Proportionality
Mark Leiser — Journal article
Content
Proportionality is often invoked as if the word itself supplies judgement. It does not. In EU law, the principle performs a harder task: it converts lawful power into a public account of the justification for interference. This article reconstructs proportionality as the constitutional law of technological architecture. It argues that proportionality integrates competence, market freedoms, administrative discretion, fundamental rights, privacy, data protection and technology regulation by asking a single disciplined question: Does the design of the interference remain within the limits of the reason that justifies it? The argument proceeds in four stages. 1.It traces the principle from older ideas of measured authority, through German administrative and constitutional law, Strasbourg's fair-balance reasoning, the Court of Justice's general principles, and the Charter. 2.It develops proportionality as a judicial method and distinguishes several modes of review: deferential legislative review, strict rights review, remedial proportionality, infrastructure proportionality, lawful-basis proportionality, safeguards proportionality and explanation proportionality. 3.It reads the hard technology cases as conflicts among rights and legally protected freedoms rather than as privacy against nothing. 4.It proposes a doctrinally grounded protocol for GDPR and AI Act practice. The protocol starts with interference intensity, specifies aims operationally, tests suitability with evidence, treats necessity as a design comparison, makes safeguards constitutive, requires a residual burden review, and insists on a temporal review. The article rejects both privacy absolutism, including the anti-proportionality rhetoric associated with Max Schrems, and managerial balancing, which treats rights as a source of friction in a business plan. Properly understood, proportionality does neither. It supplies the legal discipline by which public authorities, supervisory bodies, platforms, controllers and AI deployers must explain why the rights burden of a technological design does not exceed its justification.