The German foreign ministry, which commissioned the study after suspecting it was being targeted by bots, said the findings highlighted the need for governments to systematically tackle the growing number of disinformation campaigns and recognise the effect they could have on elections.
St. Petersburg is just two hours earlier in winter and one hour earlier in sommer bc. of summer time.
It is very difficult to acertain a single user to be a “bot” either as a true machine program or as a paid troll. By those metrics you can observe larger efforts. E.g. is the spread of time windows of certain accounts, which write for a specific point and argument significantly different from the overall users that engage with this kind of topic?
Is there a specific pattern how many accounts interact with specific topics, e.g. are they always “first on the scene”?
But for an individual account it is quite difficult to identify. Could be that it is just one person getting up early. Could be that this person loves to tweet over his morning coffee.
I can highly recommend this presentation on The Rise and Fall of “Social Bot” research where the presentator concluded most metrics to be used in research until then to be arbitrary and giving many examples of real users that were considered as bots by those poor metrics. It is from the end of 2021, so i assume the research has improved in the past 2 years.
The key takeway remains though. There is no simple way to identify individual accounts as “bots”.
Time zone is just one indicator that doesn’t says much by itself. However if you have a handful of indicators, it becomes easier to positively identify bots.
Fraud detection for online shops works in a similar way, where they check your location, IP, delivery address and other metrics to assign a risk score to each order.
