- Joined
- May 2, 2021
FOREWORD
I have spent hours running various texts from the Alice leaks and tweets from both the Pantsu and Ethan twitter accounts through a plagiarism detection algorithm called 'ghostdetect' and am confident I have found solid proof of socking. I cannot go through every single text in the Alice leak and do a full comparison but I encourage all interested users to go to ghostdetect.com and start comparing texts and tweets for themselves, what is most interesting is just how difficult it is to find May tweets that aren't an exact match to Ralph. Note, living with someone and starting to emulate their speech superficially would not create these results.
INTRO AND EXPLAINING SHIT
Everyone's aware of the suspicion many high level A-logs have that Ralph regularly sockpuppets Pantsu to post on her twitter and text using her phone. For a lot of cases this is obvious upon surface level analysis but impossible to fully prove. There are tools however for detecting plagiarism and ghostwriting in academia, typically plagiarism doesnt just take the form of someone copy and pasting something wholesale from someone else, "Ok bro, you can copy it just change the words so the teacher cant tell" is the typical case. In situations like this though the copying can be detected using algorithms that pull behavioral patterns from the text and quantify them as multivariate data that can be compared so that even if a person has changed up the text they copied the base writing patterns of the original author can be detected and the copier exposed. Ghostdetect is a rudimentary program that performs this operation, it's not academic grade but it will do for analyzing small tweets and texts and comparing them between just two people.
The weird faces you'll see attached to each of the ghostdetect tests are Chernoff Faces or Flury-Riedwyl Faces, they exist to make analyzing multivariate data feasible for a human being. Basically the raw data compiled from each writing sample that gets lined up for comparison would be incredibly difficult for a person to look at and compare and Chernoff Faces just convert patterns in the data into variables attached to the facial features of one of these Chernoff Faces. In academia if you were searching for plagiarism you'd use a computer to interpret the data to detect similarities but for the small text samples ghostdetect analyzes, Chernoff Faces are a very convenient tool for analyzing the similarities and lackthereof. (Yes this means that the autistically facial blind amongst us can't tell what they're looking at so you'll just have to take my word for it if I say faces look similar.
tl;dr , ghostdetect breaks down samples of text and converts certain patterns in the writing into data that is represented with a Chernoff Face algorithm and it lets us compare if two posts were written by the same person. This cannot be easily disguised by changing up words used, adding or subtracting punctuation.
REAL SHIT
First I'll show a couple examples of what two unique authors look like in ghostdetect, again it is important to note that the Chernoff Faces are an algorithm representing emergent patterns in one's writing and the two generated serve to show just how similar or non-similar the samples are. We are not looking for perfectly alligned exact copies when there is socking just strong similarities. The important thing to note though is that when there are two unique authors with no chance of socking the faces generated by ghostdetect will be completely distinct, for instance here's an excerpt from Ethan Ralph's magnum opus The Smoke compared to an excerpt of that pedophile piece of shit broken dick faggot Joshua Connor Moon's zero hedge article regarding the Melinda Scott trial on the left and on the right is a randomly selected tweet from Metokur compared to a Ralph tweet.
The faces generated all look distinct from each other, a negative result. Which stands to reason as each of these people will have different typing styles. It's important to again note that ghostdetect is not just looking at punctuation and word usage it is compiling data from various tiny habits some of which are imperceptible to the average person just looking at the text.
To show what a positive result looks like here's Josh's previous excerpt compared to another excerpt from elsewhere in that same zerohedge article.
So now that we have a good basis to work from we can look at the texts and tweets that are supposedly from May and the results are incredibly amusing.
May and Alice texts
Again I wont be painstakingly posting comparisons of every single text sent by May to Alice to Ralph's texts as that would make this post obnoxiously long and full of images, instead I would like to encourage interested autists to open up ghostdetect and start comparing texts for themselves and post their findings here. There is a treasure trove to search through on this board.
It is also important to note that some of the texts between the two are legitimately May and from the many I tested the ones that were Ralph were usually the groomy ones trying to get her to come over and have unholy sexual communion with the couple. It could be interesting to compile a list of which texts are and are not actually May. By far the most hilarious part of this to me is the fact that Ethan wrote a romantic fanfiction story about himself and May while skinwalking as her to try and seduce Alice. An instance I have included below. His texts vary in pattern sometimes as unlike his twitter he isnt always drunk while typing these but baseline patterns are consistent throughout, note nose shape, nose width, mouth, jaw and hairstyle in the Chernoff Faces.
Tweets
This is where things get very interesting, it is much harder to disguise your typing patterns on twitter posts with it's restrictions but also a bit easier to avoid your usual patterns if you tend to be long winded and are shortening down your usual sentence length. This doesnt really matter though as the more I tested the more I realised Ethan's pattern is so insanely distinct there's zero differences in some of his sock tweets on May's account. If you need refference again for what a negative test looks like at the start of this post I compared a Metokur tweet to a Ralph tweet to show what to expect.
First off it is very fucking interesting to point out May's tweets from last year come up in ghostdetect has having uniquely different authors to her old tweets.
This is already fucking weird but when you line up comparisons of her tweets using tweets from Ralph's account and cross reference them to tweets that we can confirm are May it becomes even more jarring. I have two comparisons lined up below and I need to note once more that it took a pretty fucking long time searching through May's tweets to find tweets that were not a direct match to Ralph. There are more tweets on her account that are a match to Ralph than not a match to him. The most obvious ones are from last year during the height of the Vickers fiasco which are to
no one's surprise all a perfect match to Ralph. I encourage interested users to comb through and find more examples.
IN CONCLUSION
I could go on much longer with even more examples but I think this post is long enough, I think we can say that it is one hundred percent proven beyond a shadow of a doubt that Ethan writes at the very least the vast majority of May's tweets. If any users decide to go on ghostdetect and test more please do post your results here for everyone else to see as well.
In closing, felted also sneed
(edit) :
On sample selection - if youre going to compare writing you need two samples that are somewhat close to the same size or at the very least express something in full. For instance, a tweet of Ralph saying "Make sure to watch the killstream tonight" or "Fuck that cancer faggot Jim" are too generic and simple to be used as samples for comparison in ghostdetect. When usings tweets for comparison make sure they are fully expressing something, threads of tweets usually work best and are not just an off-hand statement or meaningless rage-pigging. And it should go without saying that comparing one sample that is a single, simple sentence to another sample that is a full paragraph is also poor sample selection.
I have spent hours running various texts from the Alice leaks and tweets from both the Pantsu and Ethan twitter accounts through a plagiarism detection algorithm called 'ghostdetect' and am confident I have found solid proof of socking. I cannot go through every single text in the Alice leak and do a full comparison but I encourage all interested users to go to ghostdetect.com and start comparing texts and tweets for themselves, what is most interesting is just how difficult it is to find May tweets that aren't an exact match to Ralph. Note, living with someone and starting to emulate their speech superficially would not create these results.
INTRO AND EXPLAINING SHIT
Everyone's aware of the suspicion many high level A-logs have that Ralph regularly sockpuppets Pantsu to post on her twitter and text using her phone. For a lot of cases this is obvious upon surface level analysis but impossible to fully prove. There are tools however for detecting plagiarism and ghostwriting in academia, typically plagiarism doesnt just take the form of someone copy and pasting something wholesale from someone else, "Ok bro, you can copy it just change the words so the teacher cant tell" is the typical case. In situations like this though the copying can be detected using algorithms that pull behavioral patterns from the text and quantify them as multivariate data that can be compared so that even if a person has changed up the text they copied the base writing patterns of the original author can be detected and the copier exposed. Ghostdetect is a rudimentary program that performs this operation, it's not academic grade but it will do for analyzing small tweets and texts and comparing them between just two people.
The weird faces you'll see attached to each of the ghostdetect tests are Chernoff Faces or Flury-Riedwyl Faces, they exist to make analyzing multivariate data feasible for a human being. Basically the raw data compiled from each writing sample that gets lined up for comparison would be incredibly difficult for a person to look at and compare and Chernoff Faces just convert patterns in the data into variables attached to the facial features of one of these Chernoff Faces. In academia if you were searching for plagiarism you'd use a computer to interpret the data to detect similarities but for the small text samples ghostdetect analyzes, Chernoff Faces are a very convenient tool for analyzing the similarities and lackthereof. (Yes this means that the autistically facial blind amongst us can't tell what they're looking at so you'll just have to take my word for it if I say faces look similar.
tl;dr , ghostdetect breaks down samples of text and converts certain patterns in the writing into data that is represented with a Chernoff Face algorithm and it lets us compare if two posts were written by the same person. This cannot be easily disguised by changing up words used, adding or subtracting punctuation.
REAL SHIT
First I'll show a couple examples of what two unique authors look like in ghostdetect, again it is important to note that the Chernoff Faces are an algorithm representing emergent patterns in one's writing and the two generated serve to show just how similar or non-similar the samples are. We are not looking for perfectly alligned exact copies when there is socking just strong similarities. The important thing to note though is that when there are two unique authors with no chance of socking the faces generated by ghostdetect will be completely distinct, for instance here's an excerpt from Ethan Ralph's magnum opus The Smoke compared to an excerpt of that pedophile piece of shit broken dick faggot Joshua Connor Moon's zero hedge article regarding the Melinda Scott trial on the left and on the right is a randomly selected tweet from Metokur compared to a Ralph tweet.
The faces generated all look distinct from each other, a negative result. Which stands to reason as each of these people will have different typing styles. It's important to again note that ghostdetect is not just looking at punctuation and word usage it is compiling data from various tiny habits some of which are imperceptible to the average person just looking at the text.
To show what a positive result looks like here's Josh's previous excerpt compared to another excerpt from elsewhere in that same zerohedge article.
So now that we have a good basis to work from we can look at the texts and tweets that are supposedly from May and the results are incredibly amusing.
May and Alice texts
Again I wont be painstakingly posting comparisons of every single text sent by May to Alice to Ralph's texts as that would make this post obnoxiously long and full of images, instead I would like to encourage interested autists to open up ghostdetect and start comparing texts for themselves and post their findings here. There is a treasure trove to search through on this board.
It is also important to note that some of the texts between the two are legitimately May and from the many I tested the ones that were Ralph were usually the groomy ones trying to get her to come over and have unholy sexual communion with the couple. It could be interesting to compile a list of which texts are and are not actually May. By far the most hilarious part of this to me is the fact that Ethan wrote a romantic fanfiction story about himself and May while skinwalking as her to try and seduce Alice. An instance I have included below. His texts vary in pattern sometimes as unlike his twitter he isnt always drunk while typing these but baseline patterns are consistent throughout, note nose shape, nose width, mouth, jaw and hairstyle in the Chernoff Faces.
Tweets
This is where things get very interesting, it is much harder to disguise your typing patterns on twitter posts with it's restrictions but also a bit easier to avoid your usual patterns if you tend to be long winded and are shortening down your usual sentence length. This doesnt really matter though as the more I tested the more I realised Ethan's pattern is so insanely distinct there's zero differences in some of his sock tweets on May's account. If you need refference again for what a negative test looks like at the start of this post I compared a Metokur tweet to a Ralph tweet to show what to expect.
First off it is very fucking interesting to point out May's tweets from last year come up in ghostdetect has having uniquely different authors to her old tweets.
This is already fucking weird but when you line up comparisons of her tweets using tweets from Ralph's account and cross reference them to tweets that we can confirm are May it becomes even more jarring. I have two comparisons lined up below and I need to note once more that it took a pretty fucking long time searching through May's tweets to find tweets that were not a direct match to Ralph. There are more tweets on her account that are a match to Ralph than not a match to him. The most obvious ones are from last year during the height of the Vickers fiasco which are to
no one's surprise all a perfect match to Ralph. I encourage interested users to comb through and find more examples.
IN CONCLUSION
I could go on much longer with even more examples but I think this post is long enough, I think we can say that it is one hundred percent proven beyond a shadow of a doubt that Ethan writes at the very least the vast majority of May's tweets. If any users decide to go on ghostdetect and test more please do post your results here for everyone else to see as well.
In closing, felted also sneed
(edit) :
On sample selection - if youre going to compare writing you need two samples that are somewhat close to the same size or at the very least express something in full. For instance, a tweet of Ralph saying "Make sure to watch the killstream tonight" or "Fuck that cancer faggot Jim" are too generic and simple to be used as samples for comparison in ghostdetect. When usings tweets for comparison make sure they are fully expressing something, threads of tweets usually work best and are not just an off-hand statement or meaningless rage-pigging. And it should go without saying that comparing one sample that is a single, simple sentence to another sample that is a full paragraph is also poor sample selection.
Last edited:
