Study: 80%+ of the Twitter accounts that spread false info during 2016 election are still active, continue to push more than a million tweets per day
Enough with the “whack-a-mole” claims that as soon as you ban one fake news site, another one pops up: A report released by Knight …
Context & Ripple Effects
The Knight-backed finding lands two weeks after companion research showed [[a:933472|Facebook engagement for identified fake-news sites fell by more than half between the 2016 election and mid-2018]] while Twitter shares kept climbing — making this report the missing half of that comparison: the accounts themselves never left.
It also reframes the 'whack-a-mole' defense platforms used after 2016. Banning individual fake-news domains did not dismantle the distribution layer, because the same Twitter accounts that pushed false information then were still posting at scale two years later.
First-order effects
- Twitter's post-2016 enforcement posture is directly implicated: over 80% of the accounts flagged for spreading false election information remained active and continued emitting more than a million tweets per day, meaning site-level takedowns left the human amplification network intact.
Second-order effects
- The persistence explains why later studies keep finding concentrated spreaders rather than diffuse botnets — consistent with earlier work showing humans, not bots, primarily drive the faster and farther spread of inaccurate news, and with the finding that a tiny user fraction accounted for most fake-news exposure.
Third-order effects
- If the pattern holds across elections — as the claim that roughly 2,100 voters drove 80% of 2020 fake-news sharing suggests — platform accountability shifts from policing content to policing persistent amplifier identities, pushing toward verification and proof-of-personhood mechanisms that Twitter's broad-expression moderation choices have resisted.
The trend: Misinformation enforcement is diverging by platform: Facebook suppressed fake-news publishers' reach while Twitter's amplifier accounts persisted, making account-level identity the next battleground.