/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

Survey of Mechanical Turk workers finds some feel they don't have recourse for underpayment, technical errors, or general abuse from temporary bosses

Dhruv Mehrotra / Gizmodo :

Gizmodo Dhruv Mehrotra

Context & Ripple Effects

This survey lands between two markers in the Mechanical Turk coverage arc: a UN survey of 3,500 microtask workers in 75 countries that put average pay at $6.54/hr in the US and $3.31/hr worldwide, and the COVID-era boom in low-paid tagging and survey work that followed a few months later. The through-line is that crowd work scaled on the promise of accessibility while the grievance side — underpayment, technical errors, abusive requesters — stayed unresolved.

The new finding adds the missing piece to that picture: it isn't just that pay is low, it's that workers report no mechanism to contest it. That matters for data buyers too, because the same platform has already been flagged for bot-driven low-quality survey data and, later, for workers quietly delegating tasks to LLMs.

First-order effects

  • Mechanical Turk workers absorbing underpayment, technical failures, and requester abuse have no formal dispute path, meaning losses from rejected work or platform errors land entirely on them.
  • Requesters face no reputational or procedural check on how they treat temporary workers, so the cost of abusive task design stays at zero for the party imposing it.

Second-order effects

  • With no recourse and piece-rate pay, the rational worker response is to cut per-task effort or automate it — consistent with the estimated 33%–46% of workers using LLMs on a summarization task and the earlier bot-contamination findings — which degrades the data quality requesters are paying for.
  • The COVID-era influx of jobless workers into tagging and survey microtasks enlarges the pool of people exposed to the same unmediated requester dynamics, amplifying the volume of disputes the platform has no process to handle.

Third-order effects

  • If crowd work keeps growing as a labor absorber during downturns, the absence of recourse structures points toward external intervention — labor standards or platform-accountability rules applied to microtask marketplaces — because the platforms' own incentives don't generate them internally.
  • The pattern pushes data buyers toward verification-based pricing: paying per accepted, quality-checked output rather than per submitted task, shifting quality-control costs from workers to requesters.

The trend: Microtask platforms are scaling into a default safety-net labor market faster than they are building worker recourse, leaving pay disputes and data-quality problems to compound together.

Discussion

  • @andylocal Andy Newman on x
    Gizmodo asked Mechanical Turk workers for the creepiest tasks they were asked to do. Answers: look at possible child porn, rate surgery videos, look at other people's SS / credit card info, draw their own genitals, and take many pix of their feet: https://gizmodo.com/... https://…
  • @lmatsakis Louise Matsakis on x
    Really smart to report on Mturk by posting surveys on the platform itself https://gizmodo.com/...
  • @anne_kreft Anne-Kathrin Kreft on x
    We talk about ethics with respect to interviewing & fieldwork, much less so when it comes to surveys. As social scientists, we need to think more carefully about how we flag sensitive content, how we avoid exploitation & how we handle problematic platforms https://gizmodo.com/...
  • @aedwardslevy Ariel Edwards-Levy on x
    (Incidentally, this is also the service Emerson polls are getting their online respondents through) https://twitter.com/...
  • @demandprogress @demandprogress on x
    @amazon Because of @Amazon's monopoly power, its Mechanical Turk platform workers are subject to exploitation and extremely low wages, with some people making only $5 a day. https://gizmodo.com/...
  • @kashhill Kashmir Hill on x
    To find out the strangest/worst things that Amazon Mechanical Turk workers have seen, @dmehro hired 1,1100 workers to tell him: https://gizmodo.com/...
  • @tsimonite Tom Simonite on x
    “12 percent of respondents claimed that the worst or strangest experience on Mechanical Turk was...feeling emotionally traumatized by an academic survey.” https://gizmodo.com/...
  • @dmehro Dhruv Mehrotra on x
    A few months ago we ran a pair of surveys on Mechanical Turk asking more than 1000 gig-workers about their worst, best, and strangest experiences/tasks (HITS) on the platform. The responses were surprising. https://gizmodo.com/...
  • @adrjeffries Adrianne Jeffries on x
    “Someone paid me like 50 cents to recall the most painful memory of my life,” recalled another respondent, “it fucked me up for the entire day.” https://gizmodo.com/...