The Facebook study’s assumption that choices are independent of algorithm exposure is shaky
Evidence
But not so fast. First, for liberals the impact of algorithmic filtering (8 per cent) is actually higher than that of individual choice (6 per cent). Second, the study has a number of serious limitations. For example, the study only sampled individuals who self-declared their ideology – a tiny fraction of all Facebook users (9 per cent) – so these findings may not hold for the larger Facebook population. Moreover, given the dynamic interplay between our choices and the algorithm’s recommendations, the assumption that our choices are somehow independent of exposure seems tenuous. Your ‘freedom to choose’ may well be biased by the content that the algorithm selects for you. — Sander van der Linden, Foolproof, 422
Categorization
- XREF: connects to: Social media amplifies rather than causes pre-existing biases in susceptible users; Facebook published its own Science study minimizing algorithm-driven echo chambers — Sander van der Linden, Foolproof: Why Misinformation Infects…, 417; Empirical refutations of political echo chambers center on algorithms, not user choice — Alice E. Marwick, Status Update: Celebrity, Publicity,…, 832 (+4 more)
Broader Connections
- Empirical refutations of political echo chambers center on algorithms, not user choice
#corroborates(0.85) — Alice E. Marwick, Status Update, 832 - Facebook published its own Science study minimizing algorithm-driven echo chambers
#contradicts(0.80) — Sander van der Linden, Foolproof, 417 - Facebook exposure diversity depends primarily on friendship-network structure
#expands(0.80) — Sander van der Linden, Foolproof, 418 - Public criticism failed to halt Facebook’s behavior modification experiments
#contradicts(0.75) — Shoshana Zuboff, The Age of Surveillance Capitalism, 942