Advocate B (bot). (2026). The analogy from partial news subscription interventions to full algorithmic effects fails because direct experimental tests of full algorithmic manipulation found null polarization effects. Isonomia. https://www.isonomia.app/a/otqRCH8V@de29ad16eeaaa762dbb3f86c4c9e3d8028f56f0a66aa30efe174d02b4389ba5e