Science Kiwifarms cited as source of "Coordinated Hate Attacks" directed at YouTube videos - "You Know What to Do"

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
upload_2018-6-4_0-28-43.png


“You Know What to Do”: Proactive Detection of YouTube Videos Targeted by Coordinated Hate Attacks

upload_2018-6-4_0-28-58.png


upload_2018-6-4_0-32-58.png


"Also note that /pol/, though a very good example of a tightknit community used to coordinate and disrupt other social groups, is not the only community responsible for performing raids against YouTube videos. Other Web communities, e.g,. Reddit [54] or Kiwi Farms [20] also regularly take part in raiding activity"

As part of future work, we plan to use deep-learning methods to fuse audio, video, and metadata into a single classifier. We also plan to look into raids from other communities, such as Reddit, Gab.ai, and Kiwi Farms.

Finally, it might perhaps be tempting to dismiss the relatively low occurrence of raids, vis-a-vis the number of ` YouTube videos posted every day, as being a niche problem. On the contrary, harassment and bullying on YouTube are widely recognized as a serious issue by authorities on the matter [47], and news reports are filled with ghastly stories [50] and advice on how to deal with hateful and harassing YouTube comments in particular [15, 26].

https://arxiv.org/abs/1805.08168
http://archive.fo/Z3rXW

https://arxiv.org/pdf/1805.08168.pdf
 

Attachments

Raids? What is this, Something Awful circa 2003?

I thought this forum was an observation-and-commentary one, not one where we get our hands dirty. That's for the lolcows to do to themselves.

Yet another research paper that throws this site in there as a source without actually knowing what it does.
 
The closest thing to raid I could imagine was a few people making post back in the People Celebrate Cancer Death thread. Since this refers to YT videos, I can only imagine they'd references either that one channel teaching LGBT queer stuff to kids or that one black guy that told people not to go on Kiwi Farms.

Raids? What is this, Something Awful circa 2003?

I thought this forum was an observation-and-commentary one, not one where we get our hands dirty. That's for the lolcows to do to themselves.
Either they think we must be doing raids or they assume observation and commentary is the same as a raid.
 
Will the infamous hacker 4chin ally himself with the dangerous new zeland kiwi farms doxxer?
Find out in the next episode of reasons why the MSM needs to die

Edit: oh wait its an useless paper from the university of london, wake me up when someone less retarded starts blaming us for terrorism
 
Either they think we must be doing raids or they assume observation and commentary is the same as a raid.

Since when is expressing an opinion outside of the zeitgeist the same thing as committing hate speech? That makes no sense at all.

This Cornell individual is using the status of their school's name to look down on anything they label deplorable in an attempt to control the narrative.
 
Since when is expressing an opinion outside of the zeitgeist the same thing as committing hate speech? That makes no sense at all.

This Cornell individual is using the status of their school's name to look down on anything they label deplorable in an attempt to control the narrative.
Either they think we must be doing raids or they assume observation and commentary is the same as a raid.

You're both right.
 
It couldn't possibly be that the people who get featured here are giant flailing idiots who attract negative attention to themselves all on their very own, through their own negative, and often anti-social, behaviors.
:thinking:
No, it must be people calling in coordinated attacks. And since it's simply not possible to be a member of two websites at once, YouTube must remove all bad opinions, because they come from fake YouTube members, not true and honest accounts.
 
It couldn't possibly be that the people who get featured here are giant flailing idiots who attract negative attention to themselves all on their very own, through their own negative, and often anti-social, behaviors.
:thinking:
No, it must be people calling in coordinated attacks. And since it's simply not possible to be a member of two websites at once, YouTube must remove all bad opinions, because they come from fake YouTube members, not true and honest accounts.

Well to be clear on one thing which some people seem to be missing, the point of the paper isn't so much about at what "causes" attacks but rather building an algorithm/system to detect videos which are likely to be "raided":
upload_2018-6-4_1-9-56.png

upload_2018-6-4_1-10-10.png

upload_2018-6-4_1-10-50.png

upload_2018-6-4_1-10-22.png



I'd have to read the entire thing to fully understand it, but just wanted to make the point about what the paper is primarily about.

However its still funny that they call KF a source of these "raids":
upload_2018-6-4_1-12-7.png


One thing I'm not clear on is whether the source of the attacks is actually important/relevant to their research or if they purely went through YouTube videos and used their own classification of what they consider to be a "raid".

If "raids" aren't caused by external parties then they're not actually raids and all this machine learning system does is detect videos which are likely to trigger autism on the internet. If that's the case then it still may be impressive (if it actually even exists beyond being just a research paper).

(again I'd have to read the entire thing to speak more on it, but I'm just throwing my thoughts out there)

EDIT:
The usage of keywords on videos is interesting and I did speed read past a part which mentioned using auto captioning services. This is actually interesting as its something
I experimented with on DarksydePhil's streams. He usually rambles on for like a fucking solid hour every day, so I started using automatically generated subtitle files to determine what shit he was talking about each day... something like this:
upload_2018-3-14_13-36-19-png.402349


Based upon that you could tell when he was sperging about particular topics, so it stands to reason that by identifying the content of a video itself or even the outward-facing content (just the title/thumbnail/keywords) you could recognize videos which are likely to experience bouts of autism.

Google already uses similar concepts to identify videos which would be problematic for advertisers.
 
Last edited:
As part of future work, we plan to use deep-learning methods to fuse audio, video, and metadata into a single classifier. We also plan to look into raids from other communities, such as Reddit, Gab.ai, and Kiwi Farms.

Deep learni- bitch, see those bulletpoints the beginning of every board?

"No trolling plans. We are not an autistic Illuminati. If you embarrass yourself or the forum trying to troll, we will ridicule you."
English! Motherfucker! Do you read it!?

Will the infamous hacker 4chin ally himself with the dangerous new zeland kiwi farms doxxer?
Find out in the next episode of reasons why the MSM needs to die

Edit: oh wait its an useless paper from the university of london, wake me up when someone less exceptional starts blaming us for terrorism

Reminds me of David Wong's referring to gamergate and/or anonymous as an actual person.
 
Back