Algorithmic management

Algorithmic management is the analytical-and-advocacy framing through which gig-worker organisers, civil-society researchers, and policy bodies name the category of work-direction practices in which software algorithms — rather than human managers — allocate tasks, set prices, score performance, surface or withhold work, and trigger dismissals across a platform's workforce. The framing was coined inside academic human-computer-interaction research in 2015, translated into a US digital-rights / worker-research register by Data & Society in 2019, operationalised into grassroots worker-organising vehicles by Worker Info Exchange and the App Drivers and Couriers Union from 2021, and carried into supranational and EU-level labour policy by the International Labour Organization's algorithmic-management workstream and the EU Platform Work Directive 2024/2831. The framing's role inside the corpus is structurally distinct from worker-authored framings in the same neighbourhood — msg-modern-slavery-content-moderation is the Nairobi-anchored worker register for outsourced moderation; msg-bossware is the US digital-rights register for workplace-surveillance software — in that "algorithmic management" was coined inside academic research and travels primarily through researcher-NGO-policy registers, with its corpus-relevance resting on the demonstrated uptake by worker-organising vehicles such as the Coalition of Digital Employees – Artificial Intelligence (CODE-AI) in the Philippines and the Worker Info Exchange / ADCU partnership in the UK and the Netherlands.

Origin

The framing's academic seed is the Carnegie Mellon Human-Computer Interaction Institute paper Working with Machines: The Impact of Algorithmic and Data-Driven Management on Human Workers by Min Kyung Lee, Daniel Kusbit, Evan Metsky, and Laura Dabbish, presented at the 33rd Annual ACM Conference on Human Factors in Computing Systems (CHI '15) on 18 April 2015. The paper studied Uber and Lyft drivers in an empirical field investigation and defined algorithmic management as software algorithms that assume managerial functions and surrounding institutional devices that support algorithms in practice. The definition's two halves work in tandem: the first locates the framing's substantive subject in the software itself (the dispatch algorithm, the rating algorithm, the dynamic-pricing algorithm); the second extends the framing to the wider socio-technical apparatus around the software (the customer-service line that workers cannot escalate to, the rules-of-engagement documents that incorporate the algorithm's outputs by reference, the terms-of-service architecture that converts algorithmic outputs into contractual facts). Together the two halves describe a labour-direction system in which managerial authority has been substantively delegated to software and to the institutional fabric that supports the software, rather than retained by human managers exercising professional judgement.

The 2015 paper set out the framing's working architecture in three substantive components — work assignment, informational support, and performance evaluation — through which the platform-worker experience of being directed by an algorithm could be analytically decomposed. This three-part architecture has carried into subsequent academic, civil-society, and policy literatures, and supplies the framing's load-bearing analytical vocabulary across registers.

Travel into a civil-society / worker-research register

The framing's first sustained translation outside academic literature is the Algorithmic Management in the Workplace explainer published by Data & Society in February 2019, authored by Alexandra Mateescu and Aiha Nguyen. The explainer redefines algorithmic management as "a diverse set of technology tools and techniques that structure the conditions of work and remotely manage workforces" — a working definition that extends the 2015 Carnegie Mellon scope from the ride-hail sector into retail, service-industry, and delivery-and-logistics work, and that pairs algorithmic management with the parallel category of workplace monitoring and surveillance (the co-released Data & Society explainer of the same date). Mateescu and Nguyen's translation locates algorithmic management inside the wider research-and-advocacy register that Data & Society has occupied across the 2010s-2020s — the same register that has carried Unmasking AI-adjacent algorithmic-accountability arguments — and supplies the framing's English-language working vocabulary for the US civil-society readership outside academic publishing.

The framing has carried into international human-rights advocacy through reports such as Human Rights Watch's May 2025 The Gig Trap: Algorithmic, Wage and Labor Exploitation in Platform Work in the US, which situates "algorithmic management" alongside Veena Dubal's "algorithmic wage discrimination" as the working analytical pair through which US platform-worker rights claims are now framed in international human-rights register.

Travel into grassroots worker-organising vehicles

The framing's principal grassroots-organising carriage in the corpus runs through Worker Info Exchange and the App Drivers and Couriers Union. The WIE / ADCU Managed by Bots — Data Driven Exploitation in the Gig Economy report, published on 13 December 2021 and co-authored by WIE Research Lead Cansu Safak and WIE Director James Farrar with a foreword by Bama Athreya (then a Fellow of the Open Society Foundations), is the first sustained civil-society documentation of how seven major gig platforms — Amazon Flex, Bolt, Deliveroo, Free Now, Just Eat, Ola, and Uber — use algorithmic systems to allocate work, set pay, score performance, and dismiss workers. The report uses the verbatim phrasing "algorithmic management practices by gig platforms" in its introduction and titles its substantive analytical section "Misclassification 2.0: Controlled by Algorithms" — a framing of the algorithmic-management category as the successor question to the employment-classification disputes that preceded it.

Managed by Bots operationalised the framing's substantive content into the Amsterdam Court "robo-firing" litigation arc that WIE and ADCU brought against Uber and Ola Cabs from 2020 and that resulted in the 4 April 2023 Amsterdam Court of Appeal ruling finding that several Uber and Ola automated processes — ride assignment, dynamic pricing, driver rating, fraud-probability scoring, and account deactivation — qualify as automated decision-making under Article 22 of the EU General Data Protection Regulation. The litigation's substantive theory of harm — that algorithmic systems direct platform-driver work in ways that workers can neither inspect nor contest — is the framing's clearest single legal-pleading translation, and supplies the framing's working precedent for GDPR Article 22 worker-protection claims across the European platform economy.

The framing has carried into Southeast Asian platform-worker organising through the Coalition of Digital Employees – Artificial Intelligence (CODE-AI), the Philippine multi-sector labour coalition launched in Quezon City on 25 January 2025 by seven founding organisations spanning BPO, delivery, telecommunications, and IT work. CODE-AI organises against the AI-co-pilot, sentiment-analysis, and speech-analysis tooling deployed against Concentrix call-centre agents in Manila — Concentrix's tooling scores agents in real time on tone, pitch, mood, positive language use, interruptions, hold times, and issue-resolution speed — and adopts "algorithmic management" alongside "AI-powered workplace surveillance" as the working analytical category for the practices its founding members organise against. The coalition's policy participation in the Philippine congressional working group on House Bills 7913 and 7983 (the proposed Artificial Intelligence Regulation Act) is the framing's clearest Global South legislative-engagement venue to date.

Translation into a continuous worker-engaged ratings methodology

The Fairwork Project, coordinated by Professor Mark Graham at the Oxford Internet Institute in partnership with the Berlin Social Science Center (WZB), has translated the framing into a continuous worker-engaged platform-ratings methodology. Fairwork rates digital labour platforms against five principles of fair platform work — fair pay, fair conditions, fair contracts, fair management, and fair representation — and the "fair management" principle is the framing's operational anchor inside Fairwork's audit instrument: it assesses whether algorithmic-management systems on a platform afford workers meaningful information about how decisions are made, meaningful processes for contesting decisions, and meaningful human oversight of consequential automated outcomes. Fairwork has rated 716 platforms in 40 countries since 2018 and has engaged ILO representatives directly on empirical evidence concerning algorithmic management, employment classification, and platform accountability — supplying the framing's bridge from worker-side audit data into the supranational policy register the next section describes.

Travel into supranational labour-policy register

The framing's most consolidated supranational anchor is the International Labour Organization's algorithmic-management workstream. The ILO frames algorithmic management as a category that "is growing across industries, particularly in customer service, transport, logistics, banking and health care" and that is "the defining feature of digital labour platforms", and its global gig-worker survey — finding that 37% of app-based taxi drivers and 48% of delivery drivers cannot refuse or cancel work without repercussions under algorithmic direction — is the framing's load-bearing empirical anchor for the working argument that algorithmic management materially erodes platform workers' rights and quality of life. The ILO's policy position calls for a "human-in-command" approach to ensure that "final decisions affecting work are taken by human beings" — a working policy formula that supplies the framing's substantive demand inside ILO Convention-and-Recommendation drafting toward the 2026 International Labour Conference, at which gig-work standards are scheduled for final-round negotiation.

Travel into EU regulatory law: Directive 2024/2831

The framing's most consequential legislative anchor is Directive (EU) 2024/2831 of the European Parliament and of the Council of 23 October 2024 on improving working conditions in platform work. The directive — politically agreed in March 2024 and adopted by the Council on 14 October 2024 — dedicates a chapter to algorithmic management and constitutes, in the Council's framing, "the first EU rules on the use of artificial intelligence in the world of work". The directive's algorithmic-management chapter imposes transparency requirements on automated decision-making systems used in platform work, requires human oversight of significant decisions (including a right of contestation for workers), prohibits the processing of emotional and psychological data and certain other sensitive categories, and requires regular evaluation of the impact of automated systems on workers' health, safety, and well-being. Member states must transpose the directive by 2 December 2026.

The directive is the framing's most consequential single piece of regulatory text to date. Where the WIE / ADCU robo-firing litigation translated algorithmic management into Article 22 GDPR rights claims under existing data-protection law, Directive 2024/2831 promotes the framing from a litigation-side argument under generally-applicable law into the named subject of a sector-specific regulatory regime. It is the load-bearing instance, inside the EU's regulatory architecture, of the framing's worker-rights demand being cashed out as positive law.

SCOPE-EDGE: framing-authored-at-academic-edge

"Algorithmic management" sits at a structural edge of the corpus's grassroots-and-democratic scope. Unlike the worker-authored register of msg-modern-slavery-content-moderation — coined by Daniel Motaung in an openDemocracy whistleblower interview from inside the African Content Moderators Union's organising lineage — and unlike the digital-rights-advocacy register of msg-bossware, coined by the Electronic Frontier Foundation from outside any specific worker-organising vehicle, "algorithmic management" was coined inside academic human-computer-interaction research at Carnegie Mellon and travels primarily through researcher-NGO-policy registers. Its principal carriers are Carnegie Mellon HCI researchers and their academic successors; Data & Society and similar US digital-rights research institutes; Fairwork and similar Oxford-affiliated empirical research projects; and supranational policy bodies including the ILO and the EU institutions.

The framing's in-scope status for this corpus rests on its demonstrated uptake into grassroots worker-organising vehicles whose work is substantively in scope. Worker Info Exchange and the App Drivers and Couriers Union have adopted the framing into the Managed by Bots report, into the Amsterdam Court litigation arc, and into the Stichting WIE International dynamic-pay collective action of November 2025. CODE-AI has adopted the framing into the Philippine BPO and platform-worker organising layer alongside its "AI-powered workplace surveillance" working register. The framing's regulatory anchor in Directive 2024/2831 and its empirical anchor in the ILO's global gig-worker survey supply the framing's load-bearing case for inclusion: workers organising against the practices the framing names now operate inside a legal-and-regulatory infrastructure that uses the framing as its working analytical category.

Why it has carried

Three features have made the framing durable.

First, it names a category of practice rather than a particular product or a particular workplace. Where the adjacent bossware framing anchors on a software product field (the vendor, the install base, the feature list) and where modern-slavery / neo-colonialism anchors on a specific labour relation in a specific supply chain, "algorithmic management" names the practice itself — work direction by software — at the level of generality that makes it portable across ride-hail (Uber, Lyft, Bolt, Free Now), food delivery (Deliveroo, Just Eat, food-delivery operations across Southeast Asia and Latin America), BPO and call-centre work (Concentrix, Accenture), warehouse logistics (Amazon Flex), and the wider customer-service / transport / logistics / banking / health-care category surface the ILO names. That portability is the framing's principal feature, and the feature that has carried it across press, academic, advocacy, and legislative writing over more than a decade.

Second, the framing arrived attached to an academic working definition — software algorithms plus surrounding institutional devices — that legislators, regulators, and advocates can operationalise without rewriting. The EU Platform Work Directive's chapter architecture, the ILO's "human-in-command" policy line, Fairwork's "fair management" rating principle, and WIE's "Misclassification 2.0: Controlled by Algorithms" framing all rest on the same underlying analytical decomposition the 2015 CMU paper set out. The framing's academic origin is in this respect a strength rather than a scope-edge weakness: the analytical decomposition is portable into regulatory, advocacy, and litigation contexts that worker-authored framings would need to translate before they could carry equivalent legislative weight.

Third, the framing arrived early enough in the platform-economy's expansion that it has accumulated more than a decade of empirical research, policy adoption, legal-pleading translation, and regulatory codification under the same analytical category. By 2026 "algorithmic management" is not a framing under contest; it is the working analytical category of EU regulatory law, ILO labour-standards drafting, and the platform-worker organising lineage that runs from WIE/ADCU in the UK to CODE-AI in the Philippines. That decade-of-accumulation gives the framing a load-bearing infrastructure — empirical, legal, regulatory, organisational — that the corpus's worker-authored framings of more recent vintage do not yet match.