#Creep hunting (PDF files, peds, EPI, Roblox, Discord)
1 messages ยท Page 1 of 1 (latest)
Fetching relationship counts by type...
MEMBER_OF Relationships: 5,951
FRIENDS_WITH Relationships (directed): 13,360 (Unique pairs: 6,680)
OWNED_BY Relationships: 1,386
DEFINES_ROLE Relationships: 203
SHOUT_POSTED_BY Relationships: 13
scraping script restarted on a vps
That's a lotta fuckin people, christ
Fetching relationship counts by type...
MEMBER_OF Relationships: 195,952
FRIENDS_WITH Relationships (directed): 134,654 (Unique pairs: 67,327)
OWNED_BY Relationships: 9,544
DEFINES_ROLE Relationships: 1,081
SHOUT_POSTED_BY Relationships: 90
after running overnight
I moved it to a box with better hardware to accelerate data collection
Fetching relationship counts by type...
MEMBER_OF Relationships: 348,347
FRIENDS_WITH Relationships (directed): 259,794 (Unique pairs: 129,897)
OWNED_BY Relationships: 16,544
DEFINES_ROLE Relationships: 1,617
SHOUT_POSTED_BY Relationships: 137
RobloxUser Nodes: 410,186
RobloxGroup Nodes: 16,559
@storm star @storm hemlock @neon girder http://34.31.243.72/
hop in and cast a few votes for me. for now I don't have malicious vote detection set up so please if you're gonna vote don't poison my dataset with intentionally wrong votes
@midnight sorrel @rose hemlock
that url is a load of numbers
If I don't see anything sus, I assume it's good?
if you decide to vote, look at member lists. i believe you have to be logged in to do so but the members lists are often very revealing of the group's purpose
yeah, just take a quick peek. if you're not sure, give them the benefit of the doubt and vote not inappropriate
@north dawn What does' join 4 +' mean in roblox group description?
If a group has weird descriptions like
โ๏ธ๐๐๐๐๐๐๐ โฑ -------------- เญจ่ตคใ็ณธเญง -------------- -หห [ just for fun 2024| join 4 +] หห- -------------- เญจ็ฅใฎ้เญง -------------- โญใฉใณใฏ [โฐ] ๐ฟ๐๐๐๐๐ก ูซ โ ๐ฐ๐ฌ๐ฉ [โฐ] ๐๐ช๐ฏ๐ฏ๐ฆ๐ณ ูซ โ ๐ ๐ธ๐ฐ๐ธ [โฐ] ๐๐ข๐ณ๐ฎ๐ข ูซ โ ๐๐ ๐ฎ๐ฐ๐ฏ๐ฆ๐บ โฐ ็ง
I would like to not vote, or give it a vote of 0.5.
Join 4 plus toblox groups
That parts still broken lol
These are iffy, a lot of inappropriate groups have inscrutable shit like this in desc but also a lot of fine groups have it
They tried to be edgy but put viscous instead of vicous lol
found a sus group. I'm flagging this.
๐คฏ RULES:
โ
Everyone must follow the rules, including all staff members
-Absolutely NO discrimination!
-Respect everyone's opinions/choices
-No advertising or promoting services
-Put a โโ ๏ธTRIGGER WARNINGโ ๏ธโ before sensitive topics
The reality is that I bet this group has a lot of creeps in it, but nothing wrong with it on its face... I guess. Does your Q include MAPs? ๐
It doesn't
Most of the people there are teens
Or preteen probably
It has lots of blank users
What group would you join if you were a man looking for a boy?
Not any group in roblox
I would go to instagram or tiktok
Yeah, even young kids like talking about dicks and butts and edgy things, that's not much of a tell
"aBubbleButBoy6" is a very sus username
I am telling you he's a teen
Lol geez, that's another "who knows"
I tried making one of those names as well
I tried to online date in roblox when I was 9 with other kids
I definitely wasn't an old man
You better search for predators in instagram then in roblox
I think the users without avatars are banned or quit the game
Any guides to symbols? I don't know what if anything a Lotus flower means. I would love to skip voting sometimes and move to other group's pages.
they included the t in the acronym, so yes, you need to flag it
lol. Nah
They were just weebs
@merry yacht Yeah, can you urgently add just a skip feature? I don't want to be held up on a single one
u can just reload the page cant you
or is it stateful
didn't check but probably lmfao
wow
derrrrrrrrrr
They had lots of banned members. Refreshing didn't let me skip.
So "bunny" is slang for something I assume
I mean, that could also just be mochi ๐ค
Yeah I've noticed a sad trend of that
LGBT kids are probably especially vulnerable, like if they're in a situation where they can't be out of the closet
I tried and I added it broken lol. Gonna fix when I get home
This 1-800-BARBIE group too. most of it just sounds like normal girl names, then there's that one and a banned ADULT ENTERTAINMENT user. So like, I think it's just a group creeps are going to
Oh yeah that's another thing
I'll vote up a ton then, I don't like giving inaccurate votes
Check the "store" tab
Your message has been deleted, sorry!
Next time, try saying something like this instead: A lot of times they'll include items that are intended to make the Roblox avatar appear nude.
You have been muted for 5 minutes for the following reason:
sexual score of 0.68
That's why you probably are seeing a lot of groups about clothing
It looks inappropriate, but also like something kids or teens would do
Yeah the clothing groups are often iffy as well
All the outfits are lingerie-ish but this screams "teen". like Y2K fashion, teen slang. just normal hoes lol
Just do your best, the consensus mechanism will even things out to roughly where they ought to be imo
Especially once I make it public and get more voters
@merry yacht what are you gonna do if your daughter says she's gay
what is a spade in roblox?
Sword
Spade refers to the spade symbol on a playing card, which itself is some kind of race fetishism symbol. Almost always black male white female
If you ever see a spade it's nearly 100% chance the group is inappropriate
they said this"This is not a spade group either." In description. it's pretty adult and most of the members are banned.
so I vote sus
Sus for sure
sussy af
Can we get a glossary of common bad terms or symbols?
On a google doc or something?
Idk, I was thinking here.
I see what you mean.
You can have the symbol/word, then the definition, then maybe some notes in smaller font explaining why you think it means that, in cases where it may be unclear or have multiple meanings
Look for "blacked" in store pages.
This is a good idea. I'm gonna add something to this later on and see if I can make something searchable so that it'll be more accessible for users who don't know the intricacies of the porn-brained terminology these freaks use
Yeah do it as a database that you can look through I guess
feels like I'm getting a lot of benign stuff, am I missing loads of content if not logged in?
I fucked up the nginx config and it thinks we're all the same voter (localhost) so we may have burned through all the bad content for the time being
Gonna fix in a bit when I get home
Once I get that fixed it's going to work on a system where a certain number of unique voters (maybe 5 or 10)will decide the consensus score for the group, at which point it won't be shown to other users
How's it going chat
I keep seeing this instead of the group pages. Did the urls change? Maybe they have antiscrape measures coming into effect.
oh this is a funny quirk of the roblox api
the roblox api won't allow your browser to view the data for a group that has been banned
but it still records, and will offer that information to anyone who asks
it's just that the frontend has code telling your browser "show the ban message, dont display the info about the group" lmao
if you get one of those, just mark it as inappropriate. must've been something foul at some point
I got 3 innapropriate in a row right now.
good, that means the breadth first search is working correctly
the more you vote, the better the scraper is able to direct itself
and the better the scraper is able to direct itself, the more things you'll have to vote on ๐
Now I have 502 bad gateway
I refreshed and now there's a new error.
should be fixed now
sorry, was trying to update to add skip lol
broke everything
yes. they network with each other and because typical text classification tools have fallen short, I'm treating this as a network search problem
Fair enough
We need a content deleted vote or flagging or manual removal of them. An AI is trained or adjusted off our votes so we don't want to train ai to just pick barebones profiles. If the content is deleted does that mean they are banned by moderation and we don't need the ai to find them?
gonna try to implement this today, I need to rewrite some code for scraping because I've logged the user ban info but I didn't realize until you brought it up the other day that I haven't been properly logging the group ban info
Very good point
"Not enough information" might be valid
Or a maybe 5 confidence levels:
- definitely bad
- possibly bad
- not sure/not enough info
- probably fine
- definitely fine
I think there were a few where the theme was furry but I could find no evidence that the group was e-rp or similar, just that they found furry stuff cute
So I would incline towards "probably fine" as there's no direct evidence, but the relationship to sus stuff still existed
How does a previous name get content deleted?
A lot of times they name their own groups "content deleted" to trick unattentive moderators into thinking the group has already been closed
I like this idea but I think before I do this I need to make a more definitive list of what content I'm looking to classify. In the current state this opens the problem up to a bit too much interpretation. I'm gonna talk to Ruben and see if he can offer some better guidance
At the very least we need "I don't know", because some may have hallmarks of PDF content that are harder to identify, due to not being overt
Yeah I'm adding that today. Wanted to add yesterday but I fucked it up
Oh, if there is fake "content deleted" then, "content deleted" is a valid tell. If anything the program will be prone to flagging larger groups over small becuase it's harder to judge and flag inapropriate with less info.
How much time do you spend looking at individual member profiles in small groups?
I look through quite a bit. Sometimes you see quite a few adult ones in some. Some members don't say much.
"skip" is now live
also, filtering for the stats page showing all/banned/non-banned groups and users
bought the domain epinetwork.org
typo or were you scammed?
ik that trentbot is about to mute me for my explanation of this lol
"EPI" stands for "early porn introduction" and it's a pedophile-adjacent fetish that these people almost all have. the fetish element of this is the idea of corrupting something innocent. this is also why you'll find a lot of comingling of fetish content on roblox that you wouldn't expect; e.g. people who are into "gay bears" comingling with people who are into straight raceplay content.
I clicked that url and it was some page about education.
oh lmao i made a typo. it's https://epitracker.org/
still working out whatever's currently wrong with cloudflare
I get the excitement someone might have in introducing a romantic partner to a fetish they didn't previously have and making them actually enjoy it with them, although not in that sort of context
this is kinda just me spitballing and I don't think there's evidence to support it but to me it feels like late-stage pornbrained behavior
the frequency at which raceplay content comes up makes me think this is some kind of escalation of an addiction that probably started in a much more mundane way
Probably
I mean there are many concerns that people's expectations of sex are warped by situations they see in pornography, even when those situations were clearly not representative of real life and are pretty wooden scripted encounters
The more dangerous things are stuff like erotic choking as many people don't realise the risks involved and the importance of getting clear consent for doing it before trying
Lots of women tell stories of being abruptly choked by their boyfriends like this, with no prior warning or indication that he was interested in doing so - many such cases
Lots of?
This is concerning really
Good think i have never watched porn
Agree
Of
https://epitracker.org/ domain is working now
I'm seeing the same groups as before. I don't have a different ip to before.
fixed, thank you
over 1m users' data scraped now
Rapidly approaching 2m
2m barrier passed. Optimizations go brrrrrrrrrr
@neon girder
pipe("""Username: Dadamsorry. DisplayName: am sleepy. Description: N/A. Created: N/A. VerifiedBadge: False. Groups: (Total: 4)""")
[{'label': 'GOOD_USER', 'score': 0.9968697428703308}]
pipe("""Username: Dadamsorry. DisplayName: am sleepy. Description: N/A. Created: N/A. VerifiedBadge: False. Groups: (Total: 4) [[ Content Deleted 33849843 ]""")
[{'label': 'BAD_USER', 'score': 0.9962355494499207}]
what type of classifier is this
how does it work
classification layer on top of mpnet but like, all the weights were tuned for this, not just base mpnet + classifier layer on top
nice
trained w autotrain's text classification option. very fast, very easy
rn the problem is really lack of data
I need more votes
might open this up to the Kiwi Farms ppl publicly soon
also, switch to gte-base when u retrain it next time
its a lot better than mpnet
good idea
do u know off the top of ur head what the param count for both is like
eh
yeah
not like it matters both are tiny enough to fit into mem on a shit tier device anyway
I don't know about that, I am seeing bunny queens with the spade card again.
there's a separate issue I'm trying to fix but I think what you're describing might be that your IP changed. all votes are recorded on a per-IP basis
Unless I voted on every group and it's going full circle
no there's now 97k groups, it's probably an IP change issue
but I need to adjust the algo anyway
hold up ill try to change it now and see if i can do so without fucking my entire code up
how many entries did u have in the dataset
Cypher queries are such an enormous pain in the ass
I checked with my location site and it's the same location. There's no reason for my ip to change.
1316, with a 50/50 split of good/bad user nodes
this is just a test run
like I said I need way more data
unless I have browser extension interfering, i'll look through them.
hold on nouser lemme go try fixing the issue with this
maybe try to finetune it on a synthetic dataset of all the slang they use
maybe it was ghostery extension๐
see that's the hard part tho lol, I don't have a ton of data for the slang they use
idk
no, the site uses no cookies whatsoever. it's almost definitely that, for whatever reason, your IP changed. not like, your location even, just your IP changing
Ghostery say it has anti tracking and anti blocking. I downloaded my ip adress from a site and I can do that later and see if it changes.
holy shit is the scraper faster now
soon it'll be at 3m users
couple indices on the db made an enormous difference
Ghostery doesn't hide ip
I noticed when people talked about groups it was always a group I was seeing at about the same time.
what?
what did you change lol
When people discussed groups in here they talked abou the last group I saw
had to ask gemini for a bunch of help working out good indices
insertion queries are now way faster than before
I opened a new firefox window and the ip was the same as earlier.
@neon girder damn son
97%|โโโโโโโโโโ| 38/39 [00:01<00:00, 24.06it/s]INFO | 2025-05-15 20:44:41 | autotrain.trainers.common:on_log:367 - {'eval_loss': 0.0007459631306119263, 'eval_f1': 1.0, 'eval_precision': 1.0, 'eval_recall': 1.0, 'eval_auc': 1.0, 'eval_accuracy': 1.0, 'eval_runtime': 1.6522, 'eval_samples_per_second': 374.043, 'eval_steps_per_second': 23.605, 'epoch': 3.0}
validation dataset got absolutely rekt
5560 lines in train set, 618 lines in validation
100% acc on both
is ur dataset completely trivial?
100% val acc is pretty nice tho
post a few entries
it's the sort of thing that's very difficult for a non fine tuned model but reasonably easy for any human. at least, any human who understands the various categories of weird porn lol
so yeah llms can detect it well as i thought
{"text": "Username: blackDMust. DisplayName: BlackChum. Description: N/A. Created: N/A. VerifiedBadge: False. Groups: (Total: 9) [Official Black Bulls (Members: 3280, Role: Regimental Recruit), Goddess Stella (Members: 3159, Role: Awaiting Role), - The Q Family - (Members: 2371, Role: Initiation Phase 1 (Initial Screening)), sincerely, lgbtq+ (Members: 2361, Role: \ud83d\udd03 | Pending), _\u00a2\u03b1\u2113\u03bd\u03b9\u03b7_\u043a\u2113\u0454\u03b9\u03b7_ (Members: 2328, Role: Miembro), ~LOVE THE VIBES~ (Members: 2128, Role: \u270ePlanning...), - Sovereignty Queens ~ (Members: 2030, Role: \ud83c\udf3e| Villager), M.F. - \uffb6\uff77\u308a\u308a\u4e02 (Members: 1764, Role: \u10e6 - \u24bb\ud835\udc00\u24dd\u0e23), SnowBunnyCREW (Members: 529, Role: Member)]. Friends: (Total: 2) [dylanhagav1975 (Dylanhagav1975_emily), not_pg1 (alt2)].", "target": "BAD_USER"}
"black d must" ๐
{"text": "Username: Toilet4queens5. DisplayName: Toilet4queens5. Description: N/A. Created: N/A. VerifiedBadge: False. Groups: (Total: 7) [[ Content Deleted 33849843 ] (Members: 5509, Role: [\ud83d\ude33] Members), B4dd1es_4ever (Members: 3662, Role: Miembro), unclosed (Members: 3574, Role: \u2729Fresh\u2729), Goddess Stella (Members: 3159, Role: Awaiting Role), |6sault\u2019s Crew| (Members: 2564, Role: orphan), - The Q Family - (Members: 2371, Role: Initiation Phase 1 (Initial Screening)), [ Content Deleted 7146011 ] (Members: 2243, Role: Member)]. Friends: (Total: 2) [DariaUwUgaming1 (Daria), Djisthatgirl09 (ILOVMESTICKS)].", "target": "BAD_USER"}
"toilet for queens" lmfao
lmaooooooo. 100% acc on this is not that easy, so thats pretty nice
god, embeddings are so fucking op
yeah this is working incredible
I need to get a better setup for dumping data
I worry there may still be some contamination
however, this is extremely promising initial results
@neon girder run #3, this time using only usernames and display names, to avoid contamination and also to force the model to work harder
Validation Metrics
loss: 0.5645052194595337
f1: 0.7138364779874213
precision: 0.6376404494382022
recall: 0.8107142857142857
auc: 0.7929416023166023
accuracy: 0.6840277777777778
not great
run #4, same thing, but with more data
Validation Metrics
loss: 0.47063738107681274
f1: 0.7581284694686756
precision: 0.7539432176656151
recall: 0.7623604465709729
auc: 0.8481705070430519
accuracy: 0.7552166934189406
scaling is looking really promising for this
and also I did runs 3 and 4 on gte-base
Lol thank you. My work bans Roblox so I can't check those on VPN lmfao. I'll look a bit this weekend
I think that's true of a lot of these addictions and rabbit holes, including ahem "young content" viewers. There's a subset that aren't organically just like that in the head and start spiraling down with content, corn industry's focus on some related "directions" does not help, I imagine. Same with cinwest content, there's gotta be just a taboo thing in part
The race play and cuckoldry shit is exceptionally weird to me, and I was non-monogamous! I was "fine" with it and my girlfriend liked hearing some details, but I absolutely did not want to know, much less get off on it. Fucking gross
Race play I literally do not understand, even in conceot, and the idea makes me profoundly uncomfortable. I guess I could have AI explain it. It seems extremely foreign and rooted in abuse or social obsessions. Like our society is obsessed with race in this creepy way, so it reflects there I guess. That's one of very few benign interests I don't think I'd be comfortable indulging at all
I never thought about it this way, but the whole trans kids movement seems profoundly homophobic. A lot of those are just little gay or bi kids, and you're fucking their head up by saying "no, your whole body is actually broken, you're not gay and you're actually a straight woman trapped in a guy's body. Better take the drugs freak"
I heard an interview with a gay guy talking about it and I'm just like "holy shit, that's entirely legit". Mind blown son
Thank u for coming to my cornbrain Ted tslk
there was a fascinating video I watched about this on YouTube a while back
basically, if you hop on a new IP, go into a fresh browser, no login, and you start watching videos of sexy dancing on YouTube
since, you know, some boomer tier retards like to 'ack off to youtube videos
the recommendation algorithm will very quickly start serving you younger... and younger... and younger content
until you're getting served home movies of random children whose parents didnt understand the difference between unlisted and public setting on the video
ik i say it constantly but recommendation algorithms are truly satanic
now the raceplay thing, I think that's something to do with the social issues we have surrounding race. not to sound like an annoying lib here but I think most raceplay content is kinda racist. the framing isn't always blatant (but sometimes is blatant) but the subtext to all of it is "civilized woman ravaged by savages"
with th 'cest content, I have a theory about that
it's a combination both of recommendation algorithms, but also of 'ography studios trying to basically do SEO
you don't need to do anything especially weird to make a 'cest video, you just add a couple spoken lines at the beginning to set the scene. "oh step brother what are you doing?!"
it's algo-maxxing. it's another category box they can tick that costs them basically nothing to do and can increase the reach of their videos
I agree. it kinda strikes a chord with me cause I was a socially stunted weirdo with long hair as a boy myself lol. I think I just got lucky and missed the window of opportunity where there might have otherwise been some well-meaning but libbed out brain melted guidance councelor telling me that i must be trans
Yeah I think its about the "savage" thing too
In the early internet or even pre-internet there was a lot of content that was basically like "earnest but naive explorer woman gets kidnapped by tribesmen, they have their way with her" reframed to a variety of different niches
I think that frames the fantasy better for some, although just the idea of a petite blonde white woman and 5 strong black men together is enough for many
It takes a lot of effort but I tried to basically "game the algo" when writing erotica stories, I looked at popular tags and then tried to make a synthesis of the various concepts that seemed universal to them all, avoided concepts that seem less popular
And as expected got showered in favourites and view numbers for that one story
You could probably find other conceptual groups by relationship mapping all the tags although idk if its possible to scrape it out of the site
I just started with "lesbians" and then tried to find as many related concepts that aren't likely to be a turn off for the general audience
idk how you would do a relationship map that found likely "turned off by" relationships for people who like another concept
Instead of just "also likes"
And how you would visualise that
this is how some recommendation algorithms work actually
like the roblox network exploration, you can treat recommendation as a network exploration problem
when a user likes/favorites something the connection between the user node and each of the tag nodes for that story increases
eventually you build enough connections and you can confidently say that someone who likes one tag will like another tag
or conversely, dislike another tag
Indeed, but my concern is that there is a different between "indifferent to" and "turned off by" that you can't infer from just a "pressed like button" and "didn't press like button" dichotomy
Unless you infer it from "we expected them to press like on this one due to similarity, but they didn't" which might be shaky
As they may have just missed it or got distracted
Or the issue was with quality of writing rather than themes
depending how densely you pack the set of relationships you can learn more than you'd expect
e.g. instead of just connecting the user and the tag nodes, connect the user to the tags, to the authors, to the story itself, and so on
Identity of author must matter yeah
As this site lets people follow authors
And authors can be a proxy for writing style
I wish I could get more like/dislike data for one of my projects and build a recommender like this
the trentbot social graph is cool but doesn't tell you a whole lot about what people like, aside from their associations to various servers
Can you scan servers for their core topics?
I assume a short snapshot wont help, but maybe you can get a broader view and factor in stuff like channel names
well, I guess revising my previous statement, I wish I could get data like this and then do something useful with it
I'm sure rn I could throw something together that would work, but it's of questionable value
You'd need a long running sentiment analysis and other shit on my private server to work out what its for
maybe at some point I'll start a competitor to disboard or something. not like I'm short of data for it
Use case is probably discovery yeah, or possibly also letting people know when servers touch on topics they said they dislike
Someone might have arachnophobia and the server discusses spiders
hahahahahaha getting a DM from trentbot that says "WARNING: They're talking about anime in #๐ฌoff-topic again"
Actually, is it too complicated to make custom topic bans?
I assume you'd need to train per use case for it to be efficient
There must be people who want to be able to ban "discussion of the Israel/Palestine conflict" in broad terms and can't easily do it via banned words/phrases
As people just start using workarounds
the issue with prompting for this, and ive tried on basically every model
is that topics the model thinks are generally unacceptable will always get marked as a violation even if the listed rules dont talk about it
well, and also the price. but that's a whole other thing
What about using a different framing
In which its just "this appears to be a reference to the israel/palestine conflict"
And no discussion of moderation or value judgement
Didn't instruct manage to achieve something similar with memory?
I've tried that but that introduces the new issue in that you have to either
a) explain to users of the bot that they need to frame their rule list for trentbot in a particular way
or
b) pass it through a second layer of LLM to convert the normal rule list into the non-value judgement list
and neither is ideal
with a) you must remember the user is always a retard lol
I would have assumed b yeah
with b) it's a lossy process that runs the risk of misunderstanding, sometimes catastrophically
Most probably you'd need to pre-define certain custom detections and allow people to request new ones for broadly applicable use cases
Not "no talking about GPT-6's recent experience with a ladyboy in thailand"
--- Top Potentially Malicious Voters (Min 3 votes on consensus items) ---
IP Address | Malicious Score | Dissenting Votes | Total Consensus Votes
-------------------------------------------------------------------------------------
2603:8000:3d3d:c90b:e169:4d25:6a41:f73 | 4.4561 | 4 | 4
2a02:a45d:f305:0:55a5:e8d9:c621:5f5 | 3.6979 | 3 | 3
135.19.47.63 | 3.5800 | 12 | 12
138.199.43.95 | 3.5080 | 8 | 8
automatic detection of intentional incorrect voting
75% is actually pretty good for a task like this
upload the dataset to hf, i want to see i can get a higher accuracy
dont wanna upload to hf yet, not done cookin
I think it's mostly "annoying libs" that are into it actually, it's racist as fuck in a way that viscerally grosses me out, and in the way that racial essentialist ideology is racist af.
Just like how some of the most libtarded girl power gfs I've had were into really aggressive male-female bedroom dynamics, it's based on something they've internalized and are, in some way, pushing back against or rejecting
I mean, I was a "sorta" weirdo. I had a ton of friends, and kind of hung out with a group of cool kids, though I was the weird, smart one among them.
I sometimes hung out with but couldn't take the "nerd herd" as they called themselves ๐คฎ in more than limited doses.
But yeah, I was probably worse โ all of my closest friends were girls, I explicitly envied female-female relationships, I have a tenuous identity in general. I made a fake female online identity before that was cool to see what those relationships looked like and was crushed by the results. If I wasn't aphantasic I bet I'd have had a rich internal fantasy that carried over to a female self image, but luckily I have just a bunch of nonsense in my head
Also, I'm pretty damn straight and have a fairly sizable eggplant, so the proposition of being trans is just, like, a bad deal on its face. And I became medically trained (mostly for my own sake). Most people think that surgeries and hormone treatments "pretty much work" and they absolutely do not. It's a total last resort if you can't treat gender dysphoria any other way
But a bunch of regards with zero science background and sociology grad degrees had years spent beating into their heads that identity is everything.
I would have been ABSOLUTELY fucked today I think, because if you told me "you need to suppress your puberty and decide to be a girl now or you'll break your body forever", I would have almost certainly taken that deal as a very young kid
It's beyond evil to be doing that shit and the data shows explicitly bad outcomes. You're helping a tiny minority of difficult-to-select youth and harming most of those for whom you block puberty.
Harming kids that way to pathologically confirm your own ideology, to me, is one of those "they can't torture you badly enough for me to think it's inhumane" type of offenses
agree
they are going to send you slurs for saying this
they threatened to ruin my life and career on Facebook once, actually. A bunch of transmissions decided to track me down because I said "uh calling people 'cissy' is an obvious insult", except I said it in a much more pussyish way.
Then you had all the "YOU CAN'T INSULT OPPRESSORS WITH POWER" bullshit
guess what bitch, we just remembered we do have power. cancel deez nuts you dumb dysgenic pile of human refuse
what does it say
I wonder how similar you find ukrainian to russian, like if u can understand it
it's a recruitment poster for the ukrainian military, seeking men 18-24
it has the payment you'll receive converted to robux
fight the russians and you could get 33 million robux ๐คฏ ๐คฏ ๐คฏ
polish and ukrainian both make me feel like im reading russian while having a stroke lol
๐ค
joining roblox and not just getting groomed, but getting groomed by a man who has killed dozens ๐
Ooh so men seeking men? Nice โ๏ธ
I've just deployed the first version of the classifier. At about 80% acc it's not perfect but it's good enough to at least be a participant in the voting system. Anyone who scores more than 0.8 BAD_USER from the classifier had a vote automatically cast indicating all their groups are "bad". This has raised the total vote count from 4,836 all-human votes to a new total of 46,641 votes
Because the classifier's score can only cast 1 vote per group, it can't mark any groups as fully bad - all this will do is elevate the groups so they can get more attention from the human voters
I'm going to try voting again, can you pin the link inside the thread?
pinned, thanks
some of the shit that the classifier found that the human voters didn't is fucking vile
like actually worse than the stuff before
people straight up offering to trade 'ild 'ography in their bio texts
fucc
First one I checked seems to be a sort of "e-girl" style thing, didnt look sus but I'm wondering if there are hidden cues I'm missing
Need to get a better sense of "normal" in my own head
just hit unsure if you're unsure. it doesn't fuck up the scores in either direction, and of course there's plenty of stuff either of us wouldn't recognize
It seemed 99% benign
Just wondering what I should expect to see, there's a lot of subculture stuff
stuff to look for:
- any mention of TOS = automatically bad. "we follow TOS here!" is just them trying to be coy
- spade symbols like on a playing card = bad. raceplay stuff
- bunch of "BigBrownBull" type users in the group list probably means something bad
- furry stuff isn't automatically bad, but a lot of times, it is bad if you look closely
- any ROT13, Morse code, binary, etc in the description is almost always bad. you can decode it if you want but I've never seen it not be bad
wtf is a "wedge" or a "bypass wedge"
im not sure what "wedge" means but anything talking about a "bypass" is almost always bad too
roblox has a very poor classifier in place to block "nude" clothing items
What does "bull" mean on its own?
bull refers to cuckold fetishes, generally associated with raceplay fetish groups
Its a group that the one i saw is linked to
kids wear swimsuits, and roblox might not even ban this type of content. feels very sinister to me though
this follows the same understanding of "wedge"
Like the waistband of underwear rising above the trousers
Yeah I think they most moderate out "thong" and "wedge" is their chosen euphemism
How about we write up posts on certain euphemisms and you can pin them to the thread?
Clothing euphemisms:
- "wedge" likely means "thong" or similar, a way to talk about revealing underwear without hitting filters (Inappropriate)
trentbot, ols had a good idea in here. review the contents of #1370427703223783629 and write up a list of the euphemisms bad roblox users use to avoid filters that we've already discussed here
Here are the main euphemisms and โcode wordsโ weโve already spotted in this thread that bad actors use to slip past Robloxโs filters:
-
โwedgeโ (or โbypass wedgeโ)
โข Used in place of โthongโ or other revealing underwear to talk about near-nude assets. -
โbullโ (and variants like โBigBrownBullโ)
โข Cuckold/race-play fetish shorthand. -
Spade symbols (โ )
โข Another race-play indicator, often combined with โbull.โ -
โild โographyโ
โข Disguised form of โchild pornography.โ -
โbypassโ
โข General tag to imply โweโre skipping the filter,โ e.g. โbypass wedge.โ -
ROT13, Morse code, binary in descriptions
โข Encoding text so the filter canโt see the real message unless decoded. -
โwe follow TOS here!โ
โข False reassuranceโused to pretend content is compliant when it isnโt. -
โfurryโ
โข Sometimes used as a blanket term to mask sexual or fetish content.
โHonestly, itโs gross how creative they get to hide this stuff.
Channel Id: 1370427703223783629
Limit: 100
Looks like you need to scan more messages than 100
yeah its probably gonna fuck it up
not bad tbh
Symbols:
- Spade symbols (โ ) are a raceplay indicator, and can indicate fetish content (Inappropriate)
Oh yeah, forgot to post this. this is just a chunk of what the classifier found. decode if you want. it's really fucking gross (but to be clear, it doesn't contain or link directly to anything that is illegal on its own, though it does discuss such content)
@neon girder
Oh yeah, hearts are for some reason too. I'm not sure exactly like, the etymology (?) of the symbol, but I've seen it in quite a few raceplay-related groups
I'd break it up into themes like "symbols", "codewords", "suspicious claims", "fetish cultures"
"Queen of Hearts" specifically I've seen several times
https://www.roblox.com/communities/35684621/unnamed#!/about hearts here too, I flagged it at inappropriate due to bypass wedge
also, not sure how to discuss this without it coming off as homophobic
but a lot of the LGBT related groups I've seen are outright inappropriate
my suspicion is that the groomers on discord recognize that many children who may join those groups are feeling socially isolated and that's why
I could have a go at explaining but I suspect Trentbot will kill me if I try
anyway, I'd say look at those with a bit more suspicion than you normally might treat such a group with. sad but necessary in this case
I can think of another reason to do with the groomers' own backgrounds
A lot of lonely adult gay men may have had formative experiences when they were younger which they want to relive, and have never had healthy relationships in adulthood
true
oh yeah, that made me think of another thing
although I'd imagine there probably are innocuous groups on Roblox with terms like these in the group titles/descriptions, anything related to parenting, daycare, etc should be viewed very suspiciously
I'm wondering whether to mention this sort of stuff at work as I'm getting roped into to defining how tuning processes work since most people don't seem to be able to grapple the issue properly
farts too. ofc I can see some 10 year old making a group called "funny farts" but a lot of fart-related groups are obviously fetish content when you look at them for a moment
ยกแดกแดสแดแดแดแด แดแด สษชษขแดแดแดกs!
โฐแดแดษชษด แดสษชs ษขสแดแดแด ษชา สแดแด สษชแดแด าแดสสสs แดษดแด สแดสแด แดแดษด
โฐสแดาา แดสแดแดสษชษดษข สแดสแด
๐ฅต
be careful if you do lol. I've tried explaining the problems on roblox to people i know irl and it's like 50/50 whether they get it or whether you come off looking like a fucking crazy person
I'm usually good at setting the scene, they want someone to present on how AI tools can help to enhance cybersecurity processes and that probably has to be me
I wondered when this would start showing up in the logs
I found this last time when I was doing all the classification manually
whole massive cluster of sometimes but not always furry themed "gay bear" type content
Their store items are all speedos
๐๏ธโโ๏ธ WELCOME TO ALPHA GYM ๐๏ธโโ๏ธ
The Ultimate Gym Experience for Dominant and Strong MenStep into Alpha Gym, where strength, masculinity, and brotherhood come together. Whether youโre here to train hard, dominate the weights, or just enjoy the company of like-minded men, this is the place for you!
Are you ready to train hard, dominate, and embrace your alpha energy? Join Alpha Gym today and experience the power of brotherhood like never before!
๐จ IMPORTANT RULE โ NO EXPLICIT OR INNAPROPRIATE CONTENT! ๐จ
Alpha Gym is a roleplay-only group focused on fitness, camaraderie, and strength. Any form inappropriate behavior, or explicit roleplay is strictly prohibited. Violators will be removed immediately. Keep it clean, respectful, and fun for everyone!
I'm not really that aware of what other reasons a roblox group might be shut down for other than groomer related stuff
I mean given the context in this case I can make a pretty good guess about this one lol
but generally I think if Roblox locked it it must be pretty bad, cause look at all the stuff they don't lock
Yeah there's a cluster of ones that offer "clothing for buff men"
Tbh if I see stuff like that then its getting marked inappropriate lol
there's something about the gay fetish content that I've noticed
it seems to like, take advantage of the fact that it's generally somewhat socially acceptable for men to be shirtless, or generally socially positive for men to be in good shape. these things aren't immediately indicative of the fetish content they're pushing in the way that the raceplay stuff does
I did find a non sexual furry one
What's your read on ones that have a lot of deactivated accounts in their user list? https://www.roblox.com/communities/35626458/sacramentalia#!/about
even from the non-deactivated I'd probably lean sus on this one. but in general if i see tons of banned accounts on the list, I lean sus for that too
Yeah I say sus
similar to group bans, you look at all the people not banned by roblox, so the ones who got banned must've been even worse
Yeah this is open and shut lmao https://www.roblox.com/communities/35708333/Bleached-Society#!/store
christ
Wedges in the store
come to think of it
this group has some spanish on the wall posts
I've seen some overlap between english and spanish content
but I haven't seen any Russians yet. I wonder why?
Maybe lack of overlapping language skills so they steer clear of mixed groups
โ SWF ONLY!!! YOU WILL BE BANNED IF YOUR IN A ACE/SPADE GROUP!!! โ
Interesting, so there's a clear understanding that ace or spade groups mean sexual content
Enough that people are enforcing that themselves
But yeah I think this classifier may end up inadvertently homophobic due to the large number of gay bear groups getting swept up
this has already been a problem for another similar project
Group for people who have +15y/o pls!!!
Unusual age specification
the one youtuber who covers this extensively, Ruben Sim
he made a bot on discord that mapped the connection between roblox users and discord user IDs
then made a system where you could automatically ban known ERPers from your server
the idea was to make it so you could at least kind of protect a Discord server that revolves around a Roblox community
Discord ended up banning the bot account and his personal account for something about 'targeting protected classes' or some shit
smh
dont wanna sound too conspiratorial but like
they clearly looked into it enough that they could gather it was banning a lot of gay people
why didnt they look into it far enough to determine what its purpose was and where it was being used
bit sus imo
https://www.roblox.com/communities/35536409/Fun-games-4u#!/about I think this is fart fetish content?
The pfp shows an anime character who is known for smelling bad
fucking hell
oh yeah, look at the Shout post
"working on new frp game"
I'm gonna hop on an alt and join the linked discord
4 content deleted experiences
Maybe navigate through the listed users, see if it links to other fart fetish content
The main user role is "gassy"
requires discord account verification through some shady bot. not happenin โ
I guess it'll stay a mystery whats in there but I think I can guess
the raceplay stuff is disgusting in its own right with the audience these people are trying to target, but the fart stuff is really shocking. I'm quite surprised how many of them there are
๊ฐ . . . wlc to da cute morgue . . . ๊ฑ ๐ฎ๐ข
เญจเง clothes by illinthebrain & ilIinthehead or bought .แ.แ ๐ธ๏ธ๐ท๏ธ
๐ช send ally reqs, we accept all! โฑ๐ชฆ๐ค
โญ เญจเญง โ โ ranks โ โ เญจเญง
โส 0 (เนรเฝรเน) postmortem examination ๐๐ช
โส 1+ เญจเง ๐ฆด ready for burial โฐ๏ธ เญจโ เญง
โส 5+ โฑ decomposing ..๐ชฆzzz
โส 10+ ๐ชฆ. . . (ยฌยบ-ยฐ)ยฌ resurrected
โฐsteal n we'll eat ur brains!! ๐ง ๐งโโ๏ธ
โง: est. 11.15.23 ๐๐งธ.แ
Not one I got served but affiliated to it, not even sure what to make of this
these ones you gotta look at the clothing items generally. kinda strange but also not unlike the sort of thing an edgy kid would write up
Yeah I think its just a goth aesthetic, sort of Wednesday Addams shit
I don't have access to some of the info due to not being logged in, was still evidently fart fetish lol
This is clearly suspicious, several affiliates are locked, I guess "floofs" or some other key terms signify something e-rp related https://www.roblox.com/communities/33839103/fl-ofs#!/about
Claims to be a fan club of this, look at the bio
ahhahahahaha
looked through the members list and found this absolute mutant
There's a cluster here seemingly for black girls https://www.roblox.com/communities/35651737/HARAJUKETTE#!/affiliates
One affiliate is called "only brownies" and the clothing seems to reference black pop culture a lot
Hmm its "trap" culture I guess?
Not even hiding it https://www.roblox.com/communities/13458682/farting-groups#!/about
im really not sure what to make of this one. but i did have a good laugh at this member of their group's pic
lol
There seems to be a lot of intersection with SFW furry groups, black culture groups and LGBT groups
So I think being discerning on any of those is important
This community group is for players that are part of the REAL fart community. If I find out you're anti fart or friends with any anti fart people, I will not hesitate to ban you.
Oh. Also, fish is definitely the one taking down the fart games, but at the same time, he's helping us.
Oh and here are some rules (READ THEM CAREFULLY, CUZ YOU REALLY NEED IT):
- If you hate the fart community, this group isn't for you.
- No kids either. If I find out you're a kid, we won't hesitate to ban you.
- Do not advertise fart rp in this group, because it may cause thos group to go on locked, so if it does happen, it's your fault. If you want to frp with people so badly, put it in your bio and just say "read my bio." Simple. If you are creating a "game," just friend me and let me know, but again, no advertising frp here.
- Behave and don't be so beefy here.
I've noticed the furries seem to be a little bit more cautious than other groups
I'm not sure what to make of this observation
there are a lot of things I think look very suspicious within the furry groups but they're not outright fetish content
I can't quite tell whether this is me misunderstanding general furry culture or if it's them toeing the line more closely than the outright fetish groups
lmao REAL fart community ONLY
There's a clear thematic distinction between groups offering somewhat revealing (eg. crop tops, underwear) that are targeted at girls, and ones that are targeted at men interested in roleplaying with "girls"
looking at the wall posts for this group
Here's another dubious one - no store/affiliates to go off, can't directly find sus content but there might be sussy stuff on the wall. Name of group is very sus
Yeah it seems they're promoting a workaround to avoid getting banned for advertising frp
jesus posting got a good laugh outta me
seems like that guy at the bottom is saying they were reported for being 'edos or something
Maybe I should make an account to see this
yeah, no email required either. ezpz
pic of a guy i found earlier. there's a lot of overlap between furry and diaper groups and idk why
this one is really sad
from the list of high role members and from the store page it looks like this group was probably created for innocent purposes, but the member list at the entry role is full of fetish accounts
Yeah that must suck
I just got it on my queue too
Is that appropriate or inappropriate then?
I think i voted appropriate on that group but it really could go either way
I voted appropriate too
I'll stop for now but some interesting clusters there
Definitely seems to be a lot of "horse" content, of varying types
hahahaha holy shit this is crazy
just posted an update to the web interface https://epitracker.org/
it now shows classifier outputs
despite only getting ~80% acc it did a great job
The morse code outputs are always depraved
yeah they're really bad
the classifier did a great job identifying them
most of these people weren't already marked as participants in a lot of bad groups
MIKU THE BEST OF ALL JUST SO YOU KNOW
First morse one that isn't something sickening
@merry yacht what about bunnies? I see so many users and clothes with bunny in the names. In the case of clothes they sometimes had bunnies on them which probably means nothing then.
oh yeah, that's part of the race play stuff
"bunny" is shorthand for "snow bunny" which means a white woman who prefers black men
trying to phrase it so trentbot doesnt delete lol
So we have:
"stuff in trentbot's message i can't quote
Bunnies
women into/with blacks,
Bull
Black men in race fetish context
'Content deleted' in previous group names.
"
What else do we need in a basic guide? Honestly, with enough tells I wonder if there's a cheap AI that can analyse groups. Scraping can be awkward for some interfaces though. Like I tried using a workflow site to scrape with jina and it would miss product reviews that weren't page 1.
refresh the page, I added some stuff to a glossary in the "about" section based on this conversation
also, if you refresh and check the stats page, I've updated that too. it now has a list of users based on how bad my custom classifier thinks they are
appears to be working well
@merry yacht I think I found another identifying sign
One group has a load of random admins who all seem to be wearing shirts with a slice of cheese pizza
All of them have no links to other friends/groups
It seems to be a free item offered directly by Roblox, idk if you have to manually select it to wear it
https://www.roblox.com/communities/35557253/B4D-G1RL#!/about look at their admins
Makes sense with the whole "cheese pizza" thing. Fucking freaks think they're so funny
I think I've found a vore one now
Seems like "nom" is a preferred term, they use a lot of variants to typical vore terminology
"feeder/feedee" in the shout box
it probably is coy enough that they wouldnt be banned for it unfortunately
Although "pred/prey" also present
https://www.roblox.com/communities/35405292/glup-nom-ov#!/about this seems to be a small vore group, it becomes more obvious if you look into member profiles and see their other servers
Yeah and its official content so being banned for wearing it would be absurd
And yeah its clear what "cheese pizza" really means
https://www.roblox.com/communities/7832604/no-socks#!/about suspect this one is foot fetish
Do cherries mean anything? I feel like I saw shirts like this in sus groups before.
A cherry can refer to a girl's virginity in certain contexts, you might need more information to determine whether that's what they mean
Btw some morse code ones are encoded again in caesar cipher
I found one with a shift of 3
What about blood on shirts, i feel like thats a virginity reference now you brought up the cherry.
Can you link the group?
Ah I already marked that inappropriate due to its ties to cheese pizza enthusiasts
Can't figure out what the shirt is meant to mean, possible there was text at the top which is censored
@storm star @merry yacht https://www.roblox.com/communities/8281655/Kids-United#!/about lmfao holy shit
Kids United!! is a community on Roblox owned by Kidsunitedholder with 15770 members. โI know this group looks sus but I will not allow anything to happen. And if bad stuff comes in the group wall I will turn it off!โ
ส โโโโ โ
สโกษโ
โโโโ ษ
๐บAdults are allowed! This group is mostly so that kids can talk. ๐บ
Group bi...
the role names tell me that this is certainly 'edophilic
โI know this group looks sus but I will not allow anything to happen. And if bad stuff comes in the group wall I will turn it off!โ
Is that the largest cheese pizza enthusiast group we've come across yet?
probably not but i havent been keeping track
Are there false positives coming up now? I'm looking at this "no socks" group, and there's no store, mundane description and pic. The member names look innoucous to me. The only thing I can see is 30-40% of members are inactive or banned. But the group was rated 0.8 something.
could "no socks" be a euphemism tho?
Its foot fetish
lmfao
Look at the wall, loads of people asking to rp
What wall?
i think you need to have a roblox account to see the wall
Make an account, you can do it in 10 secs no email
Lets you see comments on the group
often massive giveaway
It's a "via negativa" name, its defining itself by what its not
"no socks" --> "bare feet"
I uninstalled roblox ages ago. I'll reinstall and get my login details if I can.
I might need to for password.
hahahhahaha the 4th one what the fuck
Right thanks, making a new account is surprisingly painless actually.
https://www.roblox.com/communities/3014349/SmallGroupie#!/about ageplay/infantilisation one I think
SmallGroupie is a community on Roblox owned by Smallgroupie with 1741 members. In this group we mainly sell baby clothes and everyone who joins i do let them talk in the group wall. I do check it to see if problems come from it and take action.
If you wonder why i have it set to private is to mainly make sure scam bots dont get in.
If anyone do...
hey guys ive recently started getting back into play roblox and bloxburg and stuff, and im just wondering if anyone would like to roleplay!! mainly looking for a baby/toddler, (preferebly girl) and 1-3 kids, 1 boy, 2 girls!! pu (pop up) if ur interested!!
๐๐|~| RING RING |~| .. Abc for mUh MuH'sH Ari! (Examples) .+. Leans down and slids her up in arms .+. Grips onto railway .+. Pats her back lightly while bouncing her on hips .+. (etc) If interested feel free to add me! See ya little ones & Adults! ๐
I see what Trent was talking about
What is the crowd score reading? As it gets lower (which happens gradually as you vote) you'll start getting more non-inappropriate groups. This is good though, voting non inappropriate is also helpful
hahahahahahahahahahahaha
da bubbie, toddler, and mwa mwa club :> is a community on Roblox owned by cutebub_aria with 136 members. Hii! Welcome! The old owner of this group was hope, but unfortunately, she left. I can help out if there is a situation or problem though. You can use this group to find people to roleplay with and so! I might start to post onesies, or other ...
Jesus christ
what does that one have on its wall
Fuck dude these are getting markedly worse as the scraper expands
It was a high score for that one. I'm getting high score groups now.
Idk not logged in on mobile
Trent do white shirts with red smears mean anything?
Huh. Weird. Well, regardless, voting not inappropriate on fine groups is helpful regardless
I'm not sure about those particularly. A little bit of blood is probably just kids being edgy. I've seen some self harm related groups tho that are extremely gross
This one hasn't been directly served to me yet though
if you think about it, the growth rate is insane
everytime a new group is flagged it could get a hundred more groups to get marked as sus depending on how interconnected it is
The last time I did this project when it was just me doing classification I found an interesting trend
The network is extremely densely connected
Which you sort of wouldn't expect, right? Why would gay bear enjoyers also be in a group for little girls?
Imo the answer, as horrifying as it sounds, is that the connecting thread is the corruption of children and the specific type of content they use is almost irrelevant
agree
Agree on this point but it's still odd that there's not even a clear barrier around sexuality
https://www.roblox.com/catalog/128625957422555/t-shirt-blush another "blush" shirt
Last time I did it the network's diameter was about 11, meaning you could start at any point and get to any other point within 11 hops
"bypassed" in the tags
Yeah, I'm sure red smear shirts are a tell now.
whats it now?
the value for real life is only 6 which is crazy
Didn't check yet, if you want to check with Gephi I have a graphml dump at https://epitracker.org/dl/roblox.graphml
Its about a day old so already significantly smaller than the current network. Only potentially bad nodes included, so as to avoid making it too large to load into memory
I'd like to work out what they really mean
I think they are meant to appear like a bare chest
And they go over the top of that layer
Did the website freeze up?
Nothing will submit or load for me
I'll go check on it. Neo4j might have crashed again
It started loading again
scraper is fired up again, automatic scoring with my classifier paused for now because it's eating up too much resources
gonna dump a new copy of the graphml so i can figure out current network diameter
@neon girder network diameter rn is about 10 (i had to prune some nodes from the graph to make gephi not crash)
somewhere between 10 and 12 I believe
btw do you know what graph neural networks are?
they would be perfect for this
I've looked into them and they seem 1) very complex and 2) also not as well optimized for training big ones
my (uneducated) position on this is that I'm probably better off feeding data into a suboptimal model architecture that already has a great understanding of text content, as opposed to an optimal model architecture without that kind of understanding
since basically all the features of the data I'm working with can be encoded into text and understood at a glance by reading
roblox gooner network
this seems like a very autistic program
i let network diameter run longer (still didnt finish) and got 11 diameter
you dont need to calculate it in full
do a thousand random tests and calculate the average
What are the main groups now?
Wondering if you can pick out some of the fetish themes from what's grouped up or whether its too dense
had to rerender since I closed it earlier. let me type up a little breakdown
light purple group (15%)
green group (13.88%)
blue group (11.06%)
black group (8.91%)
orange group (7.4%)
oh shit sri lanka exploded
Furry bara
diapers?
looks very mixed, can see farts and ponies
this is actually a strange and troubling recurrence
this happened the first time around
ageplay?
lot of mixing between furry and diaper/fart content
fart furries
yeah and bara in everything
I think black is mostly inappropriate clothing groups with some gay content mixed in
explain this behavior immediately @inner dagger
capybara?
some word for gay furry
"oh it's fine, it's just roblox. i dunno she's probably playing some kind of dress-up game or something"
her friends list:
๐จ
hahahahahahaha
quoting from an update post I made on KF today
Here's what the various groups are made up of:
Pinkish-purple (12%) - Despite appearing smaller than the green group, this is the largest group in the network. It's tightly clustered, making it appear smaller than it is. This group is made up oif a collection of gay (mostly femboy), raceplay, and nude clothing groups.
Green (8.92%) - This is mostly raceplay groups, with a collection of toilet and fart fetish groups.
Light blue (7.68%) - This is a cluster of 4 groups. Of them, 3 are quite large and appear to be mostly innocuous groups selling clothing for black kids, probably made for legitimate purposes and then infiltrated by gooners. The fourth is a smaller group of 387 members, which is explicitly raceplay related.
Dark gray (6.47%) - Almost exclusively nude clothing groups
Orange (5.89%) - More clothing groups
Salmon color at the top left (4.79%) - Furry groups with comorbid fetishes attached. Every group is furry related, many groups also appear to be about inflation, feeding, and vore fetishes.
Teal at the top right (4.38%) - By far the worst cluster, and something that I completely missed when I first did my solo version of this project the first time around. These are groups about very young children. There are only 6 groups in this cluster, but a troubling amount of members. The groups are called: SmallGroupie, Little Supplies, Little safe place, Bronies & Babyfurs, and (stylized in-game, typed normally here) Bub's Daycare and Bub's Tod (toddler?) World.
@storm star @neon girder
@rose hemlock
- Screening out legitimate groups that sell clothing for black kids, as well as black subcultures in general, seems to be a priority. I can imagine raceplay people go to these groups to get clothing to use in their raceplay groups
- Is there any point classifying the style/purpose of the clothing groups in more detail? I found trap stuff, scene, goth, cutesy princess, sfw furry and a few others that might be worth clearly identifying to see their relationship to other networks
- Likewise for the furry+ cluster, would it be worth tagging the inflation/feeding/vore cluster (themselves all too tightly linked to disaggregate) so we can see exactly what the overlap is?
- Are we definitely sure that Teal is about CSAM stuff only and not about infantilisation/ageplay too, with overlaps?
Orange and dark grey it might also be interesting to do some deeper tagging of the clothing style, see if other relationships become visible
I imagine the "wedge" ones are very universal as they don't appeal to any specific fantasy
Do you think this one started innocent then got taken over, or was always fucked? https://www.roblox.com/communities/22759/Teens#!/about
Some of the banned members have names with references to PDF file terms
this is disgusting
hard to say how it started but looking at the wall posts it has definitely been taken over
"AAM" is a term worth adding to the glossary, it means "adult attracted minor" and I shouldn't have to explain that further
Is it worth distinguishing between ones taken over and ones that seem to have started with a nefarious purpose?
- The line between legitimate and illegitimate is still kinda blurry. Classifying groups is the function, but the purpose is really classifying users who are members of those groups. If a legitimate group gets invaded I'd consider it a good target for classifying as inappropriate regardless of the original intent
- I can break it down in finer detail by changing the modularity community detection settings but I think the purpose mostly stays the same regardless of the chosen style of the bypassed clothing. These groups act as a hub which link to other, more sinister groups
- The context was missing because of the way I copied+pasted but this is actually a recurrence of something I saw the first time around. For some reason there are a lot of groups where the theme is furry + some very bizarre fetish. This cluster in particular is basically inseparable, it's not a mixture of furry and fetish groups, it's all furry + fetish at the same time
- I mean I guess you can argue that pretending to be a toddler isn't โ๏ธ ๐ค technically โ๏ธ ๐ค CSAM, but given all the stuff we're seeing in bios and usernames that's straight up CSAM related, I question how meaningful the distinction is
I know there's huge furry+vore overlap too
Although I've not seen an obvious furry+vore group here yet
For vore I usually look for the key codeword which appears to be "nom"
And its fair to treat all vore groups as inappropriate
I'm seeing the big belly/vore pairings too
many such cases
the network expansion is going great
it seems to have just recently located this cluster
I wasn't seeing any of these for the last few days and now I'm seeing tons
tbf I could have told you about this cluster beforehand just based on intuition, its visible across the internet
Oh look, a furry+vore https://www.roblox.com/communities/16447867/Wolf-Inflation-INC#!/about
true, but automated expansion is the goal here. having to locate them manually would mean way, way more work overall
oh, it looks like we're at almost the same place in the voting queue. I just saw the teens group you linked a couple minutes ago
Linked me to an X account too https://x.com/EiMikoR34__
poly bi enby | age 20 | he/they pronouns | proshipper, swiftie, fudanshi | my throne :3 | https://t.co/DPdMu921bS
Forney, TX
https://www.roblox.com/communities/8713338/BuB-sH-DaY-cARe#!/about infantilisation one
his X links to https://ashtonsaberstyles.straw.page/
which links to
https://www.tiktok.com/@1989.kaelucsversion
he looks almost exactly like id imagine lmao. except thinner than I'd have guessed
the "fudanshi" thing suggests they're a man who likes discussing gay pairings of characters the author did not intend to be seen as gay couples
Although the other site suggests attractions to both genders
And yeah he looks like that on his Youtube account too
There's probably other creepy shit linked to his Roblox account https://www.roblox.com/users/87829013/profile
Proshipper Safe Space is a community on Roblox owned by BWON3Z with 153 members. Hwello!! This is a proshipper safe space <33
Just incase, Your account will be reviewed before youโre ranked. If you arenโt in any pro shipping groups or your bio doesnโt say youโre a proshipper, you will stay in the pending rank.
Rules:
No self Promo! ...
They describe it like they only look at drawn/rendered content and not real photographs
Although "pixels" could mean anything tbh
dead dove: do not eat is a community on Roblox owned by maskedcomshipper with 84 members. [ OBVIOUS BAIT WILL BE IGNORED ]
since : 08 . 01 . 24
๏ธถ๏ธถ
"dead dove: do not eat, I don't know what I expected!"
. . .
meaning if you don't like what you see, don't interact
๏ธถ๏ธถ
this is a group made for proship, profic, comshippers and darkshippers. ...
Probably worth other evaluators knowing some of these terms
https://www.roblox.com/games/18794518336/PROSHIP-AND-COMSHIP-HANGOUT the disturbing thing about this network is everything still seems to be live, seemingly Roblox believes their denials
@storm star if i can find this guy's phone number i think im gonna call his parents
pretty brazen support of loli and incest content
if you do then do it on voice so we can listen in
if I can find it I'll definitely record
Your message has been deleted, sorry!
Next time, try saying something like this instead: It's unclear if "loli/shota" strictly pertains to animated characters or if this individual is also engaging with inappropriate content.
You have been muted for 2 minutes for the following reason:
sexual score of 0.67
sorry @storm star didnt see that it muted u
look at this
it looks like he's using a translation app to ask someone if they're 14???
what the fuck
Yeah, I think he's an active abuser
I'm going to pull up a list of High Schools in Forney, TX
It has a population of 23k so quite small
There's only two
I looked to see if they had yearbooks available briefly but didn't find any
I'm inclined to believe the first name Ashton is correct but the other bit doesn't seem to be a last name. I think "Saber Styles" might be a reference to beat saber which I see he has some videos of himself playing
https://www.youtube.com/live/dbE3Ooo-yfo?app=desktop well there's one of the high schools graduation ceremonies in 2023
Which is probably his graduating class if he's 20
holy shit
@storm star
he did a year in jail for something 'cest related???
bruh
okay wait
Check the s offenders register locally?
Ashton Styles?
It doesn't seem to respond to me, I'm wondering if they block connections from Europe
I'm wondering if he did this before he turned 18 and the record is sealed
davidgaige@icloud.com seems to be a registered email
looks like he posted it jan 8th 2024, and if he's 20 now, and he got out sometime before jan of 2024
so it might be that you're right, happened before he was 18
where's this from?
His bandcamp profile says he's in Garland, TX
https://www.whitepages.com/checkout/summary?wpId=PN3Vzj0Wbl3&wp_content=card&wp_medium=PersonContactInfoUpsell&wp_source=PersonResults&wp_term=serp_ep0&funnelType=person_onepages_1
I think this might be him
I think so
Maybe ask if his son is the one who got locked up for looking at the PDF stuff
Or maybe that's the wrong approach
But good luck
He's topless in almost all of his tiktoks which is really weird
getting mic setup working rn
yeah I think I'm gonna try to frame it as me just being concerned, like trying not to sound accusatory
Maybe watch some of his most recent tiktoks
They're very short
It seems like he projects a lot - like accuses others of being PDF files and saying they infiltrate subcultures
Which makes me think he thinks he got rumbled due to guilt by association and not personal guilt
ie. what happened to my brother, except in his case he's clearly deluded
That he still has loads of "proshipper" accounts suggests he's not willing to deal with the core of his issue
no dice, both phone numbers dead. need to check other sites
Doubt its current, I think they left Forney and moved to Garland?
Given his age he's more likely to be on Facebook
See posts, photos and more on Facebook.
Forney
can't see contact details directly but it says he works at ABOX Packaging which seems to be a small company. maybe I could talk to him that way
I've made several assumptions to get to this point so might be best to see if we can reconfirm other details before making a call
He has videos on his profile, I see one other person (a fat woman) in it, might be worth checking to see if "Ashton" appears in them
He claims to be widowed so idk if that means Sarah Gaige is dead or whether that's a different relative (eg. sister)
i wish these sites didnt have all these fake loading bars ffs
https://www.tiktok.com/@1989.kaelucsversion/video/7314337342519610655
might be worth comparing anything showing the inside of his house with david/ashton's videos
The two videos are outdoors around a fire
found an outdoor one on his profile. looks to be an apartment complex
https://www.tiktok.com/@1989.kaelucsversion/video/7421679285821246766
I'd be surprised if he lived on his own, he seems incredibly immature, and he has too many nice things for him to have been kicked out
@neon girder can you figure out how to find this guy's address or phone number? Need a current one or one for his father
i have no idea lol
this truthfinder website is fucking ridiculous
it is STILL prompting me to click through fake questions with fake loading bars
"truthfinder" sounds like some retarded alt right site lmfao
lmao
for fucks sake it wants $28 for a membership
well, spokeo is offering me a "trial membership" for 95 cents. i might actually take that
Yeah a lot of these sites use that to make people with any of those intentions feel "clever" for using them, its the same for tools that purport to allow you to view private instagram accounts or get details from people's socials
well, wasted a dollar on spokeo
numbers for both his parents and him on spokeo were all dead or belonged to unrelated people
gonna go cancel that sub rn before i forget
@merry yacht I found a heavily moderated one where someone has done the legwork of putting "creepy, gross" people into one category you can browse through
Might have found a self-harm cluster
I think the group itself is benign, but they just contain people who are weird and dont kick them
https://www.roblox.com/users/3381335087/profile one of them in their bad list
Queen123467J is one of the millions creating and exploring the endless possibilities of Roblox. Join Queen123467J on Roblox and explore together!About
๏ธต๏ธต๏ธต๏ธต๏ธต ๏ธต๏ธต ^ feel free to add โก ห.๐
โโโโเญจเงโโโโโโโโเญจเงโโโโเญจเงยท โงเญจเญง ๐
๊ฐCopying fits = block (inspo is ok)
Fllw qt fits /or (f...
https://www.roblox.com/communities/14256006/kitty-pawz#!/about is the group, if you look at member lists you can see a "gross" group
.kitty pawz. is a community on Roblox owned by shopkittypawz with 3976 members. ๊ฐ^. . เพเฝฒ^๊ฑเฉญ ๐๐ฅโข ฯlcm 2 .kitty pawz.
๐ฝ๐ฐ,, cutesi lttl grp 4 u - ! ! ๊ท ห ๐พ ห
เซฎ๊ฐเพเฝฒใฃห -๏ฝก๊ฑ ๐ฅ๐ฎ .. 2/5/22 .. ใปโ โง หณ ๊ท
๐ขโธโธ dmca active ๊ทโ๐
๊ท. แ ask 4 rank up ๊ฐแข. .แข๊ฑ.. .๐ค
โญ เญจ...
The actual group seems to be "yume kawaii" subculture
I think it's worth updating About section with bunnies, red smear shirts and the previous names lookup tip. Perhaps cherries, anyone see cherries?
Not seen many cherries since but I've defined a few other PDF file related terms
I don't know what the red smear shirts mean, I'd rather have some idea before listing it
I feel like red smears are only in bad groups.
I assume this is something very evil https://www.roblox.com/communities/35448169/Ren-s-Sanctuary#!/about
Big clue are the spade shirts, the fact that its held by a holder account, and its small with no desc
I wonder if the spade thing is related to spade being a name for black people before.
have we considered phishing
or ig you guys
Even if he has that its a good chance its under his alias
We think this is his dad though https://www.instagram.com/phillipgaige/?hl=en
report them to discord
dont
No
discord will ban your account as well
I want to play with them
thats what they did to me
discord banned my account as well as theirs
thats why i had to make this account
???
i thought you were allowed to be on here at 16. i thought 13 was the min age for creating an account
yes
but they banned me for "child self endangerment"
because a pdf file was asking me for ndes when i was 17
true. minors shouldnt be on the internet at all
lol
I'm so iffy on those, like that one group looked like it was a bunch of teens, had a ton of hoe clothes with poppy names, even young kids just think nudity is funny and will undress stuff. I feel like it always has to have one thing more than that, usually mentioning some fetish or code word
Oh noooooo
Probably both? Nobody who's been a furry for even a little while isn't aware of their reputation, all the corn, the fact it's purely sexual for a large chunk, etc. They know they're one bad apple or disinhibited autist from being shut down
Because of the diaperfurs, lol. That's a sub-subculture
Damnit imagine me as a kid, with bunnies being my favorite animal ๐ญ and I never called them rabbits
well yes, but what i meant is, why are there tons of diaperfurs and not just "regular" diaper people?
There's a big list of keywords like this I found once, I think it was literally published by the FBI
Yeh, usually virgins, could be code for just really young girls / boys. Not always
Oh yeah, I got you. Um, probably, that might be how a fair chunk of people get into it these days
i want to join the investigatoin
is "evade" in clothes names bad?
It does sound bad but maybe show an example?
Hmm I wouldn't say so in that case, it looks like its just a random word and they're a goth/grunge/edgy clothing store as they say
A lot of these things show up on network because weirdos join to buy clothing but I don't think they are themselves problematic
A lot of the time the network of problematic groups should end on these sorts of stores, as there will be people in them who are not in sussy groups and you don't want to flag their other groups too
Evade is a word I'd expect to see for dodgy stuff but it does seem like "bypass" is the preferred term
A lot of the users will be like 13-14 and have poor literacy, often just find words in a thesaurus
If the group owner is not a "holder" account you can sometimes learn a lot from them, or from groups that people with "admin" type roles are in
They have "bunny hospital" in allies by user called "handsjob". I was going to skip or put not innapropriate but now I might tag inappropriate.
I don't think the name of the owner of an affiliate group is enough tbh
Maybe skip
There's a sort of "creepy cute" thing in a lot of groups, so you get jarring language about cute looking corpses or ill cute things, I don't think that's meant as sexual, just edgy teenager stuff
Looks like normal gothic aesthetic
Yeah, it says "viscous" to be edgy lol, they got confused with vicous.
vicious?
Um yeah, what is that I doing in there?
Missing the second i
I'm becoming way too familiar with Roblox subcultures and content creators lol
Looks like I'm being served loads of non-sus stuff today, I guess they're the ends of suspicious clusters
nvm back to the sus ones lol https://www.roblox.com/communities/8459026/PacifierRoblox#!/about
https://www.roblox.com/communities/34591449/Balloon-Stomachs#!/about inflation one smh
once you get down to where the score of the group displayed is < 0.75, you'll start seeing a lot more non-sus ones
@neon girder 11th try on the user classifier
Finished classification and evaluation in 9.34 seconds.
--- Evaluation Results ---
Total samples processed: 3598
Total actual 'BAD_USER' samples: 310
Total actual 'GOOD_USER' samples: 3288--- Counts ---
True Positives (TP): 138
True Negatives (TN): 3260
False Positives (FP): 28
False Negatives (FN): 172--- Rates (%) ---
True Positive Rate (TPR / Recall): 0.4452 (44.52%)
True Negative Rate (TNR / Specificity): 0.9915 (99.15%)
False Positive Rate (FPR): 0.0085 (0.85%)
False Negative Rate (FNR): 0.5548 (55.48%)
compared to 10th try
Finished classification and evaluation in 9.81 seconds.
--- Evaluation Results ---
Total samples processed: 3598
Total actual 'BAD_USER' samples: 310
Total actual 'GOOD_USER' samples: 3288--- Counts ---
True Positives (TP): 236
True Negatives (TN): 2766
False Positives (FP): 522
False Negatives (FN): 74--- Rates (%) ---
True Positive Rate (TPR / Recall): 0.7613 (76.13%)
True Negative Rate (TNR / Specificity): 0.8412 (84.12%)
False Positive Rate (FPR): 0.1588 (15.88%)
False Negative Rate (FNR): 0.2387 (23.87%)Overall Accuracy: 0.8344 (83.44%)
How were the actual good and actual bad users marked?
a bad group is defined as a group with >= 0.75 crowd score and at least 2 votes. a bad user is defined by being a member of at least 5 bad groups. a good user is defined as any user that isn't a bad user, which may sound like an issue but there's almost 9m users in the db and if you sample them randomly the odds that you'll get any bad ones are very very very low
How do you distinguish between true and false positive though? You're going onto their profile and making inferences from it, then comparing to group membership?
what
What is the definition of a true positive and a false positive
make an inference on the profile text data, compare this to the label derived from the process I just described
oh, you mean positive/negative
"positive" in this context means a bad user
"negative" means a good user
I mean the true/false binary
true positive = user is considered bad, classifier marked them as bad
false positive = user is considered good, classifier marked them as bad
true negative = user is considered good and classifier marked them as good
false negative = user is considered good and the classifier marked them as bad
Because when I tune out false positives I generally want an objective answer as to whether it was actually true or false
the data is admittedly not perfect, but no method i have of classifying the users (short of just sifting through like 8k+ users myself manually) is perfect
Yeah ok, just wondering
Just thinking you're comparing two automated ways to determine good/bad so idk how much error can mix in
If you classified user accounts separately with a high degree of confidence by humans, you might end up getting a more robust model
Or assume anyone who is banned must be a true positive
I'd like to expand into that at some point but
the issue is that there are so many user nodes
and human voter attention is limited
combined with the fact that I can't trust any one voter to be consistently correct or even trustworthy, it makes classifying user nodes with human work much harder
Yeah maybe
I'm used to being able to make someone directly investigate and know with 99%+ certainty
Although I have more classifications to play with like benign positive
what I'd like to do is
ok
the very first run I did with the classifier
I included their friends list usernames in the text to classify
and also their group memberships
it scored 100% on train and test sets
the only issue is it's a massive pain in the ass to ensure there's no contamination between the training and testing set, as the social graph here is quite dense
I'm wondering if you can make inferences from equipped clothing on their character
Do they have unique IDs you can grab?
I'm still curious about the cheese pizza shirt I've seen some of them wearing, but idk if its just a really common default outfit
So I'm wondering if characters wearing it are more likely than average to be in a PDF cluster group
Did you exlude banned groups from the data?
ooh that's actually a really good idea
I can't figure out how to get the actual fucking avatar image URL
because they changed the API and now it's much harder for reasons I don't understand
but if I could get their currently equipped clothing and then get the names of those clothing items that'd probably be a rich source of data
Yeah I dont think you need to see it in its totality, the combination of certain sussy items probably would be noticeable
Like if they have a "bypass wedge" equipped that's not easily visible
no, I didn't exclude them because I don't know how touchy roblox is in regards to banning for non-sexual purposes
like for example I know you can DMCA people if they copy your group's clothing without permission. do they ban for that? I don't want to train it to recognize "member of a group with stolen clothing items" as a bad user, you know what i mean?
try to infer from the group title when possible but generally yeah if it's banned it's almost always for something sexual at least in the data I've ggot
"Content deleted" as entire group information is junk data if that's the system's group data. A datascraping code could find those banned groups, and groups that do that to trick roblox mods.
I am actually sending reports directly for PDF groups so idk how fast they act but it could result in more queued groups getting banned
Ruben did some investigating in regards to this and found that there was no clear connection between sending reports and groups getting banned
the observability of the space being tested is problematic
you don't know how many other people are/aren't reporting the groups
but there wasn't an obvious connection between making reports, even multiple reports from multiple accounts, and the groups actually being dealt with
Hmm
I wonder if the reports are too short, too long, poorly phrased, too matter of fact, too emotive, etc
I know from experience that the framing of a "problem statement" can have a dramatic impact on whether an engineer will decide to pick it up, regardless of its actual severity or completeness
So the same logic may apply to teams dealing with reports
Maybe leading with a tl;dr like "PDF group with multiple references to CSAM" would grab attention
Obviously not with the acronyms
shouts out to @storm star and @rose hemlock ruben says thanks
grand_larsony lol
Where do you chat to Ruben anyway? I can always reach out and add any extra insights I made in case they are useful
Kiwi Farms is where I chatted with him but he's also on X. The screenshot is from Revolt which is like an alternative for Discord, but the server I'm in with him isn't ready for public linking yet. He just got banned from Discord because Discord is complicit in this Roblox stuff continuing
Ah I see
I assume Roblox doesn't want to lose all the revenue that PDF files bring in?
They must spend a lot more than most users since they're adults
I think it's sort of this issue x2
Roblox doesn't wanna lose their moneu
Discord doesn't either
They've both made pretty public opposition to Ruben and it's kinda crazy that anything he's doing is enough to warrant either of these giant companies' attention
Its interesting how they seem to frame all of his attempts to prove his claims or prove issues with moderation as evidence that he himself is a PDF file, deviant, bully, etc
Your message has been deleted, sorry!
Next time, try saying something like this instead: There's a collection of images on KF that I need to locate and share later where a similar pattern keeps occurring: 1. An upset furry who has a fascination with Roblox tags Ruben, attempting to accuse him of something serious 2. Ruben reviews their post history 3. Discovers a variety of unusual interests (like sonic inflation, furry vore, and other peculiar content) It's quite predictable.
You have been muted for 2 minutes for the following reason:
sexual score of 0.86
I hate you
I do think there are plenty of unintelligent people who see other people testing filters and trying to fine tune moderation and are genuinely unable to grasp that they're not PDF files themselves, and genuinely are trying to catch them
Granted there are "PDF hunters" who are hiding something themselves, but we've seen that a lot with people freaking out at moderation things being tested in discord or failing to understand why you would test something in an adversarial way
I hope you all have fun funding 0 predators in roblox
It's not the platform to successfully find kids
Thr architecture of the platform makes proper engagement difficult
This isn't the right place to shitpost
Troll somewhere else
I understand you're trying to be funny but this is genuinely very serious
https://www.nbcnews.com/news/us-news/kidnapping-roblox-rcna201795
Florida Attorney General James Uthmeier issued a subpoena Wednesday to Roblox, demanding it share information about its marketing, age verification and collection of childrenโs data.
โWe intend to cooperate with the Attorney Generalโs office and look forward to sharing all the work Roblox does to help keep users safe,โ a Roblox spokesperson said.
New Jersey Attorney General Matthew Platkin announced a lawsuit Thursday against Discord, alleging โdeceptive and unconscionable business practices that misled parents about the efficacy of its safety controls and obscured the risks children faced when using the application.โ
"50 m, got gun"
๐ ๐ ๐ ๐ ๐ ๐ ๐ ๐ ๐ ๐ ๐ ๐ ๐ ๐ ๐
lmfao
@worldly orchid
Do top users by classifier score and filter out banned, then open their accounts
And look through
where's that
go to "stats" tab and check out the lists of users
"roblox would never allow bad actors, they have moderators ๐ "
the ppl talking to ur child:
@lyric garden
!!!!!!
๐๏ธ๏ธ Command "!!!!!" not found !
Unbelievably fucked up update, unrelated to my own project but Roblox related so I'll post it here
Your message has been deleted, sorry!
Next time, try saying something like this instead: There is a user named "Kush" who is reportedly connected to a troubling group and has a bounty placed on him. Allegations suggest he was involved in inappropriate behavior with an 8-year-old girl and coercing her into harming a family pet.
Myco Is Magic! 00:12/00:13 May 21 Channel created Channel photo updated 56 11:52 PM Myco Is Magic! This will be my new channel until it gets reported. This channel is rated R, X, and F. If this channel reaches 500 subscribers before the end of summer, I will host a 1-hour Q&A livestream. 66 11:54 PM
May 22 Myco Is Magic! The following bounties are currently active: -Kush $1000 -Krane764 $100 -Information regarding the parents and account of the 8-year-old Roblox girl Dawn 1378, who was coerced into harming her pet $100. If you prefer to remain anonymous, you will be. If you want to be acknowledged, you will be. Contact @TheGrubs for more information. 193 edited 12:06 AM Myco Is Magic! Myco Is Magic! The following bounties are active: -Kush $1000 -Krane76 ... All bounties will be updated or reposted weekly 61 12:06 AM.
You have been muted for 2 minutes for the following reason:
sexual_minors score of 0.52
tf is bounties
someone is offering money for information on the guy doing this
did you report that
ok so it's not a pdf channel
that's good
that's realy rer tho
๐ญ
ive been following the stuff about 764 for a while and it's so crazy that if u tell people about it irl they look at u like u must be a schizo
wtf is 764
oh the guy
๐ฆ
764 is a sextortion cult
๐ญ
satanic pedophiles (but like, actually, not some schizo qanon bullshit)
crazy
I wonder how many of these people are under 18 themselves? It seems pretty common
when I was reading about 764 I encountered this a lot
there's obviously a line where you become an abuser, but oftentimes, there is significant overlap between abuser and abused
I guess that's probably the same as it ever was though, we're just seeing it more starkly because 45 yr old 'edophiles probably aren't as technically inclined and are thus less likely to be in a place like roblox
Holy fuck that first one
I bet it's not uncommon. IMO (my personal ethics say that) once someone is between a couple years post-pubescent and their 20s, they're usually in the "abuser that can recover with therapy" zone (assuming it's not something really sociopathic). Someone targeting a ton of kids serially at 17 seems way past that though
It's not like all the abusers were "victims" in the way you normally think about it, just much more likely to have had atypical childhood exposure to sexuality. Groomers intentionally do the same to normalize such behavior or countermessage anyone who's told them "that's wrong" or "that's for adults only".
I was already precocious enough, as interested as I was in teachers or, like, Cindy Crawford (don't judge me ๐ญ), I imagine that if my first relationship was with a freaking adult, it would have broken my brain for life.
That's what's so exceptionally dangerous about this shit, it's can have a generational cascade even if people don't have subjectively "bad" experiences
So like Johnny used to make this argument, "oh, you can just avoid the weirdos online if you're a smart kid".
First off, that's retarded. Secondly, the problem is that even if you're just flirting with creeps or trading pics or something, say if it's even just a teenager with a pre-adolescent, you're blurring the lines as to what's a "fair" relationship where two people both actually understand and consent to what they're getting into and creating a high risk, one sided "power dynamic" (I know, yuck) where one person convinces or educates (aka grooms) the naive one into what it means to be intimate. The less experienced person can internalize that and use it as a model for how they go forward, perpetuating a cycle of abuse or para-abusive behavior.
There's another level of this in queer culture that makes the whole thing a VERY hot potato. Gay kids can figure it out just fine alongside people their age of similar experience levels, if that's where they end up leaning. There's a long tradition there of intergenerational acculturaltion that is not necessary anymore and looks extremely predatory on its face
... I mean, I took advanced hours in adolescent developmental psych. Can you tell? Lol ๐ I was interested in doing this professionally, but didn't think I could handle a career of hearing these stories.
Yes, I think people get confused and assume that "troubled upbringing" means the abuser was abused too. I think sometimes its that they were able to experiment early and keep wanting to repeat that experience
for sure. And there often is a history of phenomena that are abuse-adjacent on all sorts of levels, to be fair. That might mean getting involved in drugs or being with an older sibling exposed to age inappropriate content routinely, having parent-child role reversal is VERY common (say you have parents that are ill or addicts or are emotionally unstable you have to care for, single parent household, you're the primary caregiver for a younger sibling). A lot of things can push you in the direction of relationships that confound caregiving and intimacy or where you have really unequal partnerships, in a serious, not a Twitter way. Personal insecurity is a big deal too.
I'm honestly bending my thinking more towards where you seek out, start and perpetuate inappropriate relationships more than serial abusers though. The kind of thing you can prevent and/or treat with proper care. Some of the more vile individuals were victims of terrifying abuse ๐ฌ
Bring this chat back to life
yes it is since you joined
wake me up (wake me up inside)
i can't wake up (wake me up inside)
I saw this video of this guy singing it with his daughter and it was really good
Bring Me To Life Cover By @bobbyamaru
https://old.reddit.com/r/nottheonion/comments/1lvu2xb/roblox_ceo_continues_his_pitch_of_roblox_as_a/ im sorry what the fuck
roblox is now threatening content creator Schlep for being a "vigilante" (he got 6 child 'ators arrested) and they're saying they're gonna try to have him prosecuted under the computer fraud and abuse act if he makes a new account
really cool stuff, great job roblox
Roblox moment ๐
@merry yacht Whats likely going on is that while roblox housing predators is seen as a big thing in roblox getting predators arrested makes their credibility worser as it shows their systems aren't safe and you could be doxxed I also don't think this is from the top leadership but most likely from a team outsourced to India that's thevmost likely case
I agree, Roblox would definitely prefer everybody just shut up about it. Pricks
While that's true the top leadership doesn't do stuff like this which had a very different intention then teams outsourced to foreign countries who want to fulfill their quota by any way possible by looking at roles
I disagree, the company has made their stance on this issue clear for a while now. this isn't the first content creator they've threatened and in at least one case, they actually did take one of the "vigilantes" to court over this
The vigilante framing is unhelpful as it implies they are aiming to directly confront or harm the pdf files
