Of the period 2015-2017: > We find that trolling fell 35% on Russian holidays and to a lesser extent, when temperatures were cold in St. Petersburg. Of 2018 a

Of the period 2015-2017: > We find that trolling fell 35% on Russian holidays and to a lesser extent, when temperatures were cold in St. Petersburg. Of 2018 and after: > More recent trolls released by Twitter do not show any systematic relationship to holidays and temperature, although substantially fewer of these that have been made public to date. Full text: https://www.nber.org/system/files/working_papers/w28035/w28035.pdf https://www.nber.org/system/files/working_papers/w28035/w280...