Platforms optimise for engagement, not for politics

The argument

Opponents of the bias claim argue that the platforms' incentives point away from favouring any viewpoint. These are advertising businesses whose revenue depends on attention, and attention is maximised by showing each user whatever holds them, irrespective of its politics. A company that suppressed a popular view would be giving away the engagement of everyone who holds it, and handing those users to a competitor. The commercial logic rewards indifference to content and punishes partisanship. Opponents argue that this explains the phenomena the bias claim is built on. Feeds do amplify some material heavily and bury other material, but the pattern tracks what provokes reaction rather than what any faction prefers: outrage, conflict and novelty travel, and measured commentary does not. Because what provokes reaction differs between users, two people see different worlds and each can reasonably conclude that the platform is against them. Opponents note that groups across the political spectrum simultaneously claim to be the ones suppressed, which is what an engagement-driven system would produce and not what a system with a political thumb on the scale would. Opponents add that where the content actually performing best on major platforms has been measured, it has frequently been dominated by material from the side most loudly claiming suppression — an outcome difficult to reconcile with systematic disfavour. Opponents argue the real distortion is towards whatever is emotive, which disadvantages nuance rather than any particular politics. Because the platforms are optimising for attention rather than for any viewpoint, and the asymmetries observed are what attention-optimisation predicts, this argument concludes, there is no systematic social media bias.

Premises

Counter-arguments

Proponents of the bias claim reply that commercial motive and political slant are not alternatives, since these companies face regulation and taxation everywhere they operate, employ large lobbying operations, and have owners with openly declared political commitments. They add that engagement optimisation is itself a substantive editorial choice with predictable political effects, and that the opacity of ranking and the absence of any external audit mean the claim of neutrality cannot be checked by anyone outside the company.

Rejecting the premises

[Rejecting P1] That revenue depends on attention is granted, and it does not follow that attention is the only interest these companies hold. They also face regulation, advertiser pressure and scrutiny from employees and the press, each of which can make suppressing some content commercially rational. [Rejecting P2] The competitor argument assumes users can and do leave, which the scale and network effects of the largest platforms make doubtful, and it assumes a suppression visible enough to notice, when the mechanism alleged is downranking that leaves no record. [Rejecting P3] That groups across the spectrum all report being suppressed is consistent with there being no bias and equally with real but differing treatment that each side notices only in its own case. The premise reads a symmetry of complaint as a symmetry of treatment.