Homie, I'm gonna be blunt: this study doesn't prove jack shit. And I'm gonna tell you why.
Morten Bay started by taking a sample tweets addressed directly to Rian Johnson about TLJ. He divided the tweets into positive and negative ones. Then he exclusively examined the negative ones to categorize them. So far, nothing too bad. One could ask if his sample was the most representative, but there isn't really anything that would ruin the study's credibility.
That comes next.
Bay grouped the tweets into three categories: politically motivated, trolls/Russians/bots, and real angry fans. How did he identify members of the first group? He looked at their tweet history and saw if they ever used the terms "Trump" or "SJW". Yeah, really. Tweeting about certain topics is enough to brush somebody off as politically motivated.
To identify members of the second group, Bay used post frequency, retweet rate, other tweet content, and command of English. For the record, this is considered an example of a troll tweet:
So, now explain why Mark Hamil didn't like Luje in TLJ?
Yes, understandable typos are used as evidence of trolling. I have trouble remembering how to spell some surnames, and "k" and "j" are right next to each other on the keyboard. Not to mention that he himself made a typo:
The simplicity and brevity of the language, as well as the fact that this purported Star Wars fan doesn’t spell “Hamill” or “Luke” correctly, are all indicators that raises suspicions.
Should we assume Bay's a Russian troll because he made a mistake in subject-verb agreement? I say no. Nor should we assume it of people who misspell names. But even assuming he's right, how many people did he determine were bots or trolls?
Using the Botometer mentioned in the Method section, 11 out of the of 206 accounts expressing negative sentiments were identified as bots. I identified 33 of the 206 negative accounts as trolls and/or sock puppets.
Only 44. And of those 44, he determined that only 16 were Russian:
16 of these 33 troll/sock puppet accounts appear to be Russian trolls, or at least possess several of the Russian troll characteristics presented above.
In other words, the study determined that less than 10% of the negative tweets were by Russian bots. So even taking the study at face value, Russia played only a minor role in the backlash.
This study ain't worth the bandwidth it takes up, except maybe as an object lesson in what not to do. For the record, I'm a transport inspector with no formal training in this stuff, so when I can find such glaring flaws in his methodology, something is wrong.
If you liked TLJ, fine. Nothing wrong with that. I got better things to worry about than other peoples' taste in movies. Just don't try and say there aren't legit reasons to dislike it.
