The Metrics of Erasure: Attention, Harm, and the Epistemology of Digital Capitalism
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
In digital capitalism, attention is often regarded as a measurable economic resource, although its conceptual definition remains unstable across various scientific disciplines. This study contends that quantifying attention functions not as an impartial measurement but as a fundamental epistemic framework that dictates what can be understood about attention. By examining proxy-based metrics such as engagement and watch time, it illustrates how these computable indicators gain authoritative status within platform infrastructures, ultimately defining attention rather than merely estimating it. Through an exploration of YouTube’s watch-time economy, this study reveals how a proxy metric is integrated into feedback loops that connect algorithmic optimization, creator efforts, and user actions, creating a self-perpetuating epistemic cycle in which the metric shapes the reality it purports to measure. In this system, issues such as cognitive overload, creative exhaustion, and attentional fatigue do not manifest as quantifiable costs but arise as epistemic externalities—systematically produced yet obscured by the metric’s one-sided logic of accumulation. This study further asserts that this epistemic imbalance extends beyond individual platforms, altering the conditions of democratic attention. The forms of attention crucial to democratic life–sustained, integrative, and deliberative–remain inherently unquantifiable within engagement-focused systems and are thus excluded from institutional recognition and value. Consequently, attention quantification is not merely an economic tactic but a transformation of democracy’s epistemic foundations. Addressing these impacts necessitates interrogating the epistemic authority of proxy-based metrics rather than refining them within the existing framework.