The ideology that data always outweighs human judgement. And that more data is always better.
Literal meaning: Dataism is the ideological position that data-driven decisions are inherently superior to decisions based on human judgement, that more data is always better, and that expanding data collection is by definition social progress. The term was used by Yuval Noah Harari in Homo Deus (2016) to describe an emerging belief system.
Origin: Harari introduced the term in 2016 to describe a worldview in which data is treated as the universe’s most valuable product, and information flows as the primary measure of worth. The term circulates more broadly in critical technology studies to name a set of unspoken assumptions that drive the data economy: that quantification produces objectivity, that more data guarantees better outcomes, and that data collection itself is neutral.
The ideology that equates quantification with objectivity and data collection with progress — rendering invisible the social and political dimensions of what gets measured, by whom, and for whose benefit.
The Appeal: Dataism rests on real insights. Data-driven decisions can reduce human cognitive bias. Large-scale data analysis reveals patterns that qualitative approaches miss. More information is, in many contexts, better than less. The assumption is not irrational — it is incomplete.
The Friction: The ideology conceals the political dimensions of what gets measured. Which variables are included in a dataset? Whose experiences are easiest to quantify? Which outcomes get optimised for? Informatics of Domination — Haraway’s 1985 framework — shows precisely this: data systems encode existing hierarchies while claiming neutrality. Algorithmic Violence is the result: systems trained on historically biased data reproduce that bias at scale, while projecting an appearance of objectivity. Data Brokers are the commercial infrastructure that operationalises dataism. Situated Knowledge is the epistemological corrective: all knowledge, including data, is produced from a position. AI Literacy is that same correction at the scale of a single user: knowing that a model’s output is produced rather than found is what keeps an answer from being read as a measurement.
Why This Matters: Dataism names the ideology that renders data collection and use invisible as political choices. Once you can identify the ideology, you can ask the questions it obscures: what data is collected, by whom, for whose benefit, and what goes unmeasured?
Related terms: Surveillance Capitalism · Informatics of Domination · Algorithmic Violence · Situated Knowledge · Data Brokers · AI Literacy · Privacy as a Premium
Read more:
- Homo Deus — Harari, Y.N. (2016). Harvill Secker
- Automating Inequality — Eubanks, V. (2018). St. Martin’s Press — on what data-driven systems miss