{"id":"fb533972-113a-4831-b579-7db1c0d53887","arxiv_id":"2412.02973","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Policy experts and gig workers share priorities around fair pay, discrimination, and safety that a worker-centered data-sharing system could serve.","lead":"Researchers interviewed 11 U.S. policy experts and held design workshops with 14 gig workers to learn what a worker-controlled data-sharing system should do. Both groups identified fair pay, discrimination, and safety as top shared priorities; workers additionally wanted systems for swapping practical tips.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The shared-priorities finding may partly reflect prepopulated prompts: workers ranked researcher-supplied sticky notes and policy experts reacted to a prepared data-type list, so the alignment claim needs a prompted-vs-unprompted reanalysis.","rationale":"The reader's weakest-assumption focuses on external validity: the 11 policy experts and 14 workers constitute convenience samples, so the shared initiative list may not generalize. That is a legitimate concern, and the paper's Section 6 acknowledges breadth limitations. However, the more acute threat to the central claim is internal: the workshops and interviews were structured around researcher-supplied data types and prepopulated sticky notes, so the reported convergence between workers and policy experts could reflect a shared response to the same prompt rather than shared underlying priorities. This concern is testable from the existing transcripts because the protocol distinguishes earlier open-ended discussion from later prompted ranking. If the unprompted analysis confirms the three shared initiatives, the finding gains strength; if not, the design recommendations in Section 5 rest on a weaker evidentiary base. I do not think this pushes the verdict beyond CONDITIONAL: the paper is transparent about its methods, the qualitative analysis is internally consistent, and the concern is a verification condition rather than a demonstrated error. The reader already assigned CONDITIONAL, so the verdict should remain unchanged, with the condition made more specific: the authors should report whether the shared priorities survived unprompted elicitation.","tokens_in":30642,"tokens_out":4805,"duration_ms":52866,"concrete_test":"Re-code the existing interview and workshop transcripts to separate unprompted mentions (occurring before any prepopulated note or data-type list is introduced) from prompted endorsements of those items. For each of the three shared initiatives—pay/unpaid time, discrimination, and safety—check whether a majority of participants in each stakeholder group raised the topic unprompted. If these topics are not raised unprompted by a majority of both groups, the alignment claim is substantially weaker and the design recommendations should be reframed as researcher-framed desiderata rather than independently expressed stakeholder priorities.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that pay/unpaid time, discrimination, and safety are shared stakeholder priorities that should drive data-sharing design—depends on those priorities being elicited rather than supplied. In §3.2.2, worker workshops asked participants to rank sticky notes 'alongside prepopulated notes,' and the initiatives were 'informed by the previous interviews' and introduced at the end of each session. In §3.2.1, policy experts were handed a researcher-prepared list of potential data types by occupation and asked which would be helpful. If the final themes in Section 4 combine prompted rankings with unprompted discussion, the reported two-stakeholder alignment may be an artifact of the materials the research team brought into the room, rather than independent convergence. The paper does not report a separate analysis of unprompted versus prompted mentions, so the reader cannot determine whether the shared initiatives would have surfaced without the researchers' framing. This is an internal-validity threat to the main finding, and it is more direct than sample representativeness: even a perfectly representative sample would not fix the problem if the protocol steered participants toward the reported categories.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports a qualitative, two-stage study aimed at identifying the policy initiatives and data needs that a worker-centered data-sharing system for gig workers should support. The authors interviewed 11 U.S. policy domain experts (policymakers, policy implementers, advocates, and a policy researcher) and conducted co-design workshops with 14 U.S.-based gig workers across four domains (freelancing, food delivery, rideshare, petsitting). Thematic analysis of interviews and workshops produced 1593 codes, 118 first-level themes, 17 second-level themes, and four third-level themes. The central finding is that both stakeholder groups share priorities around pay practices and unpaid work time, discrimination and equity, and safety; the groups differ in that policy experts emphasize stress (especially for caregiving workers) while workers emphasize sharing qualitative work strategies and experiences. Based on these findings, the paper offers design recommendations for data-sharing systems, including features for data collection, education, and policy collaboration, and reflects on challenges of data integrity, invisible labor, and privacy.","tokens_in":30857,"tokens_out":3679,"duration_ms":37215,"significance":"If the central alignment finding is valid, this is a timely and useful contribution to the HCI/CSCW literature on gig work and data activism. The paper is transparent about its qualitative methodology, uses multiple coders, provides rich participant quotes, and grounds its design recommendations in participant preferences. The study also extends prior work by explicitly engaging both policy experts and workers to identify shared versus stakeholder-specific policy priorities. However, the significance of the contribution depends heavily on whether the reported shared priorities were genuinely elicited from participants rather than supplied by the researchers' materials. The paper currently does not provide the evidence needed to rule out that possibility, which is why a revision is needed.","major_comments":[{"comment":"The central claim that 'both stakeholder groups sought data to understand pay practices and (unpaid) work time and more attention toward the issues of discrimination and safety' may be partially an artifact of the research protocol. In §3.2.2, worker workshops asked participants to rank their own sticky notes 'alongside prepopulated notes,' and in §3.2.1, policy experts were handed a researcher-prepared list of potential data types divided by occupation and asked which would be useful. The paper does not report any analysis separating unprompted from prompted mentions. Because the alignment between stakeholders is the main empirical contribution, the absence of such a prompted-versus-unprompted analysis is load-bearing: it is possible that the reported shared initiatives (pay, discrimination, safety) were already present in the materials the research team brought into the room. I recommend that the authors reanalyze their data to distinguish between themes that arose spontaneously and those that appeared only after exposure to the researchers' prepopulated lists/notes, or at minimum report a sensitivity analysis that clearly describes which findings rely on prompted responses and which on unprompted ones. The limitation discussion in Section 6 does not address this issue.","section":"§3.2.2 and §3.2.1"},{"comment":"The policy expert group is heterogeneous, comprising city-level policymakers, county and federal implementers, advocacy group representatives, and a policy researcher. The paper treats these as a single 'policy domain expert' group and does not report analyses separated by role. The authors note in Section 6 that they did not separate by governing level, but they also do not separate by role type. Given that advocacy groups and academic researchers may have different policy priorities than elected officials or agency implementers, the reported 'policy expert' perspective could obscure meaningful variation. This does not invalidate the study, but it limits the specificity of the claims about what 'policymakers' want. The authors should either provide a subgroup breakdown or add a clearer caveat about the heterogeneity of this group.","section":"§3.1.1 and §4.2"}],"minor_comments":[{"comment":"Table 2 lists three petsitting participants (W1, W2, W3), but the text in §4.1.1 quotes a participant 'W5' regarding ethnic names and discrimination; please reconcile the participant IDs or correct the reference.","section":"Table 2 and §4.1.1"},{"comment":"The text 'our prior experiencedoes not remedy' is missing a space between 'experience' and 'does'; please fix the typographical error.","section":"§3.3"},{"comment":"The coding process is described, but no inter-rater reliability metrics are reported. For a qualitative study, agreement metrics are not always required, but reporting them (or a rationale for not using them) would strengthen confidence in the 1593-code thematic structure.","section":"§3.4"},{"comment":"The paper mentions that the full protocol and study materials are in supplementary materials, but the arXiv version does not include them. Please provide the supplementary materials or describe the workshop protocol in enough detail for replication.","section":"§3.2.2"}],"recommendation":"major_revision","confidential_remarks":"To the editor: the paper's main empirical claim—that workers and policy experts independently converge on pay, discrimination, and safety as shared priorities—is plausible but not yet adequately supported given the potential contaminating influence of the researchers' prepopulated materials. This is a fixable issue: the authors could reanalyze their transcripts to separate prompted from unprompted codes, or they could transparently report which findings emerged from unprompted discussion. I do not see evidence of circularity in the derivational sense, and the qualitative analysis itself is careful. The paper is within scope for the journal and the topic is timely. I recommend major revision rather than rejection, because the issue is addressable within the manuscript's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is a genuinely useful empirical map: it puts 11 U.S. policy experts and 14 gig workers across four domains side by side on what a worker-centered data-sharing system should support, and it shows where the two groups converge and diverge. The shared priorities — pay/unpaid time, discrimination, safety — are plausible, well-illustrated with quotes, and consistent with prior work. That direct cross-stakeholder comparison is the real new contribution.\n\nIt does several things well. The two-stage design (interviews, then co-design workshops) is appropriate. The authors are unusually transparent: they describe the protocol, the recruitment limits, and the codesign caveats. The design recommendations in Section 5 are concrete and tied to the data.\n\nThe soft spots are real but not disabling. The stress-test concern about prompted materials is fair. Policy experts were handed a prepared list of data types. Workers ranked sticky notes alongside prepopulated ones. The authors say they introduced the policy initiatives at the end to avoid priming, which shows awareness, but they never separate unprompted from prompted responses. So the alignment finding is at least partly shaped by what the researchers put in the room. This is an internal-validity concern for the main claim, not just a sampling issue.\n\nSample size and recruitment are also thin — eleven experts and fourteen workers, recruited through contacts, Reddit, and word of mouth. No inter-rater reliability metrics, no raw data. But for a qualitative HCI study these are within normal range, and the authors acknowledge the limits.\n\nThe stress-test note says the paper needs a prompted-vs-unprompted reanalysis. I'd stop short of calling it a load-bearing flaw; the finding is consistent with the broader literature and the authors made a deliberate effort to introduce the policy initiatives late. But the paper would be stronger if it reported whether the shared themes appeared in unprompted discussion.\n\nWho is this for? HCI/CSCW researchers, gig-worker advocates, and policy designers. It deserves a serious referee. I'd send it out with a request for the prompted/unprompted breakdown and a bit more transparency on coding consensus. I would not desk-reject it.","headline":"A useful cross-stakeholder map of gig-worker data-sharing needs, with a real but fixable prompt-contamination concern in the main alignment finding.","tokens_in":31381,"tokens_out":2149,"would_cite":true,"duration_ms":20518,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Gig workers and policymakers share data priorities: pay, discrimination, and safety are the common agenda for a worker-centered data-sharing system.","keywords":["gig work","worker data sharing","data collectives","fair pay","worker safety","discrimination","co-design","policy design"],"falsifier":"Survey a large, representative sample of U.S. gig workers and policy professionals on their top priorities for worker data use; if pay, discrimination, and safety do not emerge as the top shared initiatives, the finding does not generalize. Alternatively, deploy a prototype data-sharing system in one gig domain and observe whether workers contribute data and whether policymakers use it to inform actual policy documents.","tokens_in":30471,"feed_emoji":"📊","tokens_out":3646,"duration_ms":36018,"temperature":0.7,"pith_summary":"This paper tries to establish that U.S. gig workers and policy experts, despite working on different sides of the table, largely agree on what a worker-centered data-sharing system should accomplish: both groups want data that exposes pay practices and unpaid work time, and both want stronger attention to discrimination and safety. On top of that shared ground, policy experts want to measure worker stress, while workers want to share practical strategies and experiences with each other. The paper argues that a data-sharing system designed around these shared initiatives, with workers retaining control over who sees their data, could support both worker advocacy and policy-making. If the authors are right, this gives a concrete design agenda for building gig-worker data collectives that go beyond tax-tracking apps and actually address labor exploitation.","feed_headline":"Workers and policymakers share data priorities: pay, safety, bias","feed_subtitle":"A worker-run data collective that tracks these three areas could serve both advocacy and gig-work policy.","key_machinery":"The argument is carried by a two-pronged qualitative protocol: semi-structured interviews with 11 U.S. policy domain experts and co-design workshops with 14 active gig workers across four domains (freelancing, food delivery, rideshare, and petsitting). The central proposed object is a worker-centered data-sharing system (a data collective) that aggregates pay, discrimination, and safety data alongside qualitative worker narratives, with workers controlling access. Iterative thematic coding of the transcripts produces the distinction between shared initiatives and stakeholder-specific priorities, which then drives the design recommendations.","core_discovery":"The central claim is that both stakeholder groups sought 1) data to understand pay practices and (unpaid) work time and 2) more attention toward the issues of discrimination and safety. Workers and policy experts also showed distinct priorities: policy experts emphasized measuring worker stress, especially for caregivers and those juggling multiple roles, while workers prioritized the exchange of qualitative experiences and strategies. The paper concludes that these shared initiatives should anchor the design of worker-centered data-sharing systems, and that the system should balance worker control, data integrity, and multi-stakeholder governance to bring legislation closer to equitable gig work futures.","pith_inferences":["If the shared priorities are representative, a worker data cooperative could reasonably focus its first efforts on pay, safety, and bias metrics, leaving stress as a secondary data ask that policy experts may need to champion on their own.","The divergence between policy experts' focus on stress and workers' focus on experience-sharing suggests that real deployments will require negotiating data-collection scope; a pilot could test whether both types of data can be collected without burdening workers.","Because workers worried about data quality when manually uploading 'cherry-picked' good weeks, policy analyses built on worker-sourced data will likely have systematic blind spots unless hybrid collection methods (automatic plus prompted manual) are built in from the start.","The cross-domain differences observed among freelancers, delivery drivers, rideshare drivers, and petsitters imply that a shared system may need domain-specific data schemas and sharing preferences, which the current study maps qualitatively but does not yet quantify."],"forward_implications":["A data-sharing system that aggregates pay, discrimination, and safety data, and that also supports qualitative experience-sharing, could serve both worker advocacy and policy-making.","System designers should let workers control access (e.g., aggregate vs. individual data) while giving policymakers the summaries they need, since workers are wary of sharing raw individual data with government and peers alike.","Addressing data integrity is essential: workers' heterogeneous preferences for manual vs. automated uploads, plus off-app payments, can bias the data unless collection methods are designed carefully.","A data-sharing system should elevate worker-specific goals like experience-sharing even when they do not directly serve policy, because those features create the incentive for workers to participate.","Governance of such a system will likely require shared ownership among workers, advocacy groups, and neutral third parties, since participants voiced no consensus on a single owner."],"supporting_citations":[{"why":"Establishes the concept of worker data collectives for improving accountability and reducing inequalities, which this study directly extends to a multi-stakeholder data-sharing design.","marker":"[50]"},{"why":"Shows how worker data probes can inform policy language and align stakeholder objectives, providing the policy-facing rationale for the data-sharing system.","marker":"[111]"},{"why":"Documents gig workers' desires to engage in collective data investigations and audits of platform incentive structures, motivating the worker-side data needs.","marker":"[110]"},{"why":"Describes the Shipt Calculator, an existing worker-led pay-tracking tool that aggregated shared pay data, serving as a concrete precedent for the proposed system.","marker":"[14]"},{"why":"Reports co-design sessions where rideshare drivers articulated preferences for collective data infrastructures, a direct antecedent this study extends to multiple gig domains and policy experts.","marker":"[88]"},{"why":"Documents intrusive data collection and surveillance of gig workers by platforms and customers, feeding into the privacy and trust concerns addressed in the design.","marker":"[81]"},{"why":"Introduces Digital Workerism as a framing for worker-led data-driven research and governance tools, grounding the paper's approach to shifting power toward workers.","marker":"[13]"},{"why":"Argues that workers need to collect their own data as evidence of injustices to advocate for fair wages, supporting the paper's premise that worker data is a policy resource.","marker":"[56]"}],"fun_headline_variants":["Gig workers and policymakers align on data for pay, safety, bias","Worker data collectives target pay, safety, and bias first","Study: Worker-run data sharing can drive gig work policy","Pay, safety, bias: common ground for gig workers and policymakers","Shared data priorities could bridge gig work advocacy and policy"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the priorities voiced by 11 policy experts and 14 gig workers recruited through convenience sampling represent the broader population of U.S. gig workers and policy stakeholders.","fun_headline_variants_meta":{"raw":{"variants":["Gig workers and policymakers align on data for pay, safety, bias","Worker data collectives target pay, safety, and bias first","Study: Worker-run data sharing can drive gig work policy","Pay, safety, bias: common ground for gig workers and policymakers","Shared data priorities could bridge gig work advocacy and policy"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000162,"raw_usage":{"total_tokens":1187,"prompt_tokens":837,"completion_tokens":350,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":453,"completion_tokens_details":{"reasoning_tokens":264}},"tokens_in":453,"tokens_out":350,"duration_ms":4098,"temperature":1.0,"reasoning_tokens":264,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T22:53:40.495574+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Survey a large, representative sample of U.S. gig workers and policy professionals on their top priorities for worker data use; if pay, discrimination, and safety do not emerge as the top shared initiatives, the finding does not generalize. Alternatively, deploy a prototype data-sharing system in one gig domain and observe whether workers contribute data and whether policymakers use it to inform actual policy documents.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents intrusive data collection and surveillance of gig workers by platforms and customers, feeding into the privacy and trust concerns addressed in the design."}],"review_version":1}