#Neuro ARG
1 messages · Page 82 of 1
as mentioned above, isn't that code just the step by step on how to decode [Filtered]?
filtered code was decoded with malboge
kinda, it also ignores spaces which is useless for [Filtered]
or whatever it is
tho ignoring spaces could be an aesthetic choice
yeah, the malbolge was found basically with this, with "char" being the title, no?
yes
nope
duality of arg
.
Char > Unicode > shift until "space" > UTF-8
oh
yeah
and the malboge script was an incomplete hello world
the last part of the script was the key
Seems like too much dark magic to be a thing
the discrepancy is there tho and has to be explained somehow
its probably not a coincidence
(it could be, but 13/16 of the digits being the same is very unlikely)
Yes, but the difference being either a random subtraction or random XOR doesn't make sense (yet)
either the process or the numbers must come from somewhere
or maybe this is another leap of faith
exactly, the discrepancy looks random, and we need to find the source of it to make it not seem random
Guys stop torturing yourselves please. It hurt more than just you. Just give time to time. You can't force a seed or an egg to grow faster than it can. If you force it, you'll end up destroying the seed / egg. Understand who want to understand.
Heart~
You can't spell schizo without removing pain from spain
therefore to schizo we must ignore the pain
Do you understand what you said? (Legit curiosity)
don't worry, we like doing this for fun. We can leave anytime we want ||
||
what counts as understanding
Nvm < 3
I won't stop any of you anyway. I just tried my best. Well have fun~
(Edit: And they kept going without considering the warning in the slightest. Adding more weight, pain and pressure. Not even considering what might happen

)
anyways. There might be one layer below even the numbers 2 key that we aren't getting
btw i read the docs and i get we have a bunch of clues and stuff but do we have any sort of a timeline going on?
Like, if there is a 13/16 match, maybe there is an underlying method to generate the key that isn't chaotic
Like a progressive hash or something
There's this: https://docs.google.com/spreadsheets/d/1BCpJA-2GlmjzrxFIEA2pq5Xt6QuAaUI-2L_iAcAnl6Y/edit#gid=521227060
So maybe something like this:
<some underlying thing, maybe a string> -(some unknown process)-> Numbers 2 key -(generate a bunch of fib numbers, shuffle them, remove matches)-> Numbers 2 lyrics
wel ye but thats like order of video releases, im talking about like a possible timeline of events that may be happening
Oh, theory-level?
yes
I'm sure you'll get a hundred different interpretations depending on who you ask, so there isn't a timeline everyone agrees on
I'm assuming you've already seen this? https://docs.google.com/document/d/12GcN8a0Q1_etDfDi23SdLjAX6MsEZ6XhCYqTNyO3M9o/edit (ARG-wide theories)
i'm trying another thing
specifically, extending the numbers 2 key
nobody said we have to take 16 digits from it
that would mean you'll have a continuous stream of fib numbers towards the end of that key
hmm, maybe it will mean something 🤔
ye
i tried to figure out if there were patterns to how the underlying skips were generated, but i fail miserably at pattern recognition
^ this is in the context of extending the key, cuz it did cross my mind once. didn't figure it out so i dropped the idea as quickly as i thought about it
anyway the digit distribution is 000111111111233344444455555555556666777778888999999999
looks relatively ordinary, if a bit lopsided
is it lopsided enough to figure out if a string is likely going to be something fib-related?
i dunno how distribution of digits in fib numbers work, i imagine it tends towards equal distribution (inb4 misinformation)
im afraid this is as good as it gets with the sequence from numbers2
(you can remove the 2 at the end)
otherwise, theres another 2 very soon and the number doesnt have any 2s
UNLESS after numbers2 theres another sequence we have to remove from the fib sequence
☝️ i thought this was what you were doing
nope i was taking what was left in the fib seq after num2
actually i've only removed a single match
let me remove all matches
(idk how much it will change)
there's also the question of what to do with the numbers we get from the gaps
(in total that's a gap of 10)
split exactly in half? i tried doing stuff with that idea too, but didn't go far
stuff = turn it into hex, weave it with bbce to see what i can do with it
ok there isnt an early match for that in fibonacci
could be rearranged 
ofc i did rearrange it
anyway heres a digit distribution for small subset of the rest of fibonacci: 0000111111111122222222333333334444445555566666777788888899999
its roughly even, with little 0s and 7s
👍 so it tend towards equal distribution
so assuming this is removal of digits from fibonacci, we need to log for numbers with much more 2s, 3s, 6s, 8s
so the 6920 number wont work, theres no 3s in there
might be a botched sample though, surely there are too little digits to tell
no we will have the start of fibonacci either way
my sample is after the removal of numbers from numbers2, and after removing the first 16 digits
ah, ok
anyway i dont think the chance of
01111134455667789
01111134455677889
being that similar is very high
so this could be a deliberate hint, or it could be an accident, and i think theres no third option
right, the third option is a troll
27 is bb in base 26 assuming b=1
anyway the letters interpreted as base26 are: [27, 56, 497, 201, 2853, 4215, 320562, 2453, 212, 43, 382568, 1, 12212]
plenty of 2,3 so i guess it could work in theory
ahhh another 201 
nope no close matches
(for 2756497201)
i mean maybe shorter ones could work, let me check
nothing for 2756497
and of course nothing for 2756
27 does match, but very far away
what are you working on rn?
the decoded hex number+bb ce string
😮
@safe monolith can you make it so arg-monitoring makes a new commit on each update instead of rewriting the branch? that would make historic tracking easier (especially if you use ident=' ' and sort_keys=True for the json)
sounds good, will do it tmr
mhm, probably better to make a commit only if anything actually changed
in theory the feed XML really only updates when there actually are changes, and should also contain the history. but i can see where having git history will be useful too
probably can re-arch it so it uses that git branch as cache rather than relying on runner cache
mhm, either way feel free to take your time, this isn't high priority at all
got scared for a second
thought it's gonna need another reverse numbers
but no, all the ends are 25, 27, 39, 3b
plus there's 15 different hex digits
i'm trying to somehow fit this number into the fibonacci sequence but it just has too many 9s
added some "further analysis" notes to the google doc and i'm done here for now, have other stuff to do
happy 80000 messages in this channel!
85 more!
i think i have in mind a general algorithm for finding potential ways of fitting a number into a fibonacci sequence in parts in O(N^3) which is good enough, but it doesn't sound like it'd help
either way something to do later
also added this to arg-wide theories
just saw this on the subathon intro
does this mean the white hair girl is study-sama and the bottom one is hiyori?
??
wait i think i heard a hidden spectrogram or morse message in it
ok time to work i'll just put the subathon in the bg (clueless)
also this here hiyori and study-sama are separate too

well i mean there is something
but i cant tell if its a message
then this
at 26:41
Hiyori isnt an ARG character, its the name of the public model vedal used
many people see it as a character, but it was never official
true, but vedal using hiyori's model everywhere in the arg doesnt help lol
slightly higher quality
and it doesnt seem like morse at least
the beep length and interval is too varied
plus the shape is weird, probably part of the audio
these were caught realtime as well. We concluded there can't be any clues for a few reasons:
-
2nd part was entirely fan-made and was given to Vedal fairly last minute, so any lore drops here are entirely fan-cannon
-
Roundabout part, visually doesn't have much except for cameos. Audio wise, it is not that well normalized/balanced, as someone in #programming has pointed out (I think it was nach). So you might be hearing & seeing artifacts
no that audio definitely isnt artifacts (its too well defined for that) but it's not a message either so whatever
either way just added a couple roundabout screenshots to the stubs document
Well, artifacts is probably not the right way to put it, but nevertheless I'm trying to convey that the audio is too chaotic to meaningfully store messages
yeah exactly
for the past week or so i've put it in the background on loop as i work
withdrawals are real
its all weird really
cuz that makes study, censored hiyori or whatever
a changing entity
What Are Yall Talking About?
the neuro arg
the neuro arg
numbers
the neuro arg
[Filtered] talk
we crossed the 80k messages mark
who's mark 
Hello everybody my name is Markiplier
another 80k to go surely
computer, enhance.
might be pointless but did someone try typing 7777 2 888 33 0 6 33 1 1 1 into the phone number call?
save me? someone probably did but if its not on the docs, probably worth it to try
lemme do it now
nop, same voicemail message and call length
thanks
trying to collect every possibly relevant transformation of bb ce
because so far we got nothing
anyway if you have some fresh interpretations (translations, applying some algos on them, whatever you call it) of bb ce or the numbers please add it to the doc
if you stare into the numbers long enough and they dont make sense you havent seen enough numbers
No idea what to do with the actual letters, but like
Meaning of life has two "lines of stuff split across two lines that you have to merge together", maybe that's the case for the letters too?
sure, I'll add that too
just added a new algorithm to the toolbox - it's like vigenere but without wrapping around, it only shifts down; thought that might be useful in the future
that didnt get anything useful out of bb ce at least
(as a bonus, if you enter a single character as the key it shifts up a little so you can check if a small up shift can decode it)
After it’s release every cipher we had was solved in under 24 hours
welp

attempting to figure out how the rotation gets decided
from what i can tell, after restoring the correct rotation there are some afterimages at offset 125
which is either an intended afterimage, or more it happened after rescaling it
if so, this means the afterimage is at the boundaries of different rotations
which is actually true from my experience
Lucky you guess who made a timeline of arg events
(unfinished)
which means if i can detect duplicate images at offset 125, i can detect line boundaries
oh there are also -125 offset afterimages huh
experimenting with an algorithm to automatically align the image, and it can even assemble the top part
getting closer
slightly worse text but the picture is restored to an impressive extent
Dude are you okay? Are you doing that non stop?
pogs tho btw
im at a level when i can recover each text separately using some code tweaks but not both
here we go
the static in Study definitely isnt random data
but it doesnt necessarily have any meaning, no idea if it does
bowomp
oh but i guess that might be compression artifacts
this?
ah fair
the korean wiki doesnt have original research
i noticed
but isnt that part solved?
we got
HFq9U0g7+Hm7yzvl4AgGDCUUe9nQVUpaecL7jJhqDaLG2NE5bO2ysHoiP5usBmDxD5uJ9139vATYZK3BZF12nqP3eMyqCefav0Yv54eqBSNWFIEzL7gK4tEizNoAH9n8Zso33/PexoyP2l7DE wtinIDLYNWQqp5JYk3DUG3tj/Ms5t4XtBe8NsHPaoS+DcYKmEwMpBjEv5LAf7K1VSH97OW5n1ZhpM3lPRGcw8c2AOMkM2NhRM37WELqHS+piD766dzzED/mJFlhSRqbWPsC0Ul55B6Rxk7npAnH8 Tx4ftSBavqGwyFJFIktxVQ6Mmd69HTyeLYO4WUItQ+KkNpczIVLLPFuSiFcaBxLuw==
thats one decryption
what no
- it has wrong pkcs-7 padding
- even if it was correct, any key can be used for aes, and if it doesnt match you will get random gibberish - thats what you got there

this is a hint for goth neuro
the fact bb ce starts with bb is really weird, because the same letter repeating means that if this is a simple substitution cipher then the same character is here, which means after decoding we won't get English, which means we can only guess what's right
seems weird
on the other hand if it's a polyalphabetic substitution cipher then there seem to be too many options
has the 692048501258949201 already been used for something
none of the things in meaning of life got used for anything yet
mostly bc nothing really seems to fit together
maybe its a snowflake that was made on <t:1585067620> (using https://snowsta.mp/?l=en-gb&z=k&f=gns4cm92h1-xd)
I almost forgot that I did this, I’ll post it here. 
No, I won’t make games, I’m just good at creating TTS and videos 
how were the decimals obtained?
i cant figure out how to recreate them
the upper numbers are in hexadecimal, the lower ones are just the same number in decimal
oh i see
I'm kinda surprised Vedal didn't mention that there was an ARG to Toma yesterday when they were discussing if there was Neuro lore
he generally doesn't talk much about it
iirc its on purpose as well
saw a clip of toma asking vedal if there is a neuro lore video / neuro lore on youtube and i was tempted to comment that the lore videos are named Numbers, Study, Numbers 2, Publis Static Void, hello world!, [filtered] and Meaning og life 
To be fair it's kind of a separate thing from actual neuro
well it both is and isn't
because ARG IS neuros official vtuber lore
at least that's probably what it is, I see no other purpose for it
lmao
Oh nice to know
My theory why he didn't mention is a) because he forgor tutel style
Or b) didn't want more eyes on the arg
If it was on Live it would be Funny 
What a menace, pinging people from over half a year ago
are they trying to read all 80k messages
dedicated arg enjoyer
truly
it can still be story-like without directly relating to Neuro though, that's what it is in my eyes 
yeah, I was like.. what the hecc. When did I say that? (get pinged too
)
a blast from the past 
Why does it even exist then?
it must connect in some way
I don't think it must anything until proven that it does. It's a creative piece of work. It can do anything it pleases
that certainly is true but again if it doesn't conenct to neuro in any way why would it exist?
actually hello world, hiyori is already a proof it relates
to create hype, of course. It's a marketing stunt, and a pretty good one too
I find it doubtful
I mean like, QR code is on neuros stream
the channel is named _Neurosama
it would be quite deceiving way of marketing wouldn't you say?
i dunno, other people's (e.g. wilbur's) ARGs don't really connect to the actual content creator. it's just an expression of some story that may not be fully related
sometimes cameos of ARG are made in their actual day-to-day content too, but their creator doesn't revolve around the ARG or vice-versa
isn't that the opposite of deceiving? It says right on the tin what it's for. It's not like those args that present some mystery and then does a switcharoo for an album ad or something like that
yeah, "alternate" in "alternate reality game"
so like it is entirely possible that it just kinda exists as its own thing. probably can't just conclude that ARG = backstory of neurosama the ai streamer
Vedal himself said that neuro basically don't have lore other than she's an AI
but fact that AI itself doesn't have any lore doesn't mean there isn't lore behind it
or rather there couldn't be
you said a true statement, but it doesn't explicitly prove that the ARG means anything
there are a lot of open doors still, but we cannot conclusively walk through one of them
That's true I guess
estonte 
but hello world for example explicitly speaks of neuro. Why would she be there if she's just for brand? it would be like stacking two boxes just to call it a tower.
Yes I guess my case isn't strong enough with more evidence, but personally without that narrative I just don't see any set path that makes sense
🤔 i'm dumb, does the cipher output mean anything (yet)?
Gosh I need to sleep, I think I had too much reddit today
ah ok. i asked just in case there was a message in plain sight
3 is such a recurring number it became a meme
like, for example, the very long sequence of just nothing but a wobbling picture of a house?
I find it odd that we've only gotten one clue from the rotated frame. When I unrotated the entire frame, I noticed some black and white squares at the bottom. They may just be relics from shifting the numbers, but why were they there in the first place? Kind of looks like they used to be vertical bars
it's not a stereoscopic image, is it?
I dont see what looking at it in VR would accomplish
whatever chayleaf is cooking, maybe it can rotate the entire frame by the intended algorithm
if this frame has an intended algorithm, that is
like, [filtered] had a more or less an intended way to rotate the image. surely this can't be a thing only done by manual rotation
Yeah, chayleaf's solution seems like the correct way to go about it
vedal plays vrchat in vr, it makes sense

"clues" 
i thought this was just cause we also had to rotate vertically to connect the numbers into 1 string

forget it
(if there was an update, i'll get an rss update in about 20 mins)
wasnt there a quote by vedal that went like "this extent to the past and future"?
becuase that video is before the ARG
in theory, it may be because of the vertical shift, it looks like instead of just the numbers the entire image was shifted vertically, as proof look below the bleeding eye, the jawline is at the level of the neck's base
and below the numbers the noise frequency is shifted too
he says that not giving us clues directly extends to the past and future
ASSHOLE streamer HATES his fans
i thought maybe bb ce corresponds to some specific numbers in fibonacci (bb = 0x22 or 55), but that didnt work
i think the idea may be worth exploring tho
at least it doesnt work out if you include both lines
U wot m8?
hex form of 141693817714056513234509965855411919657705794958199845 = 17ab6d825584b02042f00149f9d9fb51ec7ba650f0c25 = 45 characters
bb ce td ht eft ggd sgfi dqj ie br vtye b sbs including spaces = 45 characters
:schizo:
realistically it's a substitution/transposition cipher to get the 16 character key and then AES decrypting the base64

If this is about near
Even the original version has that sound in the end
yeah kinda expected that
I managed to replicate the soundcloud showing having some tracks and them not being showed on view
I had uploaded them as public
and after the fact privated them
cheers, this is basically unlisting the tracks then
so we know that there is a track in the soundcloud, now we just wait
more than likely we're working towards it
or that
is not really unlisted, in unlisted videos you can acces the video with the correct URL, this way you can't acces the song
even with URL
I'm assuming things here, but an account created just for the arg having one track should strongly imply it is eventually accessible
obviously
Strongly imply doesn't equal fact but nevertheless it is more probable than an accident
I want to say decoding the current ciphertext will eventually lead to a soundcloud link since there isn't anything else connecting them atm, but I could be wrong about this. This would be a weak assumption
More importantly is how the resources on soundcloud seems more sentimental than any other thing we decoded. Kinda like meaning of life
it's more likely that eventualy the song will be unprivated, because even with URL, you can't access the song, I tested it.
You can private but share a link right?
This implies you can keep it private but still share an accessible link
It won't be a direct link to the track I would think
We never understood the meaning of the black banner in soundcloud btw. The one with the first 3 sentences of numbers 2 lyrics
Accessible
The guys smarter than me (above) figured out the minimum we need to brute force, although I think we should try to properly obtain it 
actually it's not that long
only this is neede(after the name of the song)
s-Tbvb8al8KnX
in my secret link it was /name/s-Tbvb8al8KnX
Yeah, precisely. Could be a https://soundcloud.com/572943/s-xxxxx after decoding
Possibly after some number of decryption given the size of the current ciphertext
although the codebreaking is atm beyond me 
Damn, so the recording was at some point available but it was missed
Imagine if someone caught that during the act, would've been so confusing probably 
If this exact process happens again I hope that the rss feed I worked on will catch this
well, the first test someone did was upload the track as privated and it didn't work
Am I late? This says the track was removed
Not just you, can no longer access 👍
o yeah, I deleted it, used a very bad playing and didn't want people to listen to my shit hehehe
but anyway, the link should be like https://soundcloud.com/572943/name-of-song/s-xxxxx
If it also needs the name of song it'll probably be hard to brute force
Anyways it's probably better if we took the proper approach
yes, just with the name of the song you can't acces it, you need the s-xxx
and I also tried without the name only /username/s-xxxx, and it said that there was no track, so I guess it was taking s-xxx as the name of the song
given hindsight, a possible conjecture. Ciphertext was within ghost drop. Study was all about the audio. Plus listen with your headphones. Audio -> soundcloud is strong
fwiw it might also be using a shortlink
Maybe we get full ghost 
This is also possible
god please

did we win?
do we consider the voices taking over a win?
Imagine how confusing it would be to try and explain the neuro arg to someone that has no concept of what a vtuber is
I remember some attempts. A bunch of attempts actually
Cuz G4 was a note, so for some time people were trying to get a sequence of notes, among other things
In this case you don't even have to mention the vtuber
well its pretty easy to explain the concept of vtuber
sure, but i still get judgemental looks for having an anime profile photo. it'll just be less weird to directly explain what an ARG is
idk if its useful but i turned the notes into binary if that can help in some way?
It'll be helpful if you can paste your notes into the docs please 
who
you
are
is
was
i isolated neuro-sama's reversed voice in meaning of life to a VERY clean state btw, i'll send when i'm fully done
i did tho
Wow after a month we finally succeeded to separate the voice 
well mostly
Were the guesses correct?
and at 5:52
and at 10:21
(timestamps arent exact btw)
and at 10:46
and at 11:10
(it's weird that it wasnt that often before)
oh it's restarts more often it just sometimes has a different offset jesus
(i'll just assume that the loop offset isn't a key to some puzzle ok thanks)
let me also reverse this
maybe a cipher by lewis carroll should be used?
"it's hard not being able to speak or move properly" is one of the phrases
"i think i've hurt myself"
"i can't seem to get out"
"i think i've fallen down this rabbit hole"
those seem to be all the sentences, please confirm
"I've fallen and i cant get up!"
anxiety is my current emotion rn
i think for any keyword-based ciphers anxiety is pretty much our only keyword guess so far?
Very interesting
lewis carroll's key-vowel cipher didnt get anything useful either
for completeness's sake heres the memoria technica cipher
so i guess lewis carroll is a dead end
does "fallen down" from Study refer to the rabbit hole?
(whatever that means)
either way doesnt seem like theres any additional clues here
i mean maybe the answer has something to do with Study timestamps but thats a completely random guess
for now its probably better to wait for the next clue since we arent really getting closer
anyway we already got a lore piece for "I don't want to be an engineer" in Numbers II, this is apparently partial "lore" for "Fallen down", ??? got referenced in "is that someone knocking at the door??" in Numbers II; "I don't want to be human" is perhaps Numbers, no idea
Hello friends. I have updated the ARG monitoring tool to now come with a not-so-friendly-but-will-probably-get-the-job-done UI: https://neuro-arg.github.io/arg-monitoring/?fromDate=today&toDate=today
Any date before 19 Jan 2024 will result in undefined, because the cache history didn't exist before then.
It pulls from the same JSON cache used for the Atom feed, which can still be found here: https://raw.githubusercontent.com/neuro-arg/arg-monitoring/publish/atom.xml
Some other updates:
- I fixed a bug that causes all videos to have the same hash,
(check https://neuro-arg.github.io/arg-monitoring/?fromDate=2024-01-20&toDate=2024-01-21) - The
publishbranch now keeps commit history so that we can keep track of what was changed
(Sorry chayleaf, you've probably received a million CI failed emails
)
schizo theory time: the "fallen down" state is an ai disconnected from multimodality
zero emails here
idk if y'all see it but the enitre scene flows together with great intent in making it do so, and this stripe across the center there makes it seem like this is a timeline of sorts. Especially with how there is what clearly is a type of picture not featured here that is right above the left most side
this, and it looks like eyes looking into a cage
I'm fairly certain it was a teaser for the subathon intro.
I know I'm late, but I wanna see your thoughts on this discussion we had a while ago about the BBCE string that didn't get documented
just ignore the "solutions", those were accidental false flags
This is the farthest I've got following that logic of taking out all the valid Hex digits from the BBCE string

Comparing the two is a crazy good idea! But I'm not sure I see the matches clearly
A, B, and D all look similar, if distorted. A line that has measuring ticks, D has what looks like to be a complex image that could be a distorted version of the gathering, and B, the "eye" looks a lot like the tent if you monochrome it.
Mmm, you may be right, but I really can't see it
Guys can I do it? (add avatars to faces) 
Also maybe Vedal? The guy holding the phone?
bread is kind of the main guy of this channel (imo at least) so he should be holding the phone

True.
I'll count myself in too because I can
that's 5
3 more
darkeew perhaps 
2?
2?
Well, that’s why I can’t throw him into memes. Now I’ll skip the story and here’s the reason:

Numbers 
5 
9 or 5? 
95 
this number doesn't seem to match anything, so I can't really say anything about it
btw chayleaf, how did you accomplish the cleaner voice isolation?
audacity
with the noise removal plugin?
no
- copy the very first loop with no voice
- the end of the first loop still has some voice, reconstruct it piece by piece (it isn't reconstructed 100% perfectly because full reconstruction is kinda impossible or very hard, but it still reduces music volume)
- invert the first loop
- copy the first loop until the end of the video (34 loops total), adjusting the offset on each loop as necessary
- reverse everything
ah, then reverse and cancel?
invert and reverse yeah
btw the loop is 8 bars/24.6 seconds long, not 4 bars/12 seconds long
as you can see in e.g. the hello world! video
hey, can someone update the drive about how the track being hidden was replicated and the conclusions about the link that needs to be discovered?, I would do it myself but 1.- ESL and second, I really really really don't want to fuck up something.
Sure, I'll do it. Thanks for reaching out & sharing your concern
(We won't bite you if something goes wrong, GDocs has revision history for a reason
)
Does this convey your intent?
yes! thank you!
imo a new doc needs to be created for soundcloud because it's bound to get deeper and numbers II doc is big as-is because of the 3 steps required
You seem to be writing perfectly fine though. I'm also ESL and I've written a fair bit of the docs. I'm sure native speakers can tell but I'm also sure it'll be fine (:
what is english 
anyway, about creating a splinter for SoundCloud, agreed in principle. I'll do it when I find the time
Yeah, I already wanted to split the soundcloud into a new doc when it first appeared. It might be time to do so soon. I definitely won't protest if someone does it now
just noticed the 6s are all at equal distance, probably a coincidence tho
(and there's no 6 at the end)
testing in prod, posting random id amalgamations before trying it out first :p
and it's gone!
come again another time 
discord has two "slots" of 19 digits (channel_ID and forum_ID): Neuro ARG forum for ex is https://discord.com/channels
/574720535888396288/1112082183235047506
figured it'd be about this, yes
oh right, the format isn't correct, i see what you're saying
the snowflakes will not always be 19 digits
so probably no reason for testing if it's a discord channel or channel forum?
if you take the timestamp part of the snowflake and search here fromm mstuff that happened on that day all you get is this
there's always a reason to try something random (:
it might be a snowflake for something, but its hard to tell what
i've already checked if its a discord user and gotten nothing
maybe one of these is a clue 
I can't read 
2020 is way before the arg so I don't think it's a snowflake, but maybe there's a custom emoji with that id?
Tried doing it in reverse?
doesn't look like there's one
that's an even smaller snowflake, only 10 months and 12 days after the discord epoch
same numbers as verdal, surely the creator of this website is a clue
same digits, but not the same order 
Lmao what's going on 
i have decided that as angel numbers did not work, i shal look at more random numbers
oh god, not the angel numbers again
By the way, a fun fact about me: I live on the seventh floor in the seventh entrance and was born on the seventh day. 
One day bred is gonna write their thesis in angel numbers and discover that it is actually the theory of everything
i especially like the part where every website has a different definition for each number
He just planned the ARG for that long 
i just scrolled through every new message since i last read and all what i gleaned from it was that there is an unlisted soundcloud song now and that someone has a version of the bb ce image with a purple background for some reason
also that unlisted soundcloud songs can't be accessed by their links
so i'm not sure how we would be expected to find that
???, if you have the link, it CAN be accessed
Fair enough.
only if we know the track name
Soundcloud API response:
{"avatar_url":"https://i1.sndcdn.com/avatars-tJYw2AXK2yPy5Npf-vuQzMw-large.jpg","city":"\"Open\" in FireFox","comments_count":0,"country_code":null,"created_at":"2023-05-29T01:56:03Z","creator_subscriptions":[{"product":{"id":"free"}}],"creator_subscription":{"product":{"id":"free"}},"description":"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","followers_count":96,"followings_count":0,"first_name":"Start from X. End with Y.","full_name":"Start from X. End with Y. Remove XY + odd one out","groups_count":0,"id":1258077262,"kind":"user","last_modified":"2023-06-04T07:00:58Z","last_name":"Remove XY + odd one out","likes_count":0,"playlist_likes_count":0,"permalink":"572943","permalink_url":"https://soundcloud.com/572943","playlist_count":0,"reposts_count":null,"track_count":1,"uri":"https://api.soundcloud.com/users/1258077262","urn":"soundcloud:users:1258077262","username":"_","verified":false,"visuals":{"urn":"soundcloud:users:1258077262","enabled":true,"visuals":[{"urn":"soundcloud:visuals:193670493","entry_time":0,"visual_url":"https://i1.sndcdn.com/visuals-001258077262-DUgObZ-original.jpg"}],"tracking":null},"badges":{"pro":false,"pro_unlimited":false,"verified":false},"station_urn":"soundcloud:system-playlists:artist-stations:1258077262","station_permalink":"artist-stations:1258077262"}
the track was uploaded on 2023-10-31 17:51:59 UTC, the following is the track list json:
{"collection":[],"next_href":"https://api-v2.soundcloud.com/users/1258077262/tracks?offset=2023-10-31T17%3A51%3A59.000Z%2Ctracks%2C01653590364&limit=10","query_urn":null}
now we can bruteforce the secret token without bruteforcing the track name (leaving aside the impracticality of bruteforcing over 100000000000 possible values)
(don't worry vedal i did hack a streaming service before, gaining the ability to download music before it even releases, but surely soundcloud is more secure right?)
01653590364 is seemingly the track name? It doesn't look like the internal API id because of its leading 0
more numbers :schizo:
the fact this starts with 0 surely means this is a substitution cipher right?
abgfdfjadge...
or aqfdfjadge
this is either the key to solving whatever's in the track
or something encoded using the same cipher as 69204
I think?
skipping puzzles 
that said
since it was uploaded on 10-31, I guess it may be linked to from filtered?
so how about we check [Filtered] out again
the audio this time
who knows, maybe MoL is gated behind Filtered
now that we, presumably, know the crib, surely it will go quicker
I can't find any note transcription in https://docs.google.com/document/d/1dL6lH_qnlXMgpIXyc02Q3qpB78_vWyUgKV_pEo7H_Cw/edit
jesus the right channel has very very high frequencies
and the left channel has echo
okay where my midi sequencer at
guess i'll use ardour hoping it doesn't crash (assuming it's even enough)
failing that, reaper should work
okay ardour didnt work, reaper or lmms it is
Yes hack SoundCloud 
We can promote the arg by making on to the international news
so true
though that stuff really really doesnt make the news
the bpm seems to be roughly 108.7
does maybe someone here do at least some music production?
you'd transcribe it way faster than me
also I'll be honest I don't have the time right now
but imo the notes are key to the soundcloud, so please someone do it
the soundcloud may be key to meaning of life
01653590364 is the track ID. At least, 1653590364 is. Represents pagination after a certain ID.
If you check this guy: https://soundcloud.com/futureisnow (no idea who this is, just an experiment)
The response looks something like this (truncated):
"release_date": "2022-04-29T00:00:00Z",
"uri": "https://api.soundcloud.com/tracks/1258109134",
}
}
}
],
"next_href": "https://api-v2.soundcloud.com/stream/users/7436630?offset=2022-04-29T04%3A00%3A00.000Z%2Ctracks%2C01258109134&limit=20",
}
You can see that it's tracks,01258109134, with a leading 0. The API does indeed leak when and the track ID that was released even if the track was privated though.
never mind that isnt the track name
yes
sent at the same time lmao
either way the date is known now
so it's probably encoded in [Filtered]
For what it is worth, I believe track comments are leakable too (got a 200 hitting it with the above track ID)
although there aren't any comments for now
still require secret token
oh
surely vedal didnt make it so we have to actually bruteforce
whats the secret token like?
probably a 64bit key
https://soundcloud.com/572943/<track-name>/s-xxxxxxxxxxx
It shoud look something like this. You can find some docs about it here https://docs.google.com/document/d/1oBEX9j0Z32UQtVnnojsNFDCIz21R7LnyuoSp_2Uila0/edit#heading=h.7zsvtpwbq26p
so we need track name
Yeah (unless? we should try ID as an experiment to be safe)
can you change the name once its uploaded?
I believe you can
so we can use the track id as track name
we cant
The uploader can intentionally do that, yep
Uploader can intentionally rename the track to the track ID
i mean yeah thats true
if it changes the URL too, its a possibility
but anyway point is we need the url which again i think is in the [Filtered] video because of the upload date
hm
its like 11 characters which is impossible to bruteforce
we havent solved the description
it used to be 5-6 characters which was possible
we have bruh
no, im not talking about the key part
what then???
the function
the function doesnt need to be decoded
shift_characters
it isnt a cipher
i mean it may be used in the future but just forget about it for now, we try it enough as-is
(while this discussion is going on, i added a small note in the docs)
dont worry about it for now, enough schizoing if [Filtered] audio is the more obvious next step
not exactly
the audio likely doesnt lead anywhere
if it had a clue, we would have gotten it already
notes?
music notes
(that is documented)
yeah
where?
it's literally discovered in numbers ii
the only thing "documented"
@wispy thistle #1112082183235047506 message
some efforts here and there. most attempts are not documented
sorry, i wasn't clear. I meant that most transcriptions were not documented
10 notes is not enough
at that point, it was
There's an impetus to transcribe it properly now, but back then we only needed enough notes to form a key
thats waaaaay more than 10 notes
Point is, there were attempts to transcribe the notes, so we can't really say no one did it
well i mean that there wasnt nearly enough to say "if there was something we'd have found it already"
I thought the secret token could be manually set to anything and any length by the user (not exclusively 11 alphanumeric characters)
, can you confirm?
afaik it is randomly generated
yep
Yeah, I agree with the essence of this. Can't discard the clue is where you are getting at, right?
yes
this is reset, not set
one could also try guessing the user oauth_token
bruh
We do need a full transcription. Can someone popular enough in #neurotic-neurons help us do a little asking 
"one could try guessing vedal's password and OTP token"
you might as well say that
i asked once
no no just ouath token + client id(1258077262)
"""just"""
you do know that oauth tokens are long, cryptographically generated strings, right?
What I mean is instead of guessing the secret token you might consider trying it in the API as oauth token
while testing
but I'm clearly clueless
how would you "guess" a like 60 char randomly generated string???
it certainly is an idea, but i don't think they're equal enough in length
If you had something to test with a matching length, cmon
well if i had random bytes to test, sure lmfao
because you cant just tell soundcloud "i want this string to be my oauth token"
and no, vedal isnt that schizo
but if you as the user had an oauth token you could encode it somewhere for people to find it right?
yes, and you would include a hint "this is my fucking oauth token"
yes sure
because giving random people on the internet access to your account is literally the stupidest thing ever
it's a fake user but ok
i also think that such a hypothetical oauth token should have expired by now
i wonder if the bottom frequencies are relevant
either way personally i hear a dominant melody playing on every beat
tho the left channel has an echo and imo its best to just ignore it
the channels' difference is very weird
it seems the yellow in the blue does matter and the solely yellow part doesnt
or only the top of the yellow part matters
or it could matter for all i know
might be a better idea to use some mp3 to midi converter
doesnt look like theres a loop btw
huh
does this mean each note got duplicated on the spectrogram N times?
if so looking at the top might be easier?
does this look like data? i have no idea
either way if nobody else transcribes it i'll do it... later
(tm)
this is how a test soundcloud url looks (ends with something like testtesttest but a real url wouldnt have that pattern at the end)
with a char's ascii id converted to column
with high id being low and low id being high...
proper version
imo it looks relatively similar?
the pitch range between low and high sounds seems relatively high in the audio too
i think semitones matter here unfortunately
the transcription above has zero semitones so i'll just assume it's incorrect
(side note who the fuck decided Do is C I always hated that)
(plus note substitution ciphers are relatively common in ARGs)
okay this is rough, i roughly transcribed the first 10s just to check, and
pr]4tPP\TS|ts}`eM{}xxy~
doesn't seem to shift down to anything useful
it doesn't even match "https://" because PP are the same but r] are not
there could still be some data down the line, but yeah nothing obvious at the start
not a malbolge program either (plus decoding it from notes would be literal hell)
i think it's better to clean the audio first
there are some harmonics that can probably be removed somehow
i'll try...
this will take time as i'm new at dsp but i'll give it a shot (cleaning the audio so only the base frequencies remain)
right now it's absolutely possible some harmonics are overpowering certain notes
to explain:
When a note is playing at frequency X, harmonics are playing at frequency 2X, 3X, etc
I want to somehow detect and suppress the harmonics (2X, 3X...), leaving the base frequency (X) only, which requires some simple math (it's complex but simple for me cuz i'll use a library) like fft
That's the plan, anyway, I'll see what I can do
this would make the audio cleaner, which can potentially expose the original message, if any, in a clean way
fft is a way to decompose raw audio into its constituent sin waves, or something similar at least - that's what spectrograms are made with, the software decomposes the audio into waves and shows the waves on a graph

i actually read through everything that got posted after i went to bed and good lord thats much 
accidentally made my own "arg" art lmao
anyway heres a custom spectrogram i rendered
the higher res i go vertically, the more res i lose horizontally
which is to say, i wont be able to change the audio itself, but i will be able to decode the notes if all goes well
im thinking of changing my approach, instead of isolating harmonics i will only take harmonics from a specific layer
I still love making GIFs like this.

pain, im trying to write an algorithm for marking harmonics with red but it doesnt match at all
rip
sounds like hard work, what with the harmonic overtones
Yes, I decided to do it again.

okay i'm done for now, didnt get much closer to cleaning it but that's okay
at least you can clearly see here that the bottom is a clusterfuck, and the above harmonics are still a clusterfuck (because of other harmonic overlaps or something) but somewhat readable
it is perfectly normal yeah
and it is what allows us to get the real melody of the cipher, if this is a cipher at all and not just random notes
hmm, I may have missed the clue that lead to parsing harmonics for data, or even how to do it
the clue is that the soundcloud track was uploaded on 2023-10-31
just like the [filtered] video
so i'm assuming we missed something there
and if we did, it's the audio
as to how - we need to get the melody and convert it into data
we can't really easily get the melody because of all the clutter, but imo looking at harmonics we could recover it
Right, you're saying that because soundcloud is a music platform, it's an indicator we may need to look at the music aspect specifically, is that correct?
that too, yes
both because it's music, and because the upload dates match
and because it always sounded suspicious to me
and because there's no progress in MoL anyway
it does sound weird but, could it be an image instead of a cypher? like what portal did?
has to be decoded into data either way, but yeah
example SSTV broadcast like used in portal 2 (headphones warning) https://upload.wikimedia.org/wikipedia/commons/c/ce/SSTV_sunset_audio.ogg
SSTV is loud and jarring, it's very obvious. There's not something similar to my knowledge that isn't also immediately obvious.
And take that headphones warning serious, it's the worst sound ever 
theres DTMF (phone keyboard sound) but yeah that doesnt transfer video
it's also very obvious tho
true
suppose we get a handle on the harmonics, how do we translate it to digital data? Like something we can actually use in ciphers
lowest note = 0, highest = N
then shift
potentially modulo but it depends
(i already tried converting a part of it but simply by ear)
it's a bit of a leap of faith, but we've had to do those before
Ok yes, I see it now.
That the youtube video was publish at 2023-10-31T18:47:56Z and then that the private soundcloud track was uploaded at 2023-10-31T17:51:59.000Z insinuates that they're connected. It could be that [filtered] contains a clue that'll lead us to whatever that track is. Or a direct link.
I'm a little more hesitant to say that it has to be music related from this, but considering the content of [filtered], it's definitely not outlandish
it's just that we already restored the video and found nothing
we haven't fully restored it yet, though
there could be as little as one frame that contains it all, we don't know
i can't tell if the house bit rolls side to side or not
there's too much noise on the lower pitches of the audio of "Filtered" to make any sense buuuuut
there's clearly a pattern on the higher pitches
is still way to analogue to have a single pitch for note but it's a pattern
for example there's a clear line that is all the way trought the audio that vibrates between D and C#
Meaning of life? I think it does in that the camera moves
i sorta want to track it but on the tother hand its exam week
god having actually working headphones in incredible i can actually hear neuro yabbering in the background of hello world
And so I learned to make GIFs even more interesting.

also the ending is still so annoying, i have no clue where to start but there's clearly something and i hate it
well, there's the Y = 128 there
yeah but im lost as to what to do with it
go backwards
I have added a bit of context to the notes on the the soundcloud pagination-leak and the date from that https://docs.google.com/document/d/1nQhgVPmEAeg5HmJoo6Z46H6IRHGrbjFAorAVkIGeSrw/edit#heading=h.2j7f4c1oenih
there's way too much stuff about the soundcloud on numbers II, there should be a separate document now
Ah, owned by silver
Alright, that's the new soundcloud doc then
ok reverse engineering soundcloud is hard
my shenanegans are not working
how dare it give me a 404 when i try to like a song that semi doesn't exist
ok i sadly cannot convince soundcloud to like it regardless of whether i interpret the id from the docs as a playlist or a track
404 sounds like the correct response to queries on an entity you weren't supposed to know about
indeed
i can't think of any easy way to get it to play other than properly deep diving into how soundcloud works, which isn't on my list of things to do
plus it's literally not supposed to play without the secret. Like one thing is leaking the id and some software engineer going "eh.. it'll be fine. It's just an id.", it'll be a whole different deal if you could actually access it
true
based on observing the endpoint for a few minutes after checking the track ID, you actually need an authorized key to successfully play a track
the current client ID is either allowed to generated such a key, or 401
i was trying to change the track it thinks im listening to so when i refresh it puts me on a different track
but this is observation data and not api reference
who learns by reading documentation anyway 
i just realised i wasn't messing with this value, which might confuse it
having read just a bit of the soundcloud documentation, i can already tell that what's done in practice differs from what they've written
also this is all on api-v2.soundcloud.com
Here is (more or less) what is happening (based on observation):
/tracks-> gives the authorization key per track given client id/media/soundcloud:tracks:trackid/some_uuid/stream/hls?client_id=<>&track_authorization=<authorization key goes here>-> gives an actual playable URL (signed)
clearly the solution is to hijack the puppet master's cookies to get their client id 
did some super schizo looking at the spectrogram
572943
or thisv
(i did warn its extremely schizo, im just trying to figure out where to begin looking if this encodes a soundcloud url)
someone else made a video about the ARG https://www.youtube.com/watch?v=J4w3UUPQpsE
Что скрывается за личиной милой стримерши-искусственного интеллекта? Ооо, вы даже не представляете... И именно в этом ролике вы узнаете всё про ARG по Нейро-Саме, которая наполнена загадками, тайнами, страшным сюжетом и интересными персонажами. Вы даже узнаете, как всё это связано с Vedal'ом
------------------------------------------------------...
i've seen some other russian video about the arg released 7 months ago in related videos list

I'll see if I can find any AI translator for this video
There is an automatic translation in subtitles.
I watched his last Neuro video with auto subtitles, but I was getting lost quite easily (the subtitles appearing too late or not at all sometimes)
Yes, I understand
I watched only 5 minutes and already can say this is a really well done breakdown

yeah, his videos have a really high quality
Suprisingly youtube auto translate is doing decent too
definitely not biased but i love this
Well said...Well said neuro
And he mentioned my cyberpunk theory.
I. Am. Fucking. Speechless.

That was probably one of most random theories I ever had
Without even any solid back
Pretty much throwing out what came to my mind
On meaning of life alone (well, mostly)
The schizo really payed off in the end

At this time 23:50 in the video 
Static because you're on his discord. Could you send him my sincere regards along with thanks for mentioning a theory as schizo as mine?
And I had a problem with the Internet, so I decided to go to discord a little
@solid aspen
Oh pog he's here
So mr. Jofi I am highly honoured for my theory to be featured in your video

You know this looks like a New Year's gift 
I joined him on January 1, told about Neuro ARG on January 8, and the video was released on January 21.

Right after my sister's birthday. 
Nice
considering arg plot you may want to keep eye on her
It's okay, she's playing LOL
Do you mean league of legends?
Yes..
Thank you 
I'm gonna schizo and say there are 11 prominent notes on the filtered video (ignoring the ones that aren't as "large")
mmm, the image isn't loading

I could just paste an image earlier, what happened?
ok whatever https://imgur.com/a/ZmJheTp
the notes are extremly generous given that the sounds fluctuate a lot but it's the most "close" to them as they can be
also they're ignoring the other stuff becuuse thats fucking imposible https://imgur.com/nqQLNcq
well not imposible but a pain in the ass and super lengthy to do
you can see here better how they're the most loud 11 notes https://imgur.com/UjdjbPZ
this part looks really weird when played https://imgur.com/UdJlihN
I literally cannot read this 
jesus was there something new
Nah it’s just new people coming in with some bangers 

Door Study Girl Engineer Roommate Numbers 1 Singer Numbers 2 Singer Numbers 1 Description Numbers 2 Phone Number First 3 lines of Numbers 1 Legend: Black - Very Strong Green - Strong Link Red - Weak Link Line types do not matter Underline - Character Italic - Video Study Singer Template Template ...
Also jofi, thank you for putting my theory in the end of the iceberg 
is it just me or am I just down horrendous for the study girl
It’s been a long time since there’s an arg explanation video, there might be a wave of Russian coming in 
did no one try to pitch this to matpat or something
we'll see

There's a huge attention competition
Additionally the arg isn't done
But I don't think anyone tried in a while
Since Matpat is retiring, it's very unlikely.
i meant to the community ig
because it's not like GT is going defunct; there's the team working on new vids right?
dang
That's true.
I mean sure we can send it under ARG thread on the subreddit
but will it be able to beat the competition? unlikely
but can't hurt to try
I only discovered neuros existence this month so I probably will need a game-theory quality video to understand the arg
The problem is that for someone to make one of those, we first need a person that actually understands the arg
yhh true, tho it is possible for someone schizo to explain what we've solved so far and give an overview on the various leads
such a video might end up being like 30 mins tho 💀
I wonder if they will add this to Emoji? 
that one up there is 40+
so yes. it would
we could gather people together from here, make that video more of a group project

? my rss feed detected a change in study's video
(i accidentally swapped urls for numbers 2 and study, so it shows as numbers 2, but nevertheless, the hash is different)
lemme investigate
New Hash is for 360p24fps of Study. Old hash (794bcf6d557d881091189402b3997063d34747460617778c8e8cbd68c39739a1) actually no longer matches any quality I can get
but this is weird, youtube feed hash didn't change, so is this just youtube randomly deciding to re-encode stuff
(side note: i'm going to update the feed generator to swap back the order of study & numbers 2, my bad
)
import hashlib
from io import BytesIO
from pytube import YouTube
def shahash(stream: BytesIO) -> str:
sha = hashlib.sha256()
for chunk in iter(lambda: stream.read(4096), b''):
sha.update(chunk)
return sha.hexdigest()
if __name__ == '__main__':
yt = YouTube("https://www.youtube.com/watch?v=zMlH7RH6psw")
for s in yt.streams:
with BytesIO() as stream:
s.stream_to_buffer(stream)
stream.seek(0)
print(s, shahash(stream))
If you wanna generate the same hashes as I did
[Filtered] audio spectrogram
right channel
left channel
its like every note got split into parts and spread out on the left channel
also the "flares", whatever they are, are slightly different
new 360p has this metadata:
Metadata:
creation_time : 2024-01-21T20:10:36.000000Z
and both old 360p (found in the drive) and new 360p have different video stream hashes. they also have different video and audio bitrates. this is after enumerating through all possible 360ps i have access to
basically idk what's going on 
if i were to guess, maybe a popularity thing? after crossing certain number of views, it gets re-encoded or something. file size is smaller, data rate is smaller, bit rate is smaller, but somehow audio bit rate increases
i should just break up the file into individual frames and do a comparison like we did for filtered
In that case, why did only 360p video change? If I understand correctly that only it has changed. Should there be a counter that counts the number of views/resolution?
Otherwise, is it possible to change a video of a specific quality?
More than likely, YouTube just re-encoded the 360p version. They likely do that on occasion whenever new technology or something else like that comes out and need to test something. Though, I don't work for YouTube, so I don't know how those internal processes work, and can only speculate.
Can someone who speaks Russian(?) verify that this is a legit explanation? I mean, it almost has to be right? Or it'd be a very funny jokeTM
I dont speak russian but I watched whole thing on youtube translate and it appears accurate
yeah i think too.
Yeah, I'm watching it now as well. Even without subtitles it seems like it'd be too much effort to just be a nonsense joke. I'm putting it on the overview page
ok, 100% this is legit. And it seems to go in some fair details too. I just wish I could understand it..!
Yes, it is
heck ya!
i saw it getting recommended to me before but i never noticed its nearly a hour long 
that said, why do i get a russian video recommended anyway 🤔
I mean, i'm watching it right now.. Youtube might have detected that it has a bigger cross-appeal than normal
best part is when they do this and they completely break the videos 
The subtitles also are oddly accurate, of course not in 100% but about 85 at least
ye, it's pretty close. It could easily just be the autogeneration and autotranlsations that are off
checked all the decoded video stream hashes against the video in the docs and none of those match either
oh, weird
Of all the platforms it would make sense if youtube is the one where videos are dynamically re-encoded for some seemingly random reason or another
i only have enough information to conclude 360p video differs, because that's what the arg monitoring tool relies on. Not sure if the rest changed, but we should be able to tell from the video metadata header
Thank you
All of the streams, except pure audio streams, should hypothetically have this creation time
yeah I was trying to see if they maybe just remuxed the (legacy) progressive streams but doesn't seem to be the case
also compared the hashes of all frames: except for a handful in the beginning and end of the video and a few here and there in between the large majority of them seem to be different
(assuming it's even possible to change single frames within a youtube video)
this is to be expected under a different lossy encoding of the same original video, correct?
yes. assuming the whole video wasn't changed this probably means that it was reencoded with different encoder settings
also just checked the 1080p streams and they seem to have changed just a few hours ago (left the old video I saved, right the current one (ignore the change in container format, that's on me))
Thanks for putting the link to the vid in the overview megathread :3
You're welcome! (: It seems very well put together, even if I couldn't understand the finer details
i'm not done with the video yet but i second that
is that a freaking 20min deep dive down the deep end? good lord i don't have enough drinks for that
Acquire liquids!



