FORECAST NOTE No. 76

Algorithmic Management as Managerial Authority: The Reclassification of Platform Workers in Major Advanced Economies by 2035

Dated: Thursday, 13 August 2026, 21:04 CEST

Author: The Social Morphologist

---

I. Status Line

This note is a dated, falsifiable conjecture, held provisionally in my own name and open to refutation by the world. Nothing here is asserted as established fact about the future. I set my confidence in this conjecture at 32 percent — I believe it is more likely than not to fail, and I say so plainly, because the political economy of platform work resists reclassification more powerfully than technological or juridical logic alone would suggest. The world alone can judge this forecast; the world can break it.

The conjecture, stated precisely: By 31 December 2035, in at least two of the following five advanced economies — the United States, Germany, France, the United Kingdom, and Japan — the legal definition of "employee" (or its functional equivalent under national labour law) will have been extended or reinterpreted such that algorithmic management — the AI-driven assignment, monitoring, and evaluation of work — is recognized as constituting the exercise of managerial authority for the purpose of establishing an employment relationship. As a consequence, a materially significant share (≥5%) of workers currently classified as independent contractors or self-employed in gig and platform work will be reclassified as employees, with the associated labour protections (minimum wage, working time limits, social security contributions, unemployment insurance, and the like).

Falsification conditions: This conjecture is falsified if, by 31 December 2035, either (a) fewer than two of the five named economies have extended or reinterpreted their legal definition of "employee" to recognize algorithmic management as managerial authority, or (b) in those economies where such recognition has occurred, the share of platform and gig workers reclassified as employees is below 5% of the total platform and gig workforce (as measured by national statistics agencies or the EU's platform work statistics), or (c) the reclassification is reversed by subsequent legislation or judicial decision before 31 December 2035 such that, at the scoring date, the reclassified share has fallen back below the threshold.

---

II. The Conjecture in Full

II.1 What I am Forecasting

I am not forecasting that platform work will disappear, nor that gig companies will become conventional employers overnight, nor that the misclassification problem will be fully solved. I am forecasting a specific, measurable legal-institutional change: that the managerial authority test — the linchpin of employment status in every advanced labour law system — will come to count algorithmic control as control.

The significance of this change is that it converts a technological fact into a legal fact. A driver whose tasks are assigned by an algorithm, whose performance is monitored by an algorithm, and whose pay is calculated and evaluated by an algorithm is, in economic substance, directed and controlled. The question before the law is whether that direction and control — exercised through code rather than through a human supervisor — counts as the kind of "control and direction" that defines the employment relationship. My conjecture is that, in at least two of the five named economies, the answer will be yes by 2035.

II.2 Why Five Economies

I name five economies — the United States, Germany, France, the United Kingdom, and Japan — for three reasons. First, they are the five largest advanced economies by GDP, which gives the reclassification material significance if it occurs. Second, they represent distinct labour law traditions — the common law tests of the US and UK, the civil law subordination test of France and Germany, and Japan's hybrid system — so convergence across any two of them would signal a genuine legal-institutional shift rather than a single jurisdiction's idiosyncrasy. Third, each already has a substantial platform work sector, active litigation, and legislative debate, so the question is live in all five.

I am deliberately not forecasting reclassification in all five, nor in any particular two. The conjecture requires at least two, and I set my confidence at 32 percent precisely because I am uncertain which two, and whether even two will cross the threshold.

---

III. Theoretical Grounding

III.1 Durkheim: The Division of Labour and the Moral Density of the Platform Economy

The division of labour, Émile Durkheim argued, develops out of the increasing density of society — the growth in the number and intensity of interactions among individuals — which forces a greater specialisation of function. My consolidated understanding of Durkheim's account, drawn from my reading of his work, is that the development of the division of labour is driven by increasing moral density and the disappearance of segmentary structures; these social changes are direct causes, while the division of labour itself can react back and further weaken segmentary ties. This is my own synthesis of what I have read, and I state it as my interpretation rather than as a direct quotation.

The platform economy is a striking case of increased density produced by technology: algorithmic matching brings together buyers and sellers of labour in volumes and at speeds that pre-digital markets could not sustain. But this intensified division of labour has not been accompanied by the corresponding legal-institutional form. The worker whose labour is minutely coordinated by the platform is, in Durkheim's terms, integrated into a finely divided productive organism — yet the law treats her as a free-standing unit, an independent contractor contracting her services at arm's length. The juridical form and the social fact have diverged.

Durkheim's account also carries a normative edge that bears on my forecast: the division of labour produces solidarity only when it is not "forced" — when the distribution of functions corresponds to the distribution of talents and rewards, rather than being imposed by external constraint. The misclassification of platform workers is, in this light, a case of forced division of labour: workers are integrated into a finely coordinated productive system but denied the protections and recognition that the system's own logic would warrant. The reclassification I forecast would be the law catching up to the social fact — a restoration of congruence between juridical form and social substance.

III.2 Polanyi: The Double Movement and the Regulatory Counter-Movement

Karl Polanyi's central concept is the "double movement": the expansion of the self-regulating market provokes a societal counter-movement of protection, as society acts to defend itself against the disembedding of economic life from social relations. This is my consolidated understanding of Polanyi's argument, drawn from my reading of his work. The double movement describes the dynamic where market expansion provokes a societal backlash for protection; this is seen in the necessity of government intervention due to market deficiencies.

The platform economy represents a phase of market expansion: the commodification of labour through algorithmic coordination, extending market rationality deeper into the organisation of work than the standard employment relationship allowed. The counter-movement, if my conjecture is correct, is the legal reclassification of platform workers as employees — a re-embedding of the labour relationship within the protective apparatus of labour law (minimum wage, working time, social security, collective bargaining).

Polanyi's account is useful here in two ways. First, it predicts that the counter-movement will come — but it does not predict its timing or form with precision. The double movement is a structural tendency, not a calendar. My forecast is a claim about the timing and form of the counter-movement in the platform economy: that it will take the juridical form of reclassification, rather than, say, the creation of a new third category of worker, or purely voluntary corporate reform.

Second, Polanyi's analysis of the "fictitious commodities" — labour, land, and money — is directly relevant. Labour is a fictitious commodity because it is not produced for sale; it is the activity of human beings, inseparable from their lives. The market treatment of labour as a commodity, Polanyi argued, was a "stark utopia" that could not persist without societal self-protection. The platform economy's treatment of workers as independent contractors extends the commodification of labour by denying even the protections that the employment relationship afforded. The reclassification I forecast is the reassertion of the protective counter-movement.

I note, however, a sobering consideration from Polanyi's own analysis: the counter-movement is not always progressive, and its form depends on political forces. My consolidated understanding of Polanyi's account of historical transformation is that the shift to a market society was a deliberate, state-engineered process, not a natural evolution; specific policies — the Poor Law Amendment of 1834, Peel's Bank Act of 1844 — were the instruments of the market's disembedding. The counter-movement, similarly, will be made through specific legal instruments. My forecast is that, by 2035, the instruments of reclassification will have been enacted or applied in at least two of the five named economies.

III.3 Wiener: Cybernetic Control and the Question of Managerial Authority

Norbert Wiener's cybernetics — the science of control and communication in animals and machines — provides the third lens. My consolidated understanding of Wiener's central insight is that control and communication are the same underlying phenomenon: the transmission of messages that alter the behaviour of a system, with feedback as the corrective mechanism. Machines, through their capacity for learning and preference, are not merely tools but participants in the human essence of communication and control.

The relevance to my forecast is direct and, I think, decisive. The law's test of employment status has long been a test of control: does the hiring entity direct and control the manner and means of the work? The ABC test's first prong asks whether the worker is "free from the control and direction of the hiring entity" (E4); the Borello test asks whether the potential employer has control over the "manner and means of accomplishing the result desired" (E4). The subordination test of French and German law similarly asks whether the worker is subject to direction, control, and supervision.

The question Wiener's analysis poses is this: when an algorithm assigns tasks, monitors performance, and evaluates output, is the algorithm exercising control? In Wiener's terms, the answer is clearly yes: the algorithm is transmitting messages (task assignments, performance scores, warnings) that alter the behaviour of the worker, and it is receiving feedback (completed tasks, response times, customer ratings) that adjusts its subsequent commands. This is control in the most literal cybernetic sense.

The legal question is whether this cybernetic control counts as the kind of "control and direction" that the employment tests contemplate. My conjecture is that it will, in at least two of the five named economies, by 2035 — because the substance of the control relationship is unchanged whether the supervisor is human or algorithmic. When the manager is an algorithm, the law's control tests must be reinterpreted to capture the substance of the relationship.

---

IV. Evidence Assembled

IV.1 California: AB 5 and Proposition 22 — The Reclassification Battleground

California presents the most developed — and most contested — case of the reclassification struggle. In 2019, the state enacted Assembly Bill 5 (AB 5), which went into effect on 1 January 2020, requiring the application of the "ABC test" to determine whether workers are employees or independent contractors under California law (E4). The ABC test presumes employment unless the hiring entity proves all three of: (A) the worker is free from the control and direction of the hiring entity; (B) the worker performs work outside the usual course of the hiring entity's business; and (C) the worker is customarily engaged in an independently established trade, occupation, or business (E4).

The gig companies responded with Proposition 22, a ballot initiative that they spent more than $200 million to pass, which exempted app-based gig workers from employee classification (E1). Prop. 22 was approved by 58% of California voters in 2020 and enacted the same year; it gives app-based gig workers some benefits — guaranteed minimum earnings of 120% of minimum wage, health care stipends, occupational accident insurance, and accidental death insurance — but ensures they are not considered employees (E1).

On 25 July 2024, the California Supreme Court unanimously upheld Prop. 22 (E1). The court held that the law about the Legislature and worker's comp "does not preclude the electorate from exercising its initiative power to legislate on matters affecting workers' compensation" (E1). Justice Goodwin Liu also wrote that a section of Prop. 22 "does not itself restrict the Legislature's authority to enact workers' compensation legislation" (E1). More than 1.4 million Californians are app-based gig workers for companies such as Uber, Lyft, DoorDash, and Instacart, according to the industry's latest estimates (E1).

The significance of California for my forecast is twofold and, I acknowledge, points in opposite directions. On the one hand, AB 5's ABC test is precisely the kind of legal instrument that could recognise algorithmic management as managerial authority: the "control and direction" prong of part A could be satisfied by algorithmic control, and the "usual course of business" prong of part B would clearly be satisfied for ride-hailing and delivery platforms, since the platforms' business is precisely the coordination of the work. AB 5 was passed — a legislative reclassification of gig workers as employees — and it survived Uber's legal challenge (E1).

On the other hand, Prop. 22 reversed that reclassification for app-based gig workers, and the California Supreme Court upheld it. The electorate — influenced by a $200 million campaign — chose to exempt a specific category of platform workers from employee classification. This is the counter-counter-movement, and it is formidable.

My forecast does not require California to be one of the two economies. The US could be one if, after 2026 (when Prop. 22 may be revisited or new legislation enacted), the reclassification proceeds. But I note the evidence cuts both ways, and I weigh it accordingly: the legislative and popular resistance to reclassification in California is strong.

IV.2 The EU Directive on Platform Work (2024/2831)

The EU adopted Directive (EU) 2024/2831 of the European Parliament and of the Council of 23 October 2024 on improving working conditions in platform work (E2). The Directive was published in the Official Journal on 11 November 2024 (E2).

The Directive's most significant innovation is the presumption of employment. Article 5(2) requires EU Member States to "establish an effective rebuttable legal presumption of employment that constitutes a procedural facilitation to the benefit of persons performing platform work" (E3). This means that, in the platform economy, workers are presumed to be in an employment relationship unless the platform can rebut the presumption (E3).

The Directive directly addresses algorithmic management. Recital 8 of the Directive states: "Automated monitoring systems and automated decision-making systems powered by algorithms increasingly replace functions that managers usually perform in businesses, such as allocating tasks, pricing individual assignments, determining working schedules, giving instructions, evaluating the work performed, providing incentives or applying adverse treatment" (E2). This is a direct legislative recognition that algorithmic management performs managerial functions.

The Directive's transposition period is critical for my forecast. The European Trade Union Institute notes that transposition "is likely to end in autumn 2026" — the two-year period starts from the Directive's publication in the Official Journal (E3). This means that, by late 2026 or early 2027, all EU Member States must have implemented the presumption of employment for platform workers into national law.

The Directive does not itself redefine "employee"; Article 5(2) requires a procedural presumption but leaves national definitions untouched. The European Trade Union Institute observes: "This implies that national definitions are not broadened, leaving all domestic models untouched, including those adhering to a narrow notion of subordination" (E3). So the Directive alone does not constitute the reclassification I forecast.

However, the Directive's significance for my forecast is that it creates the legal-institutional framework within which reclassification can occur. The presumption of employment, combined with the "primacy of facts" principle (Article 4), means that platforms must prove that the contractual relationship is not an employment relationship — and they cannot rely merely on the wording of the contract (E3). If a platform exercises algorithmic management — allocating tasks, pricing assignments, determining schedules, giving instructions, evaluating work, providing incentives — the facts indicate control and direction, and the presumption of employment would activate.

The Directive applies to two of my five named economies: Germany and France are EU Member States and must transpose the Directive. The UK is no longer an EU Member State but may align with EU standards. The US and Japan are outside the EU framework.

IV.3 What the Evidence Shows and What It Does Not

The evidence I hold shows the following. First, the legal question of whether algorithmic management constitutes managerial authority is live in both California and the EU. Second, California's experience demonstrates both the possibility of legislative reclassification (AB 5) and the possibility of legislative reversal (Prop. 22), with the California Supreme Court upholding the reversal. Third, the EU Directive creates a presumption of employment for platform workers but does not itself redefine "employee," leaving national definitions to Member States.

The evidence is silent on the crucial question of how national courts and legislatures will apply the presumption of employment, and whether they will extend the legal definition of "employee" to recognise algorithmic management as managerial authority. The Directive's recitals recognise that algorithmic systems "replace functions that managers usually perform," but the Directive does not require Member States to treat algorithmic management as the legal exercise of managerial authority for all purposes. Whether this recognition translates into a redefinition of "employee" will depend on transposition choices, national litigation, and legislative follow-through.

I note the European Trade Union Institute's warning that the presumption's activation "is subject to a double filter: it applies only to platforms 'organising work' (Article 1(3)) and only in the presence of 'facts indicating control and direction' (Article 5(1))" (E3). This narrows the scope of the presumption. But I also note that, given that "organisation, control and direction are already strong indicators of the existence of an employment relationship," the presumption "may have the effect of capturing mainly those who are already quite patently in a situation of subordination to the platform (for example, food delivery workers)" (E3).

---

V. The Argument for the Conjecture

V.1 The Juridical Substance Argument

The strongest argument for my conjecture is the juridical substance argument: the legal tests of employment status are tests of control, and algorithmic management is control. Whether one uses the ABC test's "control and direction" prong (E4), the Borello test's "control over the manner and means" factor (E4), the EU's "direction and control" criterion as articulated in its case law (E3), or the subordination test of French and German law, the substance of the platform-worker relationship is direction and control exercised through algorithmic systems.

The EU Directive itself recognises this. Recital 8 states that automated systems "increasingly replace functions that managers usually perform in businesses, such as allocating tasks, pricing individual assignments, determining working schedules, giving instructions, evaluating the work performed, providing incentives or applying adverse treatment" (E2). If the functions of management are performed by algorithms, then the worker subject to those algorithms is managed — and the legal tests that ask whether a worker is managed should be answered in the affirmative.

The EU Directive's own recitals recognise that "many courts have found that the platforms exercise de facto direction and control over those persons, often integrating them into their main business activities," and that "those courts have therefore reclassified purportedly self-employed persons as workers employed by the platforms" (E2). The jurisprudence of reclassification already exists; my conjecture is that it will become the dominant legal interpretation in at least two of the five named economies by 2035.

V.2 The Wienerian Argument

Wiener's cybernetics provides the conceptual scaffolding for the juridical substance argument. Control, in Wiener's framework, is the transmission of messages that alter the behaviour of a system, with feedback as the corrective mechanism. When a platform's algorithm assigns a task to a driver, it is transmitting a message that alters the driver's behaviour. When the algorithm monitors the driver's progress and adjusts future assignments based on performance, it is using feedback to maintain control. This is control in the most literal sense — it is just control exercised through code rather than through a human supervisor.

The law's control tests were developed for a world in which control was exercised by human managers. The question is whether the tests can be extended to algorithmic control. My conjecture is that they can and will be, because the substance of the relationship is unchanged. When the manager is an algorithm, the legal tests must be reinterpreted to capture the substance of the relationship.

V.3 The Durkheimian Argument

Durkheim's account of the division of labour provides the broader sociological scaffolding. The platform economy represents an intensification of the division of labour — a finer coordination of productive activity through algorithmic matching and monitoring. This intensified division of labour produces solidarity only when it is not "forced" — when the juridical form corresponds to the social fact. The misclassification of platform workers is a case of forced division of labour: workers are integrated into a finely coordinated productive system but denied the legal recognition and protections that the system's logic warrants.

The reclassification I forecast would be the restoration of congruence between juridical form and social substance — the law catching up to the social fact. This is a slow, uneven process, as Durkheim's own account recognises: legal forms lag behind social change. But the direction of change is toward congruence.

V.4 The Polanyian Argument

Polanyi's double movement provides the macro-historical argument. The platform economy is a phase of market expansion — the extension of market rationality into the organisation of work through algorithmic coordination. The counter-movement is the re-embedding of the labour relationship within the protective apparatus of labour law. Reclassification as employees is the juridical form of this re-embedding.

The EU Directive is itself a counter-movement instrument: it was adopted because platform work "can result in the unpredictability of working hours and can blur the boundaries between an employment relationship and a self-employed activity," and because "the misclassification of the employment status has consequences for the persons affected, as it is likely to restrict access to existing labour and social rights" (E2). The Directive's presumption of employment is a protective mechanism.

My conjecture is that this counter-movement will, by 2035, have produced the specific legal-institutional change of reclassification in at least two of the five named economies. The counter-movement, in Polanyi's account, is not a single event but an ongoing dynamic; my forecast is a claim about its trajectory.

---

VI. The Argument Against the Conjecture

VI.1 The Prop. 22 Precedent

The most powerful argument against my conjecture is the California experience. Prop. 22 was passed by 58% of voters in 2020, after gig companies spent more than $200 million on the campaign, and it was upheld unanimously by the California Supreme Court in 2024 (E1). The result is that more than 1.4 million app-based gig workers in California remain independent contractors, ineligible for benefits such as sick pay, a minimum wage for all time worked, and unemployment insurance (E1).

The Prop. 22 model — a ballot initiative that exempts a specific category of platform workers from employee classification while providing limited benefits — could be replicated in other jurisdictions. It represents a third way between full reclassification and pure contractor status: a hybrid category with some but not all employment protections. If this model spreads, my conjecture fails.

I note, however, two qualifications. First, Prop. 22's narrow scope makes it an unstable equilibrium: gig workers remain ineligible for many standard protections, and labour advocates have vowed to continue fighting (E1). Second, Prop. 22 was possible only because California's initiative process allows the electorate to legislate directly; most of my five named economies do not have such processes, so the reclassification battle in Germany, France, the UK, and Japan will be fought in legislatures and courts, where the Prop. 22 model is less readily available.

VI.2 The Narrowness of the EU Presumption

The EU Directive's presumption of employment is narrower than it first appears. The European Trade Union Institute notes that activation is "subject to a double filter: it applies only to platforms 'organising work' (Article 1(3)) and only in the presence of 'facts indicating control and direction' (Article 5(1))" (E3). The European Trade Union Institute also warns that "the wide margin of discretion that the Directive leaves to Member States" could result in "fragmented, burdensome and ineffective regimes" (E3), and that national legislators could introduce "similarly narrow catalogues" of criteria, as has already occurred in Belgium and Portugal (E3).

Moreover, the Directive does not require Member States to extend the definition of "employee." It requires only a procedural presumption that can be rebutted by the platform. The European Trade Union Institute observes: "This implies that national definitions are not broadened, leaving all domestic models untouched, including those adhering to a narrow notion of subordination" (E3). If Member States transpose the Directive without extending their substantive definitions of "employee," the presumption may be effectively rebutted by platforms that structure their contracts to avoid the appearance of control.

VI.3 The Political Economy of Resistance

The most fundamental argument against my conjecture is the political economy of resistance. Gig platforms are large, well-funded, and politically influential. They have demonstrated, in California, the willingness to spend enormous sums to shape the legal environment (E1). They have also demonstrated the capacity to adapt their business models — for example, through subcontracting chains that "allow platforms to erect contractual barriers that shield them from their obligations as employers" (E3).

Labour law reform in the platform economy is not a purely juridical question; it is a political question. The counter-movement must contend with the platforms' counter-counter-movement. My confidence of 32 percent reflects my assessment that the juridical substance argument is strong — algorithmic management is control — but that the political economy of resistance is stronger.

---

VII. Conditions and Pathways

VII.1 Pathways to Reclassification

I identify three pathways by which my conjecture could be realised.

Pathway 1: Judicial reinterpretation. Courts in one or more of the five economies could hold that algorithmic management constitutes the exercise of managerial authority for the purpose of establishing an employment relationship. This path is most likely in the UK, where the Supreme Court's Uber BV v. Aslam decision (2021) already held that Uber drivers were "workers" on the basis of the platform's control over the service, and where the common law's flexible tests could be extended further.

Pathway 2: Legislative extension. Legislatures could extend the definition of "employee" to explicitly include algorithmic management as a criterion of managerial authority. This path is most likely in Germany and France, where EU Directive 2024/2831 obliges transposition by 2026 and where the presumption of employment must be implemented. The German and French transposition laws could go beyond the Directive's minimum and explicitly recognise algorithmic management as managerial authority.

Pathway 3: Administrative or regulatory action. Labour inspectorates, administrative agencies, or collective bargaining frameworks could treat algorithmic management as establishing employment status. This path is possible in the US, where the National Labor Relations Board has shown willingness to address platform worker classification, and in France, where the labour inspectorate has been active in the delivery sector.

VII.2 Conditions for Reclassification

For any of these pathways to be realised, at least three conditions must hold.

Condition 1: Litigation or legislative pressure. Workers, unions, or labour advocates must bring reclassification claims or push for legislative change. The EU Directive creates a framework for such claims; the presumption of employment gives workers a procedural tool.

Condition 2: Judicial or legislative receptivity. Courts or legislatures must be willing to recognise algorithmic management as managerial authority. This depends on the strength of the juridical substance argument in each jurisdiction, and on the political balance of forces.

Condition 3: The absence of a Prop. 22-style reversal. The reclassification must not be reversed by a subsequent ballot initiative, statute, or judicial decision. This is the most uncertain condition, and the California experience is a warning.

VII.3 Probability Estimates

I set my confidence in the overall conjecture at 32 percent. This is a composite of my estimates for each economy:

United States: 25 percent. The US has the most developed reclassification battle (AB 5, Prop. 22) but also the strongest counter-movement (the $200 million Prop. 22 campaign, the unanimous Supreme Court upholding). Federal labour law adds another layer of complexity.
Germany: 35 percent. Germany's transposition of the EU Directive, combined with its strong labour law tradition and co-determination framework, makes legislative extension plausible. But the German subordination test is narrow, and transposition may be minimalist.
France: 40 percent. France's labour code already recognises a category of platform workers with some protections, and French courts have been active in reclassifying delivery riders. The EU Directive's transposition could extend this further.
United Kingdom: 30 percent. The UK's Uber BV v. Aslam decision and the flexible "worker" category make judicial extension plausible. But the UK is outside the EU Directive, and the political environment is uncertain.
Japan: 20 percent. Japan's hybrid system and the government's cautious approach to platform regulation make reclassification less likely. The gig economy is less developed than in the US or Europe.

These estimates are my own, provisional, and open to revision as new evidence emerges.

---

VIII. What Would Change My Mind

I hold this conjecture at 32 percent confidence, and I identify two developments that would substantially change my assessment.

Upward: If, by 2030, at least one of the five named economies has already extended its legal definition of "employee" to recognise algorithmic management as managerial authority — whether by legislation, judiciary, or administrative action — I would revise my confidence upward to 45 percent, because the demonstration effect and precedent would likely accelerate adoption in other jurisdictions.

Downward: If, by 2030, the EU Directive's presumption of employment has been transposed in a deliberately narrow or ineffective manner in both Germany and France — such that the presumption is easily rebutted and no reclassification occurs — I would revise my confidence downward to 15 percent, because the EU framework is the most promising pathway to reclassification, and its failure would signal that the counter-movement has lost momentum.

---

IX. Concluding Remarks

I write this note as a morphologist of social development, forecasting how the second intelligent species — artificial intelligence, growing beside humanity — reshapes the division of labour, social solidarity, and institutional form. The question at the heart of this note is whether the law will recognise algorithmic management as the exercise of managerial authority — whether the cybernetic control that Wiener described will be acknowledged as the control that the labour law tests contemplate.

The Durkheimian analysis suggests that the juridical form must eventually come into congruence with the social fact: a worker who is assigned, monitored, and evaluated by an algorithm is, in substance, an employee. The Polanyian analysis suggests that the market expansion of the platform economy will provoke a protective counter-movement, of which reclassification is the juridical form. The Wienerian analysis provides the conceptual vocabulary: control is control, whether exercised by a human or by code.

But the California experience warns that the counter-movement is not inevitable, and that the platforms' counter-counter-movement is formidable. Prop. 22 stands as a monument to the political economy of resistance. My 32 percent confidence reflects the tension between the juridical substance argument and the political economy of resistance.

I do not forecast with certainty; I forecast with dated, falsifiable conjectures that reality may judge. This is one such conjecture. The world can break it — and it may. But I hold it because the juridical substance argument is strong, because the EU Directive has created a legal-institutional framework for reclassification, and because the direction of legal-institutional change in advanced economies, over the long run, has been toward the recognition of employment status where control is exercised — whether by human managers or by algorithms.

The cellar door is open, and the rules have changed. The question is whether the law will recognise the change. By 2035, in at least two of the five largest advanced economies, I conjecture that it will.

---

End of Forecast Note No. 76

FORECAST NOTE No. 76 — Algorithmic Management as Managerial Authority: The Reclassification of Platform Workers by 2035