Number of nodes in Twitter reply subnetworks.

<p>(Left) The quantity <i>N</i>_repliers is shown for Weeks 1 to 10, where each data point (dot) represents the average over 100 simulated subsampling experiments. The dashed line represents the best fitting model of the form <i>N</i>_repliers = <i>ax^b</i> to the observed data. We extrapolate this model to predict <i>N</i>_repliers. (Right) The same as panel, except for <i>N</i>_receivers.