Figure 3

Ratio of layer sizes differentiates fake news from real news. The distribution of the ratio of layer sizes and its development after a period of time can differentiate fake news from real news. These differences appear already after a few hours. (A) The PDF of all re-postings in the first five layers averaged over all of the Weibo propagation networks. The p-value of Mann–Whitney is below 0.01. (B) The PDF of all re-postings in the first five layers in all networks of Twitter. The p-value of Mann–Whitney is below 0.01. (C) Distribution of the ratio of layer sizes at five hours from the first re-posting. The ratio of layer sizes is the size of the second layer divided by the size of the first layer. The p-value of Mann–Whitney between fake and real news is below 0.01. This figure considers all 1701 fake news, all 492 real news and 51 real news with non-official creators at five hours from the first re-posting. (D) Distribution of the ratio of layer sizes of all re-postings for the whole lifespan. The p-value between fake and real news is below 0.01. Here we consider all available Weibo propagation networks (all 1701 fake news, all 492 real news and 51 real news with non-official creators)