Repairing algorithmic harms requires resources the poor already lack.

Evidence

People are also differently placed to repair the costs inflicted through surveillance systems. Imagine the costs when every object, most processes, and much of one’s work and other life opportunities are managed through automated monitoring. You may need to buy a new tool when its built-in tracking system breaks down; you may need to find alternative sources for education, health care, or insurance when denied them on algorithmic grounds; or you may need to correct inaccurate data held about you in a system—for example, a credit rating. People’s ability to bear those costs will depend, as ever, on a preexisting (likely unequal) distribution of resources. And the more opaque the processes of data sorting are, the more resources will be needed to contest them effectively. — Nick Couldry & Ulises A. Mejias, The Costs of Connection, 484

Categorization

• XREF: Foregrounds ‘repair costs’ as a distinct axis of inequality — connects to the reader’s data-colonialism vs. attention-school comparison; algorithmic contestation costs parallel the reader’s surveillance literature. — connects to: Data extraction costs fall hardest on the vulnerable under a moralistic blame frame. — Nick Couldry & Ulises A. Mejias, The Costs of Connection: How Data Is…, 608; AI lacks the ability to address deep-rooted social issues effectively. — Emily M. Bender, Alex Hanna, The AI Con- How to Fight Big Tech’s…, 269

Broader Connections

• Data extraction costs fall hardest on the vulnerable under a moralistic blame frame. #corroborates (0.80) — Nick Couldry & Ulises A. Mejias, The Costs of Connection, 608
• AI lacks the ability to address deep-rooted social issues effectively. #qualifies (0.72) — Emily M. Bender, Alex Hanna, The AI Con- How to Fight Big Tech’s…, 269