35:49.400 --> 35:52.060]  But there's some motions towards this.
[35:52.060 --> 35:53.060]  This is a lot of text.
[35:53.180 --> 35:53.380]  I'm sorry.
[35:53.520 --> 35:57.120]  I just cribbed these slides from a talk I just gave last week or a couple weeks ago.
[35:58.420 --> 36:03.640]  So there's one really cool bill that if you're interested in this, you should pay attention to.
[36:03.760 --> 36:04.740]  It's called AB 701.
[36:05.520 --> 36:12.520]  So AB 701 was a bill pushed by the Warehouse Workers Resource Collective in California.
[36:13.460 --> 36:21.080]  And like I said, CCPA, California Privacy Law, doesn't allow workers to request data about their work.
[36:21.080 --> 36:27.940]  But folks who work in Amazon warehouses, like those poor folks who are wearing those module belts, get instrumented constantly.
[36:28.660 --> 36:30.420]  They have production quotas.
[36:30.500 --> 36:33.280]  They have, you know, actual rates that they need to hit.
[36:33.400 --> 36:36.600]  Otherwise, they get fired, often automatically.
[36:39.080 --> 36:45.220]  And these quotas have led to workplace injuries, skipped breaks, you know, talk about wage theft.
[36:47.040 --> 36:54.740]  And workers wanted to have some evidence about how these quotas were changing over time.
[36:54.920 --> 36:57.920]  And how these quotas were affecting their working conditions.
[36:58.920 --> 37:04.340]  So this bill allows workers, and a third party, that's the really important part.
[37:04.540 --> 37:10.520]  Because you as a worker might, like, you know, be experiencing this but may not know how to engage in this data conversation.
[37:10.720 --> 37:13.100]  You need a third party to be able to submit requests on your behalf.
[37:14.240 --> 37:20.720]  Lots of privacy, just suggested bills, regulations, don't really accomplish this.
[37:20.820 --> 37:24.040]  They don't have... they make it difficult for third parties to submit requests.
[37:24.460 --> 37:31.380]  Or they don't think about this collective idea of aggregating data that's about you, that's collected by companies or by your workplace.
[37:31.880 --> 37:32.960]  But this bill does.
[37:33.140 --> 37:37.020]  This bill allows third parties to send those requests on behalf of warehouse workers.
[37:37.020 --> 37:42.300]  And when they do, Amazon, or it's, you know, any warehouse bill, but it's really about Amazon.
[37:43.400 --> 37:54.860]  Any warehouse has to respond within 90 days with basically a bunch of data about your production quota, your work times, any sensor data used, I believe, to calculate them.
[37:55.840 --> 38:06.140]  These kinds of bills are the things that will push this work forward without us having to build all of this technology that just collects the same data these employers do already.
[38:07.140 --> 38:09.240]  Now, there are benefits to doing it yourself, right?
[38:09.320 --> 38:19.420]  Because you can do things like in that campaign where the worker asks workers questions, the organizer asks workers questions about how safe do you feel at a certain period of time.
[38:19.840 --> 38:22.500]  The employer is not going to have that information, right?
[38:24.040 --> 38:35.380]  But this makes it a lot easier for folks to be able to interrogate, question, and do what I call worker inquiry into algorithms and into these systems.
[38:37.860 --> 38:43.760]  In the GDPR, there's some stuff that people are exploring how to apply to these contexts.
[38:44.240 --> 38:50.260]  So, like, specific states basically have to come up with rules for the workplace context.
[38:50.520 --> 38:54.620]  It's kind of happening now, especially as folks look at more AI regulation.
[38:54.620 --> 39:01.320]  A lot of this is going to get lumped into artificial intelligence regulation instead of data protection, which is interesting.
[39:02.360 --> 39:13.160]  There are these things called data trusts and data cooperatives and data unions, which are all different ideas around aggregating and governing data collectively.
[39:14.020 --> 39:21.040]  Consumer reports, the consumer union, the group behind them, basically is building stuff around consumer data unions.
[39:22.280 --> 39:24.600]  But these can also be used for workers, right?
[39:24.660 --> 39:25.780]  It's got union in the name, doesn't it?
[39:25.880 --> 39:35.280]  So, the idea of a union or a worker collective or worker group aggregating that information, but having it under, you know, tightly bound under legal conditions.
[39:35.280 --> 39:36.540]  Workers are the beneficiary.
[39:36.680 --> 39:38.420]  They can't sell the data without their permission.
[39:38.600 --> 39:41.720]  They can't use it to make something that's going to harm the workers.
[39:42.860 --> 39:47.000]  These are helpful legal tools to solve really thorny issues.
[39:47.200 --> 39:50.680]  Some people try to solve these using technical approaches.
[39:50.680 --> 39:56.540]  So, you know, put data on the blockchain, audit when people access it, give people voting rights.
[39:57.480 --> 40:01.440]  But we have way simpler legal solutions to some of those problems.
[40:04.180 --> 40:14.080]  The main thing that I... the main message here with data rights is that workers need to be involved in data stewardship.
[40:14.380 --> 40:20.240]  They need access to the information that they produce while they're at work in order to use it for whatever they want, but for organizing.
[40:21.120 --> 40:27.520]  And we need more technologists to be building tools that allow workers to collect and leverage that data.
[40:28.500 --> 40:30.760]  I'm going to end on this.
[40:31.060 --> 40:48.360]  This is one of my favorite images from sort of deep dives into labor history and the history of scientific management, which is sort of like the old school intellectual grandfather of algorithmic management that we now see on platforms.
[40:49.260 --> 40:52.100]  This is called a stereo cyclochronograph.
[40:52.420 --> 40:54.200]  This was taken in 1913.
[40:55.140 --> 40:55.780]  1913.
[40:56.140 --> 41:02.580]  This is a long exposure photograph of someone's hands with light bulbs on the end of their fingers.
[41:03.060 --> 41:05.720]  This is taken by Frank and Lillian Gilbreth.
[41:06.400 --> 41:10.120]  They're sort of like the grandparents of scientific management.
[41:10.440 --> 41:21.800]  They developed all these insane technologies to measure how people did certain jobs in factories and in textile mills and I think in hospitals.
[41:22.660 --> 41:28.660]  And the goal of these photographs... these are some of the first stereo photographs that were ever taken.
[41:28.660 --> 41:33.240]  Stereoscopic photographs was for measuring worker behavior on job.
[41:34.600 --> 41:38.820]  The goal of a stereo cyclochronograph like this... I think that's how you say it...
[41:38.820 --> 41:44.680]  is to help decipher those precise movements associated with faster or safer performance on some task.
[41:44.920 --> 41:47.780]  Nobody knows what this is actually illustrating, what task it is.
[41:48.440 --> 41:50.440]  But I think the image is kind of cool enough by itself.
[41:52.020 --> 42:02.200]  And the thing I want to sort of bring to folks' attention is that back then, getting your photograph taken was a kind of event.
[42:02.540 --> 42:04.000]  Not an everyday occurrence.
[42:04.980 --> 42:08.040]  This was... think of this as like early data collection.
[42:08.720 --> 42:15.720]  Your motion getting collected by a camera, by two folks, you know, on the other side of the table from you.
[42:15.720 --> 42:16.880]  It's personal.
[42:17.340 --> 42:17.500]  It's personal.
[42:17.500 --> 42:22.240]  And a lot of these workers, when they were studied in this way, they got a picture back.
[42:22.540 --> 42:25.760]  They got a copy of their data.
[42:26.360 --> 42:31.820]  And it was something that showed them a different perspective on their own work and their own life.
[42:33.880 --> 42:41.720]  That process was not quite as extractive as the ones we have today, where people wear a belt and you never see the result.
[42:41.720 --> 42:51.640]  And the company posts a report that talks about reducing your lumbar injury, but all you feel is a vibration on your back every time you lean down to pick up a box.
[42:55.460 --> 42:59.320]  While this was used to measure their work, but this is still a photograph of someone.
[42:59.500 --> 43:03.020]  And the process by which it was generated speaks to that.
[43:03.020 --> 43:09.360]  It was personal and it presented folks with this alternate idea.
[43:10.340 --> 43:18.020]  As we build tools, as we build algorithm management tools, tools to organize, tools for workers to take hold of their own data.
[43:18.580 --> 43:20.860]  This is something I want folks to think about.
[43:21.160 --> 43:28.460]  Is how do you still make those in a way that still keeps the person there and keeps it reflective and personal.
[43:29.440 --> 43:30.560]  I think that's all I got.
[43:31.220 --> 43:32.160]  Yeah, thank you.
[43:39.660 --> 43:40.980]  If folks have questions.
[43:41.560 --> 43:42.020]  Sorry.
[43:42.320 --> 43:45.020]  There's a microphone under the spotlight.
[43:45.820 --> 43:47.160]  You can go up and ask.
[43:47.280 --> 43:50.100]  You can also yell one out if you are so inclined.
[43:51.180 --> 43:51.940]  That's cool too.
[43:57.600 --> 43:58.980]  Hi, I love your talk.
[43:59.180 --> 44:08.860]  And one of the things that you didn't exactly speak to about why people should, you know, have some access or control of their own data is just the concept of dignity.
[44:08.860 --> 44:18.240]  And I think that also like personal dignity that like you shouldn't be turned into an actuator for a machine that, you know, you're not like a claw arm.
[44:18.400 --> 44:21.760]  Like if you don't want to have sensors put on you, you shouldn't have to do that.
[44:21.760 --> 44:33.680]  But, um, my question about the, uh, was about what you showed earliest on was shipped and, um, to sort of take a devil's advocate role there.
[44:33.680 --> 44:36.880]  It said, you highlighted that 41% of people were making less.
[44:37.360 --> 44:38.960]  What percent were making more?
[44:39.420 --> 44:51.340]  And is this the type of thing that anytime you shift these things, it's going to, it's like by the nature of changing it, some behaviors are going to now be less rewarded.
[44:51.340 --> 44:53.720]  Some behaviors are not going to be more rewarded.
[44:54.040 --> 44:59.240]  And what's like, what would you say to the idea that, you know, they're still paying people.
[44:59.620 --> 45:01.340]  A lot of people were winners here.
[45:01.900 --> 45:03.980]  Um, you know, what's the problem?
[45:04.160 --> 45:04.720]  Thank you.
[45:05.260 --> 45:05.700]  Yeah.
[45:05.860 --> 45:06.040]  Thanks.
[45:06.160 --> 45:06.800]  Thanks for the questions.
[45:07.220 --> 45:10.380]  Um, I want to just respond to the dignity comment.
[45:10.520 --> 45:19.480]  One of the things that they said was, uh, that another argument for worker, uh, about around worker data and ownership is just dignity around data collection and agency.
[45:19.480 --> 45:22.960]  Um, I, I wholeheartedly agree with that sentiment.
[45:23.160 --> 45:36.670]  The idea that being, uh, instrumented is, especially without my consent in a, uh, coercive relationship with an employer, um, is, uh, a little dehumanizing.
[45:37.280 --> 45:42.860]  Um, and it removes agency from folks, especially when they don't have access to the information that gets, that they're producing.
[45:43.740 --> 45:56.540]  Um, and to your second question, uh, so you asked essentially, uh, essentially, uh, in that system, it seems like in the, when the algorithm changed, it seems like there were some folks who won out and, and some folks who lost out.
[45:56.740 --> 45:59.980]  And that might just be a natural dynamic of changing the system.
[46:00.240 --> 46:09.120]  And so the algorithm, uh, the algorithm at pay cut is maybe less interesting or less important under that scheme, under that, that lens.
[46:09.380 --> 46:12.560]  Um, and we might expect that with any kind of pay change to workers.
[46:12.560 --> 46:14.260]  You get that right, kinda?
[46:14.460 --> 46:16.660]  Yeah, the algorithm should be a cut or a pay raise.
[46:16.760 --> 46:18.920]  Yeah, pay cut or a pay raise under the algorithm.
[46:19.300 --> 46:20.660]  So, yeah, right?
[46:20.900 --> 46:24.580]  So, so the algorithm, I, I've, I've, I've speculated on what it does.
[46:24.720 --> 46:38.760]  I'm pretty sure that what it did is it actually estimated the amount of time that it would take a shopper to go out and grab the items in a shopping list and deliver them by using, like, the actual spatial characteristics of each target store and where they stocked items.
[46:38.760 --> 46:45.820]  Um, and so there's some folks who don't walk as fast as their, you know, average walking speed concepts.
[46:46.000 --> 46:49.400]  So they probably made poor predictions for those folks.
[46:49.720 --> 46:58.080]  Um, for example, I'm, I'm sure that there's other things that went into, uh, them not being able to hit the right pay amount for some folks.
[46:58.200 --> 47:02.900]  But it's important, I, I think two things sort of change the framing of that, because I've been asked that quite a bit.
[47:03.160 --> 47:18.900]  Um, the first is that the old pay scheme, the V1 pay scheme, although sort of like the baseline that I'm looking at, um, it's not like, uh, they, their, this new pay change was not necessarily, they, they argued that their new pay change was raising pay across the board and was fair.
[47:19.380 --> 47:21.500]  Um, A, that's certainly not the case.
[47:21.860 --> 47:25.840]  Uh, and certainly not, like, certainly not fairer.
[47:26.100 --> 47:30.560]  Um, we, I didn't see that same distribution of, like, winners and losers under the old pay scheme.
[47:30.560 --> 47:31.940]  It just, that didn't exist.
[47:32.240 --> 47:34.440]  This new one created this divide.
[47:34.860 --> 47:46.980]  Um, and that difference in pay, for reasons that I, I don't know, but are, seem arbitrary to workers, um, strikes me as particularly unfair.
[47:47.540 --> 47:53.640]  Um, and then the, the, the second response is that, that's still 40% of workers who got a pay cut.
[47:53.640 --> 47:54.660]  That was unannounced.
[47:55.060 --> 48:03.740]  Um, even if this is not really about, is it ethical or is it good to institute a new payment algorithm and blah, blah, blah.
[48:04.020 --> 48:07.920]  Like, they did not disclose that folks were gonna get a pay cut.
[48:08.080 --> 48:09.420]  They didn't talk about it.
[48:09.500 --> 48:10.380]  They didn't share that information.
[48:10.540 --> 48:11.240]  Workers didn't know.
[48:11.740 --> 48:15.240]  This is the only way that workers found out, is by aggregating that information.
[48:15.520 --> 48:19.420]  And it's the only way that some workers knew that they were getting a pay cut, because they could just think about it.
[48:19.420 --> 48:22.280]  But all the app shows you is your list of orders over time.
[48:22.540 --> 48:26.080]  It doesn't show you comparisons of your aggregate pay by algorithm.
[48:27.240 --> 48:30.780]  And so that, you still need to do the work to show that transparency.
[48:31.300 --> 48:33.020]  Um, yeah.
[48:33.420 --> 48:34.360]  It's a good question.
[48:34.900 --> 48:35.340]  Thanks.
[48:37.980 --> 48:38.420]  Hi.
[48:38.700 --> 48:39.600]  Thanks for your talk.
[48:40.080 --> 48:45.940]  Um, as a technologist who's interested in getting involved in this stuff, what's the best way to start getting involved?
[48:46.640 --> 48:47.280]  Uh, yeah.
[48:47.460 --> 48:47.820]  Good question.
[48:47.820 --> 48:49.580]  I guess I didn't say that.
[48:49.940 --> 48:58.580]  Um, I would say that the best way to start getting involved is to talk to, uh, local worker groups.
[48:58.880 --> 49:00.760]  So, go to work.
[49:00.820 --> 49:03.060]  The way that I started working on this, I'll share my experience.
[49:03.200 --> 49:03.420]  How about that?
[49:03.500 --> 49:05.160]  I think that's maybe, maybe probably the most helpful.
[49:05.680 --> 49:07.420]  Um, I touched on that briefly earlier.
[49:07.420 --> 49:19.920]  Is I got in touch through a friend who is kind of an advocacy nerd, um, with folks within Shipt who were doing this project with Willie.
[49:20.180 --> 49:23.820]  And I was lucky enough to meet with Willie and we connected.
[49:23.820 --> 49:32.760]  And I have the MIT name behind me, so it's easier, it's easy for me to sort of like get folks to, uh, pay attention to me and give me opportunities.
[49:32.940 --> 49:34.160]  Honestly, it's a privilege.
[49:34.540 --> 49:42.180]  Um, but at its most basic is I talked with him and a bunch of other workers about what they were experiencing.
[49:42.180 --> 49:50.140]  And I learned about what they were trying to do and I saw an opportunity where my skills could sort of match and help.
[49:50.860 --> 50:07.280]  Um, there's been plenty of other projects, other situations where workers share what they're trying to do to, you know, fight for a more equal wage or fight wage theft that don't, you know, sync with what I can offer as a researcher or advocate.
[50:07.280 --> 50:12.520]  Um, uh, so those cases you leave alone.
[50:13.080 --> 50:21.380]  But, um, you can go to worker centers and the people who organize them and folks who are organizing platform workers or other folks in your local area.
[50:21.380 --> 50:29.580]  Just with like some searching, shoot them an email or go to their office and just like say like, hey, I have 10 hours a week and I have built things that look like this.
[50:29.760 --> 50:31.800]  Is there anything, you know, can we think of something?
[50:32.040 --> 50:34.420]  Um, and what are you looking at right now and how can I help?
[50:35.040 --> 50:39.460]  Uh, it's kind of like a trite thing, just go talk to people, but it, it's honestly true.
[50:40.380 --> 50:48.320]  Um, the 15 billion dollars, um, that you mentioned in your presentation, is that mostly from, um, private companies?
[50:48.580 --> 50:54.100]  I mean, public companies like Amazon, Target, or, you know, is it like privately held as well?
[50:55.480 --> 50:56.300]  I don't know.
[50:57.860 --> 50:58.580]  I'm not sure.
[50:58.920 --> 50:59.040]  Yeah.
[50:59.680 --> 51:01.180]  Where did you get that number?
[51:01.460 --> 51:02.320]  I'm just curious.
[51:02.320 --> 51:04.380]  Um, I have a source I can share with you.
[51:04.680 --> 51:08.600]  Um, uh, I don't remember, I don't remember off the top of my head.
[51:08.840 --> 51:11.080]  Um, I have it in the original slides.
[51:11.440 --> 51:17.560]  The, the, I imagine it's across public and privately held companies.
[51:17.780 --> 51:20.980]  I, I can't imagine that that number was specific to one or the other.
[51:21.380 --> 51:26.940]  Um, although I, and I don't know if there's even a difference in rates of wage theft between the two.
[51:26.940 --> 51:27.980]  I'm not sure.
[51:28.440 --> 51:33.440]  I imagine probably it's higher in publicly traded companies if only because of the clear profit motive for folks.
[51:33.840 --> 51:35.860]  But, um, yeah, I'm not sure.
[51:36.040 --> 51:36.600]  Interesting question.
[51:39.970 --> 51:42.190]  I can't see, uh, really well with the lights.
[51:42.390 --> 51:45.210]  So, if people are raising their hands or something, you might not be.
[51:45.310 --> 51:46.950]  But if you are, I can't really tell.
[51:46.950 --> 51:48.330]  So, if you could go to the mic.
[51:51.730 --> 51:52.250]  No, okay.
[51:56.980 --> 51:57.480]  Okay, cool.
[51:58.480 --> 51:58.940]  Thank you.
[51:59.060 --> 51:59.660]  Thanks for your time.