How far it actually travelled

A subject the papers are about. The loosest grouping, and the one to reach for last.

How false claims move, and how much of the traffic they are. Allcott's prevalence figures are deflationary; Del Vicario's answer is that the average is the wrong statistic, because diffusion is governed by homogeneity inside clusters rather than reach across a population.

That disagreement is methodological rather than political, which is what makes it worth holding. One side measures how much of a population's news diet false stories actually were, and finds a small number. The other says a small population average is compatible with intense local circulation, because sharing happens inside homogeneous clusters where the relevant denominator is the cluster and not the country. Neither refutes the other, and a reader who takes only the first away has been misled by a true statistic.

Chew and Eysenbach sit before the vocabulary existed -- H1N1 tweets in 2010, analysed as content rather than as misinformation -- and are useful for exactly that reason: the phenomenon was being measured before 2016 supplied the word that made it a subject. Mills's SPIN framework is the attempt to say what makes something spread at all, independent of whether it is true.

The set's weakness is worth stating plainly: the review and agenda pieces outnumber the measurements. Lazer's *Science* paper calls for a research programme, Broda draws interdisciplinary lessons, and Wang maps the literature scientometrically -- three records about the state of the field against a smaller number that add evidence to it. A reader wanting to know what is known should start with Allcott and Del Vicario and treat the rest as orientation.

Allcott is held twice; this set holds `allcott2017socialb`, and the two records are linked.