Shipt’s algorithm squeezed gig workers, who fought back • Hacker News Board

Re: Shipt’s algorithm squeezed gig workers, who fought back

#91

Post by pxx » Thu, Jul 04, 2024, 5:34 PM UTC

oefrha wrote: post #26

or_am_i wrote: post #3

I am not defending Shipt and there is no doubt gig workers are in a very vulnerable position. However, the data analysis results as presented in the article do not support the article's main point. "40% are getting paid at least 10% less" is not unnatural to expect whenever pay is redistributed, especially since some 30+% are getting at least 10% more. Imagine a _hypothetical_ situation where Shipt is 100% on point a…

> 40% are getting paid at least 10% less That statement is wrong either way. Looking at the graph, ~22% got a >10% cut while ~36% got a >10% raise. Overall ~43% got a cut while ~57% got a raise. And if there’s any doubt, later on they dropped the “at least 10% less” qualifier for the 40% figure in text: > But we felt that it was important to shine a light on those 40 percent of workers who had gotten an unannounced p…

it seems clear to me from first principles and from experience [0] that the result from voluntary compensation sharing skews low compared to the actual population. it's the people who want to complain about low wages who participate in such schemes.

that's not to say sharing wages is bad. this can still provide upward pressure on said wages. but the people who participate in the beginning are likely from the lower end of the distribution.

[0] websites like levels.fyi seem to consistently skew low when estimating higher percentiles

Re: Shipt’s algorithm squeezed gig workers, who fought back

#92

Post by alwa » Thu, Jul 04, 2024, 5:34 PM UTC

toss1 wrote: post #70

mft_ wrote: post #29

Earlier quoted context omitted.

It's an interesting question, as we have a spectrum from 'little to no transparency' through to 'full transparency' (which is pretty rare), and in the middle sits the usual approach of 'communications-team-led messaged quasi-transparency'. Difficult to know (without more info) where Shipt would have appeared on this spectrum, but given the issue, they're probably somewhere towards the 'insufficient transparency' end.…

YES. If Shipt is actually trying to incentivize better performance, it seems the best way is to be completely transparent about the rewards algorithm. "Short high-value trips are now somewhat de-rated, and trips requiring more effort now have improved rewards, specifically ..." or whatever. This "communications team" approach did everyone a disservice if Shipt mgt were really trying to improve results. OTOH, if the a…

Although as far as that graph reflects the the study’s results, the new distribution looks almost perfectly balanced: even more people experienced a “10%+” bump than a 10%+ reduction post-update.

By what mechanism do you suggest the worker-screwing is happening here?

Re: Shipt’s algorithm squeezed gig workers, who fought back

#93

Post by croemer » Thu, Jul 04, 2024, 5:39 PM UTC

> The system used optical character recognition—the same technology that lets you search for a word in a PDF file

That's not correct, at least for "digitally-born PDFs" that were made on a computer and haven't been scanned. In that case, the PDF can be parsed directly, without OCR, to get text. That's what a tool like PyPDF2 does, for example.

Re: Shipt’s algorithm squeezed gig workers, who fought back

#94

Post by dsr_ » Thu, Jul 04, 2024, 5:51 PM UTC

sweezyjeezy wrote: post #49

dsr_ wrote: post #44

Earlier quoted context omitted.

So... they've turned it from "we provide a good quality service at a fair price" to "over time you will have a chance to pick favorites, but not consistently". That's enshittification for your use of the service, and enshittification for the workers, too.

Not sure I understand your argument - from a customer perspective why is the service worse, and why do you believe that it has BECOME worse?

If it were as good or better than before, nobody would want to pay attention to when to schedule deliveries based on the deliverator. Since this person specifically wants some deliverators over others, that means that the quality of the service is less good. They need to spend time considering scheduling, where they did not before.

I don't worry about which USPS mail carrier delivers my mail -- I know it will be consistent and good enough. I happen to know who my usual carrier is, because I work from home and she likes to say hi to cats if they are in the front window. I also know the face of the usual UPS driver and the usual FedEx driver; they aren't here 6 days a week, but often enough that I recognize them.

In none of those cases do I expect a quality change based on the driver. I expect competence, and I get it so often that the exceptions really stand out.

From the Shipt workers' perspective, they now need to worry about customers discriminating among them rather than just getting the job done.

Re: Shipt’s algorithm squeezed gig workers, who fought back

#95

Post by alwa » Thu, Jul 04, 2024, 5:51 PM UTC

demondemidi wrote: post #82

theptip wrote: post #51

A clear case of adverse selection in the old pricing model. From a game-theoretic perspective in a gig marketplace you don’t want jobs that are strictly better, else sophisticated market participants (workers) will select the best ones leaving chaff - and a worse experience - for the less sophisticated participants. What you are looking for is preference optionality, eg one Uber driver might prefer not to do very lon…

I used to fly into Pittsburg and take an Uber to my grandparents house 40 miles away. For years I could always get an Uber but then after the pandemic my trip was rejected so frequently that I had to start renting a car. So basically no one wanted the gig and Uber never told me my ride was unreasonable. Something changed in the algorithm to benefit drivers, Uber, or both, at the expense of the customer, and that info…

I’m curious: would you have felt better if Uber had just rejected your ride request upfront (“Sorry, we can’t offer rides that far” or something)?

Or is this attitude of “hey, it’s a long shot, but let’s give it a try and see if anybody takes the job” closer to the attitude you’d like to see? If the latter, how would you communicate that to the users?

I had an occasion a long time ago where I needed to request an Uber for a ride similar to what you’re describing. At that time, apparently the driver didn’t find out the route until they’d committed to the ride. The guy swiped to say he’d picked me up, and more or less broke down in tears when he found out where I was needing to go. He lived 40 miles in the other direction, was going off shift, and would be driving the whole 90 miles home without any prayer of a passenger to cover the time or cost. In that case I ended up giving the guy a generous amount of cash to cover the imposition, but I couldn’t bring myself to use Uber for that route in the future.

Until recently, when I had to use Uber for that route again. This time it seemed like they’d gotten much better at accommodating drivers’ preference optionality: the guy who picked me up drove over 110mph all the way to the airport. Apparently when you drive like that, especially in an EV, the more miles the better…

He explained that he could dial into the app that he preferred longer trips and trips between areas that happened to be connected by this lawless highway.

Re: Shipt’s algorithm squeezed gig workers, who fought back

#96

Post by alwa » Thu, Jul 04, 2024, 5:52 PM UTC

croemer wrote: post #93

> The system used optical character recognition—the same technology that lets you search for a word in a PDF file That's not correct, at least for "digitally-born PDFs" that were made on a computer and haven't been scanned. In that case, the PDF can be parsed directly, without OCR, to get text. That's what a tool like PyPDF2 does, for example.

It sounds like they were parsing screenshots that workers submitted by SMS

Re: Shipt’s algorithm squeezed gig workers, who fought back

#97

Post by alwa » Thu, Jul 04, 2024, 6:10 PM UTC

dsr_ wrote: post #94

sweezyjeezy wrote: post #49

Earlier quoted context omitted.

Not sure I understand your argument - from a customer perspective why is the service worse, and why do you believe that it has BECOME worse?

If it were as good or better than before, nobody would want to pay attention to when to schedule deliveries based on the deliverator. Since this person specifically wants some deliverators over others, that means that the quality of the service is less good. They need to spend time considering scheduling, where they did not before. I don't worry about which USPS mail carrier delivers my mail -- I know it will be cons…

This “Shipt,” though, involves an opportunity for some degree of relationship to make a difference, right? Your mail carrier must deliver your package, the package is the package, it’s either delivered or not. Maybe there’s a small margin around the edge where one carrier is nice to the cats and the other isn’t.

These Shipt people, though, have to interpret your preferences and essentially act as your agent as they decide what to pick from the store shelves on your behalf. Sometimes they make decisions that you probably would have made, sometimes less so; sometimes they’re confident that you understand each other, sometimes they’re nervous and want to hassle you about each of 10 different little decision points. When you find somebody I work well with, isn’t it a positive that you get to try to keep that relationship for future transactions? Isn’t this the same dynamic underpinning virtually every in-person service, from your hair cutting human to the tradies who do work on your house to the dry cleaner?

For that matter, doesn’t it create a perverse incentive if worker doesn’t believe that trying to understand my preferences will ever pay off? That it’s a one-off game rather than an iterated series of games, and effort to excel and bring human judgment to bear is wasted because there’s no way to reward it?

Doesn’t the enshittification tend to require as a prerequisite that a platform is successful at alienating service providers from service recipients (and from each other) like that?

Re: Shipt’s algorithm squeezed gig workers, who fought back

#98

Post by NovemberWhiskey » Thu, Jul 04, 2024, 6:11 PM UTC

tossandthrow wrote: post #5

or_am_i wrote: post #3

I am not defending Shipt and there is no doubt gig workers are in a very vulnerable position. However, the data analysis results as presented in the article do not support the article's main point. "40% are getting paid at least 10% less" is not unnatural to expect whenever pay is redistributed, especially since some 30+% are getting at least 10% more. Imagine a _hypothetical_ situation where Shipt is 100% on point a…

You are looking for this in the article > It wasn’t a clear case of wage theft, because 60 percent of workers were making about the same or slightly more under the new scheme. Your statement that "some 30+% are getting at least 10% more" assumes that there is no wage theft - which is not cut in stone.

I mean, it wasn't "a clear case of wage theft" as these weren't wage workers; and in any case it's not like they weren't told ahead of time how much the order would pay.

Re: Shipt’s algorithm squeezed gig workers, who fought back

#99

Post by EVa5I7bHFq9mnYK » Thu, Jul 04, 2024, 6:13 PM UTC

tbrownaw wrote: post #48

astrange wrote: post #46

Earlier quoted context omitted.

Uber doesn't do that because the drivers wouldn't take ones they don't like. Remember, consumers are using platform apps like Uber because they don't trust the drivers on the other side.

> Remember, consumers are using platform apps like Uber because they don't trust the drivers on the other side. Oh, I thought it was because "push button to summon car" is, like, super convenient.

That too, but taxi drivers are indeed among the most notorious for scams.

Re: Shipt’s algorithm squeezed gig workers, who fought back

#100

Post by EVa5I7bHFq9mnYK » Thu, Jul 04, 2024, 6:18 PM UTC

alwa wrote: post #89

If the whole point of the algo change is to correct an unfairness by which a strict fee+cart value approach doesn’t reliably reflect the amount of work somebody’s being asked to do, isn’t this exactly the outcome we expect? That the people who were putting more work in now get more money, while the people who were benefiting from sniffing out the “easy” jobs now make something more in line with everybody else’s compe…

Apart from the workers and the company there is another important actor here - the clients. I think the point of the algo change was to better serve clients with smaller orders.