Good luck with your investigation, I guess.Carthage is a stand-in for China. Historically, the quote is known as the original Thucydides Trap.
Good boy looking that up on Wikipedia, you did a good job with cut and paste. Did you have to use a mouse or did you do all that with your own thumb? Either way, you sound smart, have a cookie.
I've given up talking about Blacklist Universe videos here on the Farms, there's an instinct over evidence crowd in this forum that doesn't want to be convinced. Not sure what more could be said to get anyone to seriously consider the evidence. Us BLU Watchers moved the discussion over to Whatsapp, I'll check back in from time to time with what we learn there.
The last thing to leave everyone with: anyone notice this thing called Mike S. Miller split off into multiple YouTube channels once we started discussing his identity? Happened a couple weeks after the last post. One channel where the content is focused on religion, the other where the content is focused on (empty) commercial ventures. And the religion channel is 'Unapologetic' - which seemed to me like a tell.
Something about the language bothered us, so we decided to perform stylometric analysis of transcripts from the videos using R-stylo and python. To do this, we created a corpus of transcripts using 50 BLU videos along with 50 others from YBZ, Doug Tenapel, TUG and YF. All transcripts are single-speaker (no interviews), under 2 hours in length, and everything is in English. Here's a short run down of the data in the corpus. For reference, word tokens are unique combinations of words that appear often and in close proximity.
View attachment 1203402
So this dataset is intended to be used a certain way. We work with the assumption that spontaneous speech patterns should map back directly to the speaker and that certain phrases that appear in a lot of videos - like "Comicsgate Creators" - are present with every author and won't unduly influence the result set. This approach has limited applicability with video transcripts, it doesn't pick up on non-verbal cues and mistranslations are a thing. Despite those defects, it's possible to get good indicators that suggest relationships.
With that in mind, we took the transcript from a recent Unapologetic Mike video and compared it with the corpus. A copy of the transcript is attached, you can see the actual language used. Using the full transcript, we got these results:
View attachment 1203441
Cred means credibility, how certain we are that this is the same author. Conf means confidence, how confident are we with the score. This initial pass is pretty raw, while the UM transcript does map to Blacklist Universe - it's also a close match for Yellow Flash. More importantly, stylo doesn't have a lot of confidence in its top choice. After some trial and error, we massaged the data, taking out the first 5 minutes of the UM transcript to get these results.
View attachment 1203454
When we remove the first 5 minutes of the UM video, the transcript statistically identifies as being authored by Doug Tenapel. Blacklist Universe is a close second. Both with a high degree of confidence. Remember, we're looking at word clustering, not content length, frequency or repetition. This means there are identifiable patterns of speech being repeated in the UM video that maps directly back to both authors, words and phrases they repeat all the time.
We can refine the script to look at lexical markers and I'm certain this will identify more closely with the BLU transcripts. However, even with that, the BLU Watcher group does believe this analysis is significant and merits additional scrutiny. We're still trying to answer the question what we are looking at and what additional tools can be brought to bear in search of an explanation.
We've ruled out the con-man hypothesis, which I never accepted in the first place. There's too much organization to the efforts of the person we're seeing in the videos, I don't believe this is just some guy running fake crowdfunding campaigns to steal people's money. He puts in too much effort for too little return, there's some non-monetary purpose behind the actions we see that can't easily be explained by ego or reputation.
I find the performance art hypothesis less compelling given the evidence of direct coordination between Doug and the thing we call Mike S. Miller. It's not just that they are repeating the same phrases often enough to achieve statistical relevancy, it's that no one else is. The tight correlation between BLU and Tennapel would be easier to explain if they were occasionally hitting on topics everyone else is talking about and we would see that in the results above. Stylometric analysis reveals they are basically developing a language internal to their videos that's well-enough defined to suggest that (in terms of the words used) common authorship exists.
So where does that leave us? I have my theories but there's still too many unknowns. The BLM Watcher group is expanding the size of the corpus and developing blind / null tests to refine these results. We're doing some interesting things with NLP and stillframe object identification that should allow us to identify common themes and maybe automatic identification of the people appearing in the videos. We're starting to think about data visualization and how to best represent what's going on here, we're thinking there needs to be a timeline of topics that demonstrates when the person in the videos is talking about what, and how often.
All I really know is there are very good reasons to believe the thing we call Mike S. Miller is not an actual person, more like a synthetic personality broadcasting on YouTube for some arcane purpose. I want to know for sure what I'm looking at and hope better tools and data can reveal the truth.
Keep us posted.