Gemma Dipoppa
banner
gemmadipoppa.bsky.social
Gemma Dipoppa
@gemmadipoppa.bsky.social
Assistant Professor Columbia Political Science - Political Economy, Migration, Crime, Environmental
https://www.gemmadipoppa.com/
11/
🎯 The takeaway:
States use surveillance as a preventive tool against the empowerment of educated but excluded groups.

👉 As excluded groups gain political empowerment, surveillance may reproduce inequalities by silencing them exactly as they gain political voice.
November 24, 2025 at 11:41 PM
10/ Across democracy and dictatorship, and with more and less technology to collect data, the logic of surveillance was similar:

States target those combining political capacity (education) with radical grievances (subalternity).
November 24, 2025 at 11:41 PM
9/ ➡️ Result 3:

Across 5 indicators of political activism—voting, protests, strikes, holding political roles, and armed resistance—educated cohorts did not become more engaged.

Surveillance expanded preventively, not in reaction to mobilization.
November 24, 2025 at 11:41 PM
8/ ➡️ Result 2:
Who faced the brunt of surveillance? The working class. The newly educated poor were watched longer, more harshly, and more intensively, consistently with the state fearing their empowerment.
November 24, 2025 at 11:41 PM
7/ The effect persists even when individuals move elsewhere, in line with surveillance following a portable asset (education) rather than municipal-level changes.
November 24, 2025 at 11:41 PM
6/ We present 3 results:

➡️ Result 1:

Municipality-cohorts exposed to more schooling were 64% more likely to be surveilled.

The effect increases as the state expands education and disappears when later reforms equalize schooling across municipalities.
November 24, 2025 at 11:41 PM
5/ 🎲 The shock:

The Casati Law mandated primary schooling for 2 years everywhere but extended it for +2 years in towns >4,000 inhabitants and cohorts born post 1854.

We show the reform reduced illiteracy and use it in a difference-in-discontinuity design by population and cohort.
November 24, 2025 at 11:41 PM
4/ 📃 📄 Descriptively, educated people were more likely to be watched. But education may proxy for background, status, or other fixed attributes.

We need a shock to education that affects otherwise similar people.
November 24, 2025 at 11:41 PM
3/ Unsupervised LLM on 1,200+ police files shows that mobilization capacity – and a particular marker of it, education, – together with potential for subversion are recurring traits noted by the surveillance state. 📚
November 24, 2025 at 11:41 PM
2/ 💡 We propose that states strategically target those combining capacity to mobilize with grievances for radical mobilization – educated but subaltern individuals perceived as most threatening to state stability.

This idea is rooted in descriptive data:
November 24, 2025 at 11:41 PM