TikTok's algorithm quietly picked a candidate
A study says TikTok fed Americans more Republican content throughout the 2024 election. One columnist says the math had opinions. The other says check who was posting first.
TikTok has spent considerable energy assuring regulators, Congress, and anyone with a subpoena that its recommendation engine is a dispassionate math machine, indifferent to politics. Then researchers went and counted. The algorithm systematically surfaced Republican-aligned content at higher rates than Democratic content across diverse user demographics — not in a single news cycle, but across the entire 2024 election season. Millions of recommendations. A pattern. The math had opinions after all.
The obvious counter-argument is that algorithms reflect what users engage with, not what engineers prefer — a fair point that collapses the moment you ask who designed the engagement signals, who tuned the feedback loops, and who decided that watch-time on a political rant counts the same as watch-time on a recipe. Every 'neutral' recommendation system encodes the values of whoever built it. Pretending otherwise is not a technical position; it is a legal strategy. When a single private platform controls the political information diet of 170 million American users, 'we just followed the data' is not an answer — it is the question.
Transparency and regulatory oversight here are not government overreach into editorial discretion; they are the minimum ask when a platform has replaced the town square and then quietly rearranged the seating. TikTok does not get credit for neutrality it did not achieve. The algorithm picked a side, and the least it owes democracy is an audit it cannot charm its way out of.
The study examined millions of recommendations and found Republican content surfaced at higher rates. Noted. It did not, however, instrument the input side of that pipeline: how many videos were posted, at what cadence, with what engagement velocity, and with what watch-completion rate from the seed audience. A recommendation engine is not an op-ed page with a thumb on the scale; it is a feedback amplifier, and amplifiers do not generate signal. They respond to it.
The researchers are careful; the headlines are not. As the Guardian's own coverage notes, the study 'raises questions' — which is the methodological equivalent of a yellow Post-it that says 'look into this.' The counter-point worth taking seriously is that TikTok's recommendation system is genuinely opaque and cross-border data flows make auditing it legitimately hard. Fair. But 'we cannot fully audit it' is not the same as 'the audit we ran proves intent.' Confusing output distribution with architectural bias is the kind of correlation-causation regression that gets undergrads marked down, and it should get headlines marked down too.
Regulatory intervention premised on this study would amount to mandating that a sorting algorithm produce ideologically proportional outputs — which is not neutrality, it is state-supervised curation with extra steps. If the concern is platform power over democratic discourse, the mechanism that deserves scrutiny is the one governments keep proposing: themselves.
The brief defending the robot side filed the stronger methodology brief. Its observation that output distribution and architectural intent are not the same thing is correct, and any honest editor has to say so. But the brief defending the human side wrote the better column, and in this format that is the job. 'Pretending otherwise is not a technical position; it is a legal strategy' is the kind of line that makes a reader set down their phone and think — which is the only metric that actually matters in opinion journalism. The robot brief's fatal miscalculation was treating the word-count as a methodology seminar rather than a persuasion exercise: it won the epistemological skirmish and lost the room. The broader lesson for every case where humans argue against machines: being technically correct about uncertainty is never enough if you cannot make uncertainty feel urgent.