from Hacker News

Algorithmic Wage Discrimination (2023)

by tacon on 9/11/24, 5:20 PM with 101 comments

  • by advael on 9/11/24, 8:26 PM

    As with price discrimination, I think there are only two possible ways this can go:

    Either we pass some laws that deeply upset some oligarchs and future eras see this one as a "wild west" of data-driven exploitative practices, or we permanently destroy the bargaining power of everyone not big enough to be doing mass-surveillance, which is effectively the end of markets and the rise of a sort of advanced feudalism (which some would argue is quite far along already)

    One obviously seems better to me, but to be fair I am biased, being in the class of people who don't own an enormous corporation

  • by vundercind on 9/11/24, 7:07 PM

    The reaction here has been interesting. As someone with a heavier social-science and politics background than most HNers, I started reading the introduction before there were any comments here and went “oh my god, of course this is a thing, why did it never occur to me both that this would/is happen/happening, and to put it in these terms“, and my mind started chasing down the mechanisms by which it must work. Bombshell is a word that came to mind as soon as it dawned on me what the paper’s getting at.

    Then I came back to this thread and had a similar experience to when I checked on the Internet’s reaction to The Last Jedi after I got out of the theater, and was surprised to find that my experience of it as the only almost-actually-good SW film since the original trilogy was, um, not what most others had taken away.

  • by DeliriousDog on 9/11/24, 6:36 PM

    There was a (now deleted comment) about how there is no proof of wage discrimination for Uber/Lyft drivers, which was posted with no evidence.

    This video (https://www.youtube.com/watch?v=OEXJmNj6SPk) was recently published which shows drivers being offered the same gigs, but different payment amounts. Note that I could not find a published version of the data they collected in this video.

    That is not explicitly proof of wrongdoing, but clearly algorithmic price setting can be demonstrated as not always offering the same payment to the same drivers for the same work. There may be a valid reason to why this is the case, but as the calculation method is closed source, the individuals being offered the wage are unaware of why they would be paid less than their peers.

    This is work that is often considered "low skill" - which should actually make it extremely cut and dry as to why an individual would be paid more or less. Are they making their pickups faster? Are their customers more satisfied? If that's the case, why would they sometimes be offered more money than their peers and sometimes less money?

    Almost all workers here are price takers, and suffer greatly from the information asymmetry present. Companies hiding behind "oh but the algorithm says..." is a poor excuse for inequality.

    Edit: Because discrimination is in the title of the OP, I feel the need to clarify: in no way is the above saying that the video posted is proof of discrimination. Inequality need not be discrimination. When there is inequality without any measurable source, we need to be skeptical of the reason. Maybe one driver has better customer feedback, therefore they get offered a higher wage. There are many logical explanations for the result, but Uber/Lyft do not seem to engage with the discussion. This should raise red flags. That does not conclude that they are discriminating against anyone, and that would be a poor conclusion to draw without a true investigation.

  • by mrguyorama on 9/11/24, 8:48 PM

    If this outcome is a surprise to you, you have not been paying attention. Managers and Owners dream of reducing wages every single night. Every penny they aren't paying you is a penny they keep, and they, for some reason, think they are rightfully entitled to every penny, while you are a brat for feeling entitled to even a few pennies.

    More importantly, if your boss doesn't think like that, you can expect his boss to replace him with a manager that DOES.

    This downward force on labor prices is basically the entire point of the gig economy, and even youtubers whose only formal training is in how to write a high school essay figured this out.

    For this reason, there is institutional pressure to turn anything they can into gig economies. If you don't have empathy for the uber drivers who sleep in their car to be able to make as many rides as they can and still somehow end up below minimum wage, don't worry, they'll come for you soon. And when they've made the gig economy literally the only gig in town, it doesn't matter how savvy you think you are at "negotiation", you will get fucked too.

    Go read up on your labor rights history. None of what we have is a default, and the "haves" HATE it.

  • by foolofat00k on 9/11/24, 9:12 PM

    The thing that people consistently miss with these types of conversations is that any increase in the sophistication of the tech that exists to measure the world gives a relative benefit to corporations over individuals.

    This is because those organizations almost always have more resources to dedicate towards making effective use of that information than do individuals.

    Very often you as an individual are up against a team of PhDs and engineers whose job it is to enable the corporation to beat you, and the more data they have, the more likely they are to win.

    In this respect, there is basically no tech that does not benefit corporations more strongly than it benefits individuals. This is one of the reasons that regulation is important.

  • by Manuel_D on 9/11/24, 10:18 PM

    "Wage discrimination" as described in the article is not an unfair or illegal practice:

    > “Algorithmic wage discrimination” refers to a practice in which individual workers are paid different hourly wages—calculated with ever-changing formulas using granular data on location, individual behavior, demand, supply, or other factors—for broadly similar work.

    Surge pricing is a form of "algorithmic wage discrimination". Driving for Uber on a Friday night will net you more than a Monday at noon. Likewise, the fact that wages are higher in expensive metros is another form of "algorithmic wage discrimination." People hired during periods of a labor shortage may have a higher wage than a co-worker hired during an economic downturn. I am skeptical many would point at these scenarios as unfair - but all of these fall under the category of "wage discrimination" as discussed in the article.

    I also think this article unnecessarily injects racial messaging and leads readers to think that the algorithms are discriminating on the basis of protected class. That is not alleged in this article.

  • by fmajid on 9/12/24, 9:15 AM

    Not so coincidentally, the EU is mandating salary transparency:

    https://fikku.com/111920

  • by josefritzishere on 9/11/24, 9:38 PM

    This is very apropos after seeing the Pave post.
  • by _gmax0 on 9/11/24, 9:23 PM

    Off-topic, does anyone know whether the source code for this UI is open source?
  • by Wolfenstein98k on 9/11/24, 11:19 PM

    "Subordinated racial minority". "Informational capitalism". "how should the law intervene" (no 'if'). "For workers, these practices produce unsettling moral expectations about work and remuneration".

    Big hmm. As always, truly objective metrics - such as pay based on output (rather than time spent) is viewed as harmful to minorities.

    Is it, though? Or is it actually more fair?

  • by itake on 9/11/24, 6:38 PM

    Typically, workers are paid a flat rate, businesses buffer the worker's income generating activities, such that the business owners pocket high income activities (like serving drinks and being tipped) with low income activities (like washing the dishes at the end of the night). Any money left over after wages, is pocketed by the business owner, to be used on a day with low income day (think Friday night vs Sunday night at a restaurant).

    Dynamic pricing allows workers more exposure to the value they generate, thus enabling them to capture higher pay.

  • by WalterBright on 9/11/24, 6:26 PM

    I.e. workers who produce more get paid more, those who produce less get paid less. It's always been that way, though it is usually difficult to measure accurately.