Moderation and ranking decisions are discretionary and made in private

The argument

Proponents of this position argue that bias follows from how the platforms are built. Every feed is ordered by rules that decide what is shown first, what is shown rarely and what is removed, and those rules are written by companies, applied by a mixture of staff and automated systems, and not published in any form that would allow an outsider to check them. Where enormous discretion is exercised without disclosure, proponents argue, the question is not whether it is exercised evenly but whether anyone could tell. The argument is that the opportunities for slant are structural rather than exceptional. A judgement must be made about what counts as harassment, misinformation, hate speech or coordinated manipulation; where the line falls is contestable, and the people drawing it share particular backgrounds, locations and political assumptions. Downranking is especially consequential, since content that is quietly shown to fewer people produces no notice, no appeal and no record. Proponents note that platforms have repeatedly changed these rules under public and governmental pressure, which concedes that the rules are choices rather than neutral mechanics. Proponents add that the companies have interests of their own, which discretion lets them serve. Platforms face regulation, taxation and antitrust action in every country they operate in, employ large lobbying operations, and have owners with political commitments they do not conceal. An institution with the power to set what hundreds of millions of people see, and with its own stake in political outcomes, has both the means and the motive to favour some viewpoints. Because ranking and moderation are discretionary judgements made privately by interested parties and are not auditable from outside, this argument concludes, there is a social media bias.

Premises

Counter-arguments

Opponents of the bias claim reply that discretion exercised in private is not evidence of partisanship, and that the platforms' commercial incentive runs the other way: suppressing a widely held view would forfeit the engagement of everyone holding it to a competitor. They argue the ranking actually observed tracks what provokes reaction rather than any faction's preferences, which is why groups across the political spectrum simultaneously report being the ones suppressed.

Rejecting the premises

[Rejecting P1] Undisclosed rules and discretionary amplification are not in dispute. The premise establishes that platforms could favour some viewpoints, while the conclusion requires that they do, which is a different claim calling for evidence of the direction and the pattern of the effect. [Rejecting P2] That the judgements are contestable and made by people holding particular assumptions describes every editorial process, newspapers and broadcasters included. Without a measured skew it supports the possibility of bias rather than its presence. [Rejecting P3] Holding commercial and political interests, and holding the means to serve them, does not show the means were used. The premise supplies motive and capability, and the inference to systematic favouring treats those as sufficient.