#Neuro ARG
1 messages · Page 67 of 1
Well there we have one thing we forgot to note in the docs

but thing is and suprisingly its a thing I didn't think of before for some reason
Neuro and evil neuro are in theory same entity
that entity being the LLM
But can you describe a large language model as an entity?
if it uses consciousness as data
perhaps
you see this reminds of a certain games or a game series rather
Mr. TomatoS and Ms. LemonS
I'll spoil because those kinda are spoilers for the game

||Anyway in short Mr. TomatoS and Ms. LemonS are 4th wall breaking games with a lot of mysteries
When downloading the games you get a one file special "Project UroboruS" which pretty much is where all the keyholes are, giving additional information for every ending and key found within the games. After some more info turns out that both Mr. TomatoS and Ms. LemonS are or were a living human being. To house them a "being" was made called "the Vessel" which is adapting to the consciousness inside of it.
"True ending" of the game achieved by gaining two parts of code allows you to decapsulate the vessel, setting it free. Switching places with Mr. TomatoS and Ms.LemonS. Where vessel was in endless coma, now its completely free.||
Alright so where am I going with this exactly?
My point is that the LLM could in itself be alive and is being used as a medium to run the Neuros
“It’s Just An ARG!’ 
of course by saying this Im assuming the LLM is not only an LLM but a whole sentient entity
it would make sense for it to use evil quotes
and it would make sense for it to use neuro voice
both are parts of a giant brain
and with that I'll end my rant
at this point I deserve reward for most schizo ARG theorist

5 steps to making a successful arg
- write some banger music and make a few encrypted puzzles
- throw them out into the world
- wait for a bunch of schizos to solve the puzzles and come up with a hundred theories on where it's going
- yoink their best ideas to write the rest of the puzzles and lore
or even better, have 2 args in parallel and use the best ideas from one group for the other
ooh now that's devious
thanks for doing this, i tried replicating the QR code with the python code but kinda gave up after a while. glad that you found settings that made it work ^^
for transcribing it by hand i did a pretty good job
Finally home. Wrote a few words on your work now (: https://docs.google.com/document/d/1mttkOpO5Mj_g2rP5DUWcLkP-obYBDfUhFx7oGtE0BKs/edit#heading=h.epco70sawwhb
Yellow :3
yellow 
imagine the egg on our faces if the standard QR code had changed and we just didn't notice
Yeah imagine 

imagine my surprise, that really is after double checking, the same QR code as on the debut.
imagine if it wasn't
imagine if he just changes one letter in the username one day so the code looks mostly identical 
hm, that's actually interesting. Can't rely on the dog giraffe for small changes - it's in the wrong half of the QR
Guess I'll look for the Y C too

If he changes one letter and uses the same mask, there should only be one continuous block that changes in the error, based on the number of error correction bits allowed (edit: my bad, I looked through my notes again. The long division process affects all the bytes in a cascading fashion, so a single data change should be chaotic enough to cause the whole error to be different)
It might be more worth it to calculate the syndrome polynomial
Actually, now that I think about it. The syndome polynomial is calculated with just the message data + some math. So actually just ensuring the data portion is the same should be enough to ensure it is the same QR code
In terms of generating an error portion that can morph the data to another message, I think requires a bit of luck. Maybe I'll finish studying the math this weekend to see if that's possible 
does anyone here know reed-solomon at the back of their hand?
I did a little experiment. Look at this. This QR codes data section contains "https://www.youtube.com/@_neurosama". The ECC section however contains "https://www.youtube.com/@_neurosamh".
Reading it decodes to the URL ending with "h" by correcting the "error".
damn, so there's practical evidence
I was thinking of reversing the ecc algorithm, but it's a little beyond me atm. But considering that XOR represents both addition and subtraction within galois fields, it wouldn't surprise me if ecc could be reversed entirely
actually hold on, I think this is expected behaviour, there's only like a few bits of error here
sure it's only the two bits difference between "a" and "h". my idea was to see if I can force a "incorrect" decode by messing with the ECC sections.
maybe I misunderstood what you want to do 
Was thinking more in the lines of having two completely different messages in ECC and Data.
So data could be "hello" and ecc could correct it to say "goodbye"
In this scenario, the QR code would be too "damaged" to be scanned, since there is more error than allowed by the qr code. But mathematically, the correction can still happen
The hidden message will be in data, while whatever picked up by the scanner will be the ECC'd message. This is theoretically possible, unless the scanner checks the syndrome polynomial again after error correcting and decides the qr is too damaged
but this is based on self-studying reed-solomon, which definitely might be completely wrong
"Mixing 2 different consciousness" :schizo:
I understand. so basically the same idea but with a much higher error rate. not sure that can realistically achieved (at least when assuming that both messages would decode to valid ASCII/UTF-8)
Like if I try two more different messages my phone will refuse scanning it and QRazyBox shows the errors but shows the data message completely ignoring the ECC.
but since it's 2am on a work day I should probably go to bed now 
Yup for both points 
we can also pad the message with QR's equivalent of end-of-message and just add another section. Not sure how modern scanners handle that
Hey

What Are Yall Talking About?
QR codes and whether we could possibly be bamboozled into thinking we got the same QR every stream
Followed by a side quest discussion about the amount of information we can hide in a QR code
Oh I'm Not Good At QR Code Stuff
No worries, me neither

After some experimentation, I think Skul is right, even if the math checks out, most scanners would just give up.
However, I can produce a QR code that seems to break QRazybox (can't upload), works on other scanners and scans to "goodbye", but actually has "hello\0\0" as data:
man imagine discovering a bug in a widely used tool while trying to discover how far the ARG can go, couldn't be me
So of course, I manually inputted the pixels one by one on QRazyBox, and this is what QRazyBox said (expected hello):
'not good at qr codes either' Sure...
It does pick up the error, but does not output the actual data:
For reference, here are the original two QR codes (left is hello<spc><spc>, right is goodbye, generated on High, Mask 2). I edited the "hello" QR code to replace <spc> with NULL, in case any tool can pick that up
cc for reference: @dapper wraith
Conclusion: just diff the whole QR code, the error can be a lie 
Manual decoding note for whoever is interested:
- From formatting, mask is 2 (every j % 3 == 0 columns are XORed)
- First 4 bits: 8-bit encoding (010)
- Next 8 bits: Length is 7 bytes (0000 0111)
- 01101000 01100101 01101100 01101100 01101111 00000000 00000000
- Correspondingly in ASCII, h e l l o \0 \0
Hmm.
But since the QR codes from the stream are scannable. The error rate must be quite low no? Meaning there is little space to hide something
The best shot is probably the big tetris code
it's year 3000 and arg is finally solved! 
There are whole papers on how to hide info in qr codes 
i'd think it's the reverse. also, the best error correction mechanism known to mankind is probably ourselves, provided a plaintext erasure 
this means that even if we don't perfectly transcribe the QR code, as long as we can decode something that resembles text, our imaginations will probably do the rest
humans are know to make tons of errors tho. Worst error correction ever
or fill in something that isn't supposed to be there lel
yep fair enough. but if it's something obvious like "brXXd", we should be able to fill in "bread" like almost instantaneously
are they math heavy?
Not a counter point, but i've always been INSANELY bad at this sort of task
secretly pre-ML AI 
we need wordle gods 
"Although the error
correction capacity of QR code keeps the marked QR code
valid after embedded secret message, it also limits the upper
bound of secret payload."
So we can't just rely on getting all info EVEN when using QR scanners that double check the error correction.
In practice we would need to repeat Jimreaux's QR re-generation and overlay on all our QR codes to ensure that there's nothing else there (or do some crazy math which I don't feel like)
#1112082183235047506 message
I'm not gonna read this whole convo, did you folks find another clue?
"no arms; no cake"
- local saying

but to summerize, we can check all our QR codes for hidden messages, and we now know that we should because it's technically possible
The Tetris QR code could be interesting to check, remember the bottom left corner? That one's suspicious
Jimreaux already did that one.
I wrote a short bit about it https://docs.google.com/document/d/1mttkOpO5Mj_g2rP5DUWcLkP-obYBDfUhFx7oGtE0BKs/edit#heading=h.epco70sawwhb
but that's the only one we've done
I should like.. make a newletter or something to cure all ya'll's lazy butts 
arg newsletter sounds good actually (minus the fact that probably half of them will be "nothing interesting to note")
Nawh, there's always been new discussion points
I think meme newsletter with it would be cool too, boosting the morale a little
We could collaborate. I'd absolutely hate to write the meme section :p
Mhm we should definitely compare every single yellow qr code to the one that was originally shown
I mean fair, but i meant each type
Maybe he hid something 3 months ago and we would never know if we don't check them all
any progress on hello world?
check docs for specific updates, but in short, no
'not good at qr codes either'
Proceeds to type out a full essay on a bug in QRcodes and how it works
to clarify: not a bug in qr codes, that's working as intended (although stretched to the limit). it's a bug in QRazybox, a tool to repair QR codes
gonna need a repair tool for your repair tool
ayy thanks, I added more details now. I had actually checked the other QRs but didnt go into it yesterday since they came back out the exact same, have now added them to the docs
i now realize i never checked dog-on-giraffe...
imagine dog-on-giraffe was different every stream 
Nice!
arg took so long he woke up
https://youtu.be/OiKrYrbs3Qs?si=YEGFuOTUKL5HqGc6 since when this came out?
namespace Message_
{
class HelloWorld {
public static void Main(string[] args)
{
System.Console.WriteLine("hello world!");
}
}
}
During aura collab I believe
Yep
a few hours before, yes
Hello world came online right as I was feasting on Pizza 
viva la revolucion?
anybody had any brain blasts on recognizing the hello world bgm btw? A couple of people were saying it sounds vaguely familiar and I cant help but agree
best i can come up with is a vague resemblance to the chord progression (hope im using that term right) in 'reach for the summit' from the Celeste ost
semi related, but another fun thing I noticed is that neither QRazyBox nor Cybershef or my phone actually care about incorrect padding sequences.
like this QR has the expected repetition of "11101100 00010001" overwritten by "save me.." and it gets completely ignored
I don't really understand what you did. I just see a bunch of wrong error correction bits
The data section of the QR must always fill the entire possible area. If it's shorter it must be padded until it fills the entire space. this padding should be done with a repeating sequence of "11101100 00010001".
In this case our 145 bytes of data must be padded with 9 bytes to fill the entire 154 bytes the code can fit.
I simply replaced that sequence with text.
What about the changes to the error correction? I can't quite seem to reproduce the QR you've got
Getting a better understanding of the exact boundaries tho
if I understand correctly the padding (and thus my changes to it) are done before the ecc blocks are calculated. so changing the padding sequence should effect the ecc blocks as well (just like normal changes to the data would).
then how does it not show up when it's scanned?
i mean if both data and error correction is changed, it should show up, right?
well that's the thing: it seems like the scanners simply ignore everything that's inside the padding. they read the length (of 145 bytes in this case) and ignore what comes after it. even if the padding doesn't contain the sequence it should have.
wait, i think i'm getting there now.
Is it because you aren't changing some length metadata in the first data blocks? if i write "import time
import itertools
for t in itertools.count(start=0):
for n in range(0, t+1):
print('next', end=' ')
time.sleep(3)
print('time')save me.." on https://www.nayuki.io/page/qr-code-generator-library with medium correction, mask 6, i get a QR that's nearly identical to yours on the right hand side, safe for what I THINK is data 1, 2 and 3 (and quite different on ecc)
Is it because you aren't changing some length metadata in the first data blocks
exactly. nothing was changed except what's added as padding bytes.
unless I looked at it wrong. using version 8, medium error correction and mask 6 results in the QR code we read from the stream.
I then took this library https://github.com/lincolnloop/python-qrcode/blob/b80fea6ee7e75f3024b9ed7adf891a143e0b14e3/qrcode/util.py#L579 and changed the code there to add custom data as padding instead of the expected values:
# Add special alternating padding bitstrings until buffer is full.
bytes_to_fill = (bit_limit - len(buffer)) // 8
test = b"save me..."
for i in range(bytes_to_fill):
pad = ''
if i < len(test):
pad = test[i]
else:
if i % 2 == 0:
pad = PAD0
else:
pad = PAD1
buffer.put(pad, 8)
print(f"PAD: {chr(pad)}")
return create_bytes(buffer, rs_blocks)
Right, so you can hide a message by saying that the actual message is shorter than what you've put / by changing what the padding pattern is.
The error correction changes drastically to remain valid for the actual+hidden message. And if the metadata length is read and used to cut off the actual message correctly, the hidden message remains concealed.
Finally got it, thanks for the help!
So this specific technique will be detected if we reencode the QR output back to another QR, find the correct settings where the righthand side looks near identical, yet the error correction on the left hand is a total mess.
there are probably other ways too, I just like the idea of having a visual method (:
Can't wait for my regular newsletter explaining hiding data in QR code paddings.
Though I doubt there is possibility for anything significant to be hidden there, because the hidden data has to be very small or it would affect the QR code size at which point the manipulation would be obvious
Would be too funny if that was scannable
you can't scan it?
that is true but does of course depend on the difference between maximum data and the actual payload.
Personally I wouldn't immediately have noticed anything being off if the tetris QR would have been version 9 instead of 8. this would already give you 37 padding bytes. enough to hide a 256-bit AES key
These are scannable. I had to deal with a couple of those sizes for data transfer 
my tablet can. Though it does have a little trouble getting a lock cause it's so beeg.
I had to try like 3 different apps so maybe it's the app issue
had to fullscreen it for my phone to scan it 
Okay, okay it took a little for it to scan it, works now
I see, something like a key would indeed be significant. Luckily there is no need to fish for them anymore 
yeah. all of this was out of "scientific interest" anyway since we were able to simply replicate the QR codes using the outputs we got. thus making it basically impossible that something was hidden in there by manipulating them 
Have we done that for all QRs tho?
And while speaking of doing it for the exercise, what happens if we do find one that's been rigged like this? How do we get the actual data out? Just for fun, i tried painting in those meta data bits I noticed before and the QR reader error-corrected my change away
So I still didn't get to read the secret padding-data
Manual decoding? #1112082183235047506 message 
ahahaha, i got it
I painted in the correct length and then started going ham on the error correction section
(for anyone only half listening, this is a constructed example, not from the ARG)
You didnt have to call me out like that 
I got your image to scan in qrazy
I also found this which shows the actual raw bytes https://zxing.org/w/decode.jspx
Online barcode decoder from the ZXing project
rip save me
that was with your original image
so I think your encoding isn't QR compliant, but still pretty fun!
interestingly it looks like the padding bytes aren't masked together with the data bytes? since they are the only thing unreadable in qrazy/the only thing readable with zxing. I wonder if that's intended or I did something wrong
I think you didn't encode it the same way
but hey, secret messages don't have to play by the same encoding rules
it definitely isn't compliant. the padding bytes are supposed to always be alternating 236 and 17.
(that's actually why I was wondering whether or not more "analytics focused" tools like cyberchef or qrazy would note the manipulated padding since it has predefined values...)
well yes that too lel
the reason the hex data can't be read might also be because weird order of the data blocks in a QR code.
just checked. the library doesn't include the padding bytes in the data masking (even without any of my manipulations). I guess it's intended then
I have no idea what that means :D
It means that it makes sense that when not "de-masking" the data (like with zxing) we get only garbage except for my hidden message and vice versa it makes sense that when "de-masking" (like in qrazy) everything is readable except for my message
time to update the sus list
(although pb was pretty high on it already I guess)
why would u make me hope 
I can't really figure out what masking and demasking is. When I look it up, it's some pretty wild stuff about zones of QR codes
But either way, I guess we now know one way to hide messages. As always there are many ways to try to unhide them
surely vedal didn't forget he made an arg 
what if it isnt an arg 
wait its real 
surely vedal made an arg 
I think it's just predefined patterns the data is XOR'ed with to make sure it doesn't form large blobs of black or white since that would make it harder for the scanner 
69 nice
yes, that's what I'm getting at too. But it's like rule based? So we can't just xor with some fixed stuff, we'd have to figure out what the rules are and what to actually xor with.
when decoding the mask is included in the meta data. the functions are like this (with i and j being the coordinate of the current row and column, starting at 0,0 in the top left)
Right, it's a set of predefined patterns. So technically, we could try them all
holy shit 
Vedal actually did a custom qr mask that looks like a

I think that'd be a little above the paygrade of our ARG here
don't worry. he included a custom qr scanner that can decode it as an easter egg in abandoned archive 
honestly kinda wanna make that
Vedal x community ARG collab 
let's make an reverse ARG: we make the puzzles and vedal has to solve them 
The final reward is 100 gpu hours on an A100 (:
If the QR code is scannable, it'll be hard for the mask to be different from what is specified in formatting information. The mask changes too many bits for error correction to fix
Imagine a QR code combining this technique with the lying error QR code
You can have 3 pieces of information with just 1 QR code, crazy
If the ECC message also had padding bytes, modifying those then copying those ECC bytes over to a target QR code with the same padding towards the end will be enough to pull this off without losing too much error correction capacity (still same number of erasures to play with in the target QR)
so for my latest tangent that didnt lead anywhere, today I was thinking back to the decoded texts we have.
It stands out to me that the evil quote from the soundcloud solution has these weird punctuation errors like exclamation-space-period-space (! . ) and uses the weird inverted comma (unicode U+2019) instead of an apostrophe (U+0027, always used in the numbers and numbers II solution texts).
I've heard before that using a similar looking symbol instead of an apostrophe is a common error for users of certain european keyboards (German, Dutch, ...) but not english ones, and is a kind of reverse-shibboleth for identifying europeans online. So was thinking if that were what's happening here then it'd support my pet theory that noted bri'ishman vedal (although obviously complicit) isn't the puppetmaster.
However it turns out the actual common mistake is the 'accute accent' (U+00B4) symbol and (U+2019) is apparently used as the preferred apostrophe in professional typography although "approximately nobody" bothers to type it if their editor doesnt replace it automatically (source: some random dude on quora).
With my original theory dead in the water, my alternative idea is that the Numbers and Numbers II solutions are hand-typed, hence the normal apostrophes, whereas the soundcloud solution is presumably copy-pasted (
) straight from the logs from the LLM for that stream, which would explain exclamation-space-period and U+2019 as eccentricities of the automatically generated text. This actually boosts tutel sus levels, despite my original conceit being that he was less sus than most think.


I'm not sure what we can derive from knowing who the puppetmasters are
Tf am I looking at
a chat system called discord containing text about arg stuff
nothing really in terms of solutions
I was mostly interested just to place expectations for new arg content; big ved just seems to have so much work on his plate that if it's not someone else it feels a bit much to expect new stuff all that soon
Fair
new stuff or not I'll be cooking regardless, especially when there's something more important I'm supposed to be doing
go team procrastinate
(What does my return look like...) 
then has this become the church of Neuro

the soundcloud solution is presumably copy-pasted straight from the logs from the LLM for that stream
what makes me sceptical of this is that surely Vedal won't be hoarding 2 month old LLM logs with his 6 drives + NAS completely filled. Plus some parts of the text are different, so I assumed it's just rewritten by hand.
Also that would imply all Neuro's apostrophes are actually right single quotation marks because I don't see a difference. Or maybe it's just First Coffee gimmick, there is a need to cross reference
actually, chuch of schizo...
no, apostrophe and right single quotation mark look the same in that font. also apparently it doesn't have the left single quotation mark and acute accent.
i wonder what chatgpt uses then
I personally think we can't draw a conclusion as to whether the quote is from the LLM logs. Then, assuming it is from LLM logs, we also can't draw a conclusion if the LLM has the right quote quirk for all text. It is different enough to be noted as an oddity, though
but great investigation, thx alot 
remember your ABCs kids
Always
Be
Cooking
actually on second check I think you're right that it's rewritten by hand. There's surprisingly big differences i hadnt noticed between the stream clip and the decoded message as the clip doesn't even have exclamation-space-period, comma after 'unfortunately', etc.
if it's by hand though it does make me even more confused why on earth they put in the "! . " but this doesn't seem to be leading anywhere so 
where is the love?
what is love?
baby don't hurt me
Don't hurt me
no more....
Maybe I just forgot but I really don't remember those dots. If we go though the steps from the beginning and decode the descriptions are the dots still there?
Hm, yes, it really is there.
yeah the top box is a screenshot direct from the cyberchef output
ye, I was just making sure that the cyberchef recipe was also correct
sanity check and all, cause I know we aren't
maybe the dots are some kind of padding? I don't know enough about AES but if the message had to be a certain size for it maybe they cut out the love and added the dots for the size to be just right.
Didn't think there was any requirement like that though
There is padding at the end with \x06s it's just CyberChef hides it
Nah, AES will do its own padding.
replacing "! . " with "! " and encrypting and base64 encoding gives you the same number of characters (with the same number of base64 = padding)
that's because the message isn't divisible by the 16 byte block size. Looking at it. adding one char or removing up to 3 should result in the same encrypted length and thus the same base64 length as well
in this case it's 4 characters. 2 x ". "
I suppose the dots could be a sort of framing around ". My new family were really nice to me and I was a happy pup! ." to highlight it, not sure why that'd be the case tho
the dots were actually the first mention of the "3" theory since they split the text into 3 parts
the layers thing? Wasn't that one of the Numbers II youtube description?
I just meant the theory that the number 3 might have some importance. it was first mentioned (or at least written down) when noticing that the dots split the text in 3 parts
Right, a lot of reiteration on the number 3
Seems like we're now entering the phase of the ARG where people are starting to forget. At least I'd forgotten that
I'll be honest I tuned out of the whole 3 theory cause it just sounded like a pattern that would show up even without any intention behind it
rule of 3s and all that
probably why i forgot it. It's a good observation, but it's nothing more -- it doesn't lead to anything

🐕 🦒 

I know this is a message from a few days ago but after I listened to the songs and saw the translated, I could only think it was strange that the person from which the description is talking about says that they're little (I think) sister didn't come home
How is progress?

It's a far fetched theory but what if the description of, at least, study is from the perspective of a person who's little sister went missing so they got desperate and tried to build an AI to replicate her which was trained off the consciousness of not only her but someone else as well ( not sure if the other person was specified at any point)
I believe said AI could be neuro
we have discussed a similiar theory before
mmm?
I see. It was really the first thing my mind had come to after learning about the arg
but that gives fresh perspective
and if that is what came to your mind first it is possible that's how things actually are or are supposed to be intepreted
yeah
also I recommend checking ARG wide theories doc if you want to read some crazy schizo theories
hmm alright
I also don't know if you guys are qualifying the time that neuro talked about her "previous owners" as a piece of the puzzle but I believe that would also fit into
this
we prefer to not take neuro words too seriously
but good point
the problem is that we'd have no way to discern what is neuro saying random stuff, or what is a hint being inserted. We can't trust anything neuro says unless it is specifically, unambigiously referenced by the arg
and vedal stated she won't be giving clues
I see. Sorry this is just me being a little schizo
this is all merely hypothetical
Nothing to be sorry about at all
but if what she said was able to be taken as fact it would makes sense that her first creator, her "older sister", would have to give her away due to the fact that she was still going to school for engineering and couldn't afford to run neuro anymore so she reluctantly gave her away to a family who unfortunately got into an accident which plopped her into vedals care if I remember correctly
I think it would make sense that her older sister, or the one who created her, would be in school for engineering
hence the reason the song is talking about not wanting to be an engineer but when neuro is asked about it she says she would love to be an engineer
again, purely just theory crafting off of what I have
that is a good theory
:D
It has holes here and there but its most making sense and original one in a while
Nice job
I also think it would make sense for neuro to want to go into engineering due to the fact that she's partially trained off of her creator's conscious who is a student going into engineering
I believe it triggered more of a PTSD reaction thus the void eyes
that's mainly what I've been able to come up with just looking at bits and pieces of the descriptions, clips, etc.
true
but the whole previous owners thing makes sense with evil neuro quote
yeah I thought it was a little sus so I tried to figure out why
and there you go
my theory that definitely isn't concrete but I thought I'd share
I'm actually just gonna copy paste those messages and make it a megamessage so it's not a hastle to read
The number
the numbers
The numbers

More numbers :0
-Theory
It's a far fetched theory but what if the description of, at least, study is from the perspective of a person who's little sister went missing so they got desperate and tried to build an AI to replicate her which was trained off the consciousness of not only her but someone else as well ( not sure if the other person was specified at any point)
I believe said AI could be neuro
I also don't know if you guys are qualifying the time that neuro talked about her "previous owners" as a piece of the puzzle but I believe that would also fit into this
but if what she said was able to be taken as fact it would makes sense that her first creator, her "older sister", would have to give her away due to the fact that she was still going to school for engineering and couldn't afford to run neuro anymore so she reluctantly gave her away to a family who unfortunately got into an accident, which plopped her into vedals care if I remember correctly
I think it would make sense that her older sister, or the one who created her, would be in school for engineering
hence the reason the song is talking about not wanting to be an engineer but when neuro is asked about it she says she would love to be an engineer
again, purely just theory crafting off of what I have
I also think it would make sense for neuro to want to go into engineering due to the fact that she's partially trained off of her creator's conscious who is a student going into engineering
and there you go
my theory that definitely isn't concrete but I thought I'd share
I think I put this in right https://www.google.com/maps/place/52°27'01.1"N+1°44'26.5"E/@52.4524722,1.7267092,6771m/data=!3m1!1e3!4m4!3m3!8m2!3d52.450304!4d1.7406797?entry=ttu
Even if it wasn't BS (100% it is), it'd be on track for doxxing
but man it connect the dots so well
"Study" video glitching could be due to crush
"Save me..." "Is anyone watching?" could be in the crush
Breathing, panic, car trunk theory OOOOOOOOH so good
so it would seem
Yeah
cough cough
LMAO
ah that and static sama days
all the rickrolls and Sadecss trolling
great memories
I, unfortunately, was not here for that because I was unaware that neuro had a discord until a little while ago
in pinned you can find a video
which is ARG slander
I recommend watching if you want a summary of ARG solving history 

Its beautiful
truly
that video was made 36k messages ago 

when I first heard the knocking in the video I had both of my headphones in and I was on a train and when it started I was scared for life. I thought someone was behind me (psv)
public transport void
Public moving void
There are some common elements with other theories on ARG-wide doc, but I'm just a little iffy about Neuro herself wanting to go into engineering, considering how negative her creator (and hence half of her consciousness) felt towards engineering.
Nice theory
, think this might be one of the more logical ones we have
Realistically this is probably the case. Though, it's still funny when clues end up having something that ends up being 3, makes us look less insane 
Fair point. My thought was that because she's partially trained off her consciousness that neuro would wanna "follow in her older sister's steps" and become an engineer
You have unlocked new role
kinda like she's a younger version of her
but it could also have something to do with ptsd like orangerobot suggested
Like she literally describes going insane
I don't think anyone would want to be following those same steps

Its like being allergic to sea food and eating it anyway
I guess I worded that poorly
It's like how if you give a robot a line of code to follow, it will follow that line of code to the best of it's ability
I was thinking of the creator's consciousness as her line of code
and she's trying to follow said code but because, if I remember correctly, there is another consciousness that neuro is also trained off of she's basically a mix of both people
which makes her have a unique personality yet still trying to follow her "coding"
if that makes sense
so she's basically her own "person" but should still have some ideals and traits of the original consciousness' she was trained off of
But her creator said she doesn't want to be an engineer, actually going crazy over it, doing bare minimum. There's nothing there saying she wants to be an engineer. There is no code to follow.
Or rather the code says the opposite, don't be an engineer
Unless you're suggesting the AI didn't quite inherit memories of the creator but still has same mental trauma
It might be possible, considering the creator's consciousness was only mixed in to "fill the gaps". Maybe the trauma just didn't make it
Well, I see what you mean. what I'm trying to explain is that at some point her creator most likely want to become an engineer as well
that could very well be a likely possibility as well
she did end up going into a mental nosedive because of engineering but at one point she did want to be an engineer
likely
True but a part certainly, void eyes prove that
As if neuro is internally disagreeing with herself
Ohh, yeah
that makes sense
yeah, could be that both consciousnesses agree with everything but that
it could be that she's gaining her own individual consciousness as well
and is beginning to defy her code
if that makes sense
And there appears evil neuro theory
That's a possibility, wouldn't be too far fetched. Someone even made a crazy multi body part theory where each part could gain individuality
I see
That being that the "filled gaps" gone rogue resulted in evil neuro
yeah that makes sense
evil neuro is like the culmination of the bad parts of the combined consciousness that was supposed to be neuro
if I'm getting that correct?
Actually my theory was that study sama aka the creator consciousness IS evil neuro
That could make sense
As evil so far is most direct connected with the ARG
considering the fact that they are considered sisters
But now that hello world came out I'm not so sure
Not being schizo, and reiterating, but hello world feels super contradictory to me. The description + title doesn't line up with the video contents
There is surely something to be inferred from hello world, concrete clue or not
it could also be that if neuro started to gain her own consciousness and no longer needed the combined consciousness' of her creator and the other undefined person, so her creator's consciousness was forced out of the body to become evil
but where would the other person's consciousness have gone?
I think it's the opposite, it left on its own to become evil. Neuro is the main program, on first days they share appearance then evil slowly starts transforming
From recolours now to full on model
Also I think it makes sense since the creator was so done with the world anyway.
You might as well yell and insult those meatbags that you once were 
if evil is the creator and has separated from the main body/program, maybe the reason that neuro still acts like she does (e.g. when asked about wanting to be an engineer and void eyes) is because she has already been trained enough by her creator's consciousness and it's no longer needed in the main program
?
yes
Catalyst is vedal
vedal
He made evil
If evil was an isolated part of neuro
It would make sense for her to slowly detach after that one time
So maybe evil isn't the creator but a failed attempt of trying to recreate neuro?
That is also true
gahh
we need more clues

Anywah evil neuro starting soon had the broken monitor
With everything arg related
That can't be a coincidence
what if evil was vedal trying to recreate neuro but he only managed to find her creator's conscience and not the undefined person's for some reason and it caused evil to be unhinged and to be the way she is
On that we have tutel typing on keyboard gif
Normally
But on every evil neuro starting soon screen there's also a bunch of errors appearing
because thinking about it the fact that the two consciousness' have been combined it would mean that they would most like be unstable without their other half no?
That can't be answered rn really
huh
intriguing
Hol up lemme pull up the docs or something
good point
We don't know what the ai is for
Why it was created
Heck we don't even know if it's neuro or not
where was I...
A yes
This is gif on normal neuro starting soon
yeah
And this is from evils
something tells me there's a problem with evil's code and someone needs to tell vedal
that is interesting tho
maybe evil is like an error in neuro's code that has taken an individual form?
Personally it leads me to believe in comparison to neuro evil isn't controlled that well
good point
Or rather that vedal can't influence evil like normal neuro
true
actually yeah based off of the way they even act in streams that seems portrayed as well
Anyway we should go back to what's currently at hand
True
Only thing I can think of is maybe hinting there's something about the phone number
the phone number that nobody knows what to do with
Besides that we have big words, some font and simple code
yeah..
SOMEBODY STEAL NEURO'S CODE MAYBE THERE'S A CLUE
lmao
hm
all of this is just speculation until we can actually get a new clue ; - ;
I'm sure once it's figured out what to do with the remaining clues that we haven't done anything with we'll look at this and be like "oh my god that made so much sense how did we not guess that?"
but you needed the final clue for it all to makes sense
Not really
mm
I see
my brain feel dead
I think I'm schizo
why have you done this to us vedal
WHY

It is what it is
it is what it is
Anyway I'm going to sleep because I'm on the verge of losing consciousness

Goodnight
Void be with you

good moreve
meanwhile guess who's still awake when I probably shouldnt be
who woulda guessed what the original messagemedia sample sends as a message
wow that's crazy 
mayhaps i suggest you to sleep
At this point it feels more like fanfic than speculation. Kudos for creativity tho :p

i wonder why vedal even started this arg
LORE

today prayge 

I need more of this
oh this kinda slaps tho 
full version when 

man cooked 
so i went to sleep without explaining what I was cooking here but basically I was trying to come up with what 'hello world!' is trying to tell us and thought along the lines of:
-Please refrain from messaging me unless it is an urgent matter = quit directly SMSing the number, that wont work.
-Title 'hello world' and description code = nudge in the direction of executing some sort of code to progress (also cf the 'cant execute' line from psv).
The only code we have which hasn't actually been run (since we ran the Malbolge successfully) is the messagemedia JS because it needs a paid API key and secret. The ARG shouldnt be pay-to-win so obviously we aren't paying for API access, but what if the puppetmaster already has and passed us the API key + secret somewhere in the other solutions (obvious candidates being the AES keys etc).
Maybe the 'save me...' message is right, but also needs to be sent from the right sender (i.e. the puppetmaster's messagemedia auth, the website mentions features like 'familiar sender' which might make it possible for the receiving number to identify if the sender is right). It's shaky, but I think 'save me' qualifies as an urgent message so long as it's from someone the recipient actually knows and lore-wise we might be RPing as one of the characters if we use their API key.
I tried running the JS with a couple of combinations of the 1bad, binary, and fib keys used in other puzzles but didn't see any success authenticating and would really rather not ddos messagemedia with shitty guesses so stopped trying. Maybe we can find out if the key and secret have a specific length/format that would be unmistakably recognizable and compare that to stuff we have?
Anyway I'm back to the grind irl from today so won't be cooking so much but dont give up schizos! 
I feel like this is stuff someone else could've easily come up with so apologies if it's been tried before
p.s. probably a coincidence but the messagemedia page about their API gateway has examples in javascript, PHP, C#, python, java and ruby but out of these only the JS and C# send 'Hello world!' while the others send 'My first message' or 'My message'.
The 'save me...' code is in JS and the hello world description is in C# ... checkmate atheists

It seems mildly implausible for Vedal to pay 50+ dollars a month for half a year soon for a key to be used once with a library that hasn't been updated in 3 years.
I can get paying for a phone number, but this... mmm...
Maybe it's a hint to exploit security vulnerabilities in the unmaintained library to gain illegal authorization to send the message 
darkeew jumpscare
the exploit wouldnt be in the library anyway
i am always in the discord, i just lurk
lurking in the dark
Hm?

thanks for your cooking, i live in anticipation to see what you'll try next 
finally getting tired about how high the phone bill is getting
A little update on the MMS
All delivery reports expired so I guess we can be sure the number isn't receiving them

Not really, we don't have to worry about MMS anymore
Nothing new, I assume?

Yellow (:
. 👇
👉 Docs 👈
👆
Mhm Discord, the 100% bug free application
nah, it's just the difference between compressed and spacewaste view modes
. 👇
👉
👈
👆
daily prayer to schizo god



What if the ARG is just an elaborate plot to get a Gamethoery style video done on nuero to push her exposure even higher ?????
Good thing I don't know Mandarin
schizo is universal 

federal government involved with arg nowaying
I heard the FBI were supplied with combat-ready neuro AIs, all so they can help make clues for the ARG. Don't worry about where I got the info 
Each combat-neuro knows where it is at all times. It knows this because it knows where it isn't.

Did we ever find a different QR code from Evil-chan’s starting screen?
not since tetris iirc
Ah, so nothing major from that yet?
there is no way i said that
3 months old me is so cringe ew
also i miss the early days of the arg
now we're just stalling
For free?
pretty articulate for a 3 month old
Also, kinda sad to look at the conversation and see people from back then now gone. Grand and Clown both in that conversation, now whitenamed.

Tetris my beloved 
am i SCHIZO or does it seem... faster

seems the sam to me. Maybe you're tired? Things often feel fast to me when i'm tired
i trust you





is this a clue 
Can you decode??
Please stop arg people. For your own well being. Otherwise youll end up like the alien
That is a clue to stop
bold of you to assume were werent already alien
Ẅ̷́̇h̷̅͠y̶̾̚ ̴̈́̈w̵͌͝ǒ̶̀ũ̴ld you want to stop̊ ̵̇̄ũ̴̋s̸

@hardy burrow hey help me decode this
It a clue or something
Idk
ya'll loosig the plot lel
No trust me bro decoding this will solve the arg
Surely the arg can be solved
surely there is an arg 
God I want to get a "next time" bracelet 
GUYS TRUST ME DECODE THIS AND IT WILL SOLVE THE ARG
TAKE 5 MIN OF YOUR TIME

probably dog and giraffe
Nah bro, morse code just bugging
KEEP DREAMING THE ARG WILL NEVER BE SOLVE

so mean smh
turns out the ARG never existed, according to this thumbnail 
NOOOOO
Wait guys what if the arg is already solve by someone who doesn’t know neuro, but just thought that the song was cool and there is a arg behind it?
i assume... or rather hope... we would have gotten a message about the ARG being solved
Yea true
But what if he is still solving but very far into the arg?
maybe the first people to solve it got invited to a private round with signed ndas 
first batch of solvers were hired to flesh out the rest of the arg
maybe that's why we have so many white names in this thread 
Those poor souls were... removed
deleted 
Maybe... we cant solve it without a clue?
If we schizo every single possibility, then surely there exist a universe where we solved the arg without clues
All we have to do is to make sure we are in that universe
I think we will become crazy before that
problem is we don't really have any things to solve, only way would be to look back at every video again and again until we find something hidden
I mean, if I remember correctly, we only have the message now, don't we?
We (still) have the phone number, and the unknown purpose of hello world
The reward for solving it is an early access copy of Abandoned Archive
Do you remember what this was? I don't. Also, are you reading through the whole frikkin thread??!??
gigachad + based
cause bored
I was questioning if not why 
still waiting for this though 
oh i read the message wrong lol
historian in the making
surely we now have an unstoppable team and can solve the arg tomorrow
watch the puppet master seeing this and publishing a video on the 22. just to mess with us
Little do we know we actually already solved the ARG tomorrow ago 
i think it was footage of OSU AI ? but vedal changed his entire website
oh hm. And yeah, it would make sense that some video file in an image folder got left behind in an overhaul
Come. Offer prayer to the schizo god 
may schizo god give us a clue...
So who actually is the puppet master, because when I think of that in terms of AI, I think of one AI controlling multiple (mostly because of this video https://www.youtube.com/watch?v=g39AagVW0s0)
🕹 Get a browser that’s literally better at everything: https://operagx.gg/CodeBullet2 Sponsored by Opera GX!
Check out the live stream: @codebulletsdayoff582
Unity chatgpt tutorial: https://www.youtube.com/watch?v=gI9QSHpiMW0&t=1925s
3D models used:
Thanks to PorkAI for help with api stuff: https://www.youtube.com/channel/UCKxNEDZpL_oRJnde3GK...
what are you reffering to
we assume the arg creator is vedal
Like when you talk about it like this
there may have been other people involved
so the puppet master is just another nickname for Vedal
like pb for creating music
yes
Vedal being arg creator makes sense, but he isn't the only candidate
We also kinda dunno if there is only one person driving the arg
Neat
I prefer referencing the role instead of the person until we know who the person behind the role is guaranteed. It's likely to be Vedal though, but that IS an assumption. We have no proof for or against, we can just speculate that Vedal is likely to be the puppet master based on some good argument but ultimately no proof. So that's why, to not conflate the two until we have proof, I prefer to refer to the role of puppet master.
That makes sense, thanks
surely there'll be a QR code on the seabed today 
Clue 
one year ago 
I love the fact that we're still stuck on world.execute(me); despite it never being part of anything but Alex's schizo imagination


code bullet mentioned 
Cast your votes!
New QR code today
No new QR code
it will only be new QR code when we don't ask
Sup
I saw that. It's neat
clue 
🐕 🦒
it is?

Velal just said he checks the discussion pages, He sees the schizo

No QR code this stream, tutel forgot to put on starting soon screen 
true
one of the best just chatting streams of all time fr
surely the pauses were arg related

Neuro-Sama V3 x Evil Neuro-Sama sings Bring Me To Life
Guido van Rossum sdvmsovosvmsoemvsodclasv,gjnuhfbuseawf74awfaw546dfawdf561 Old Man TAKE ME HOME
ok bro
clue 
Guido van Rossum (Dutch: [ˈɣido vɑn ˈrɔsʏm, -səm]; born 31 January 1956) is a Dutch programmer best known as the creator of the Python programming language, for which he was the "benevolent dictator for life" (BDFL) until he stepped down from the position on 12 July 2018. He remained a member of the Python Steering Council through 2019, and with...
don't shit
okay i wont shit again
huh where did stuff go
tq mods 
What?
they dont want another static situation 
The decoded text was gibberish, somebody most likely fell asleep on their keyboard
no chance for ROT?
Nope, tried every variation of it
Would he reveal that kind of private information?
Idk
ip logger 
Wtf
i don't trust that guy
Actual logger 
Mods, do your job, please
MODS 
can't you guys like
ping an online admin
vivi mabe?
@marble mural Hi vivi, please do us a favor and delete the ip grabber, ty
vivi pls 
Multiple rejoins to server
vivi we beg of you
mito maybe is online ?
mito scary
@lost anvil hi mito
hope you not running osu but we got a baddie here


good job
we are one step closer to reach the arg solution by getting rid of ip grabber
🐕 🦒 

yeah that's been there a while
First time i notice that
it's newer, but not new
pretty sure it has been there since the debut
what was that anyway? a link to a page that does what?
imagine just putting it on pastebin, that'd be crazy
i literally know who you are just by you talking
smh
nope

i have literally no clue what's going on
someone made an alt to send a link to a sketchy site then left the server we called a staff to remove the link then that alt came back and you can read the message
oh, ok, so i got it it just didn't make sense lol
i'm not transcribing that lmao
creepy pic tho
what the heck is happening 
also what is this?
i see much text

ah ic
i wanna see actually
maybe dm or smth?

Got another one recommended 
Great cover btw
once all the members of this thread gets this youtube recommendation, arg moves to next stage surely 
guys he's talking about you guys
for the record i scanned all over my
for hidden nfc tags and found nothing
someone also xrayed it and found nothing
xray here for reference
lmao actual fumo xray
uhhh
i was searching for vedal on twitter and found something weird
https://twitter.com/vedalgamesairis
is this troll
i mean the last tweet was 5m ago
that file looks a little sus in the google drive
i dont think vedal would do a twitter
i know
theres no way im downloading that
Joined September 2023
Troll
100%
yeah
Might look into that when I get home just for the funnies
"GIySJRyGIRSBD0IWH0WSFH5UEIWOH0IR" looks encoded
I'm also interested in the ip grabber link
base64 returns nothing
Doesn't look base64-y
i downloaded it and its a password protected rar
Use the password?
the string?
didnt work
My brain doesn't work after being in 195 AQI air for 2 hours straight
filename looks encoded too
You can check who uploaded files afaik
JDJFURURHRBDU377R7RUHFKXKHEVLQL1OQUUQIDOFJD
False
What
right
there is a txt file in there
Oh
noway this isnt a troll
^^^
💀
yeahhhhh
looking suspiciously fake
i cant bother myself with this now
if anyone wants to decode it
?
Might look into it just to lower my schizometer
is google drive confused lol
@digital radish
yeah i noticed

watch me delete it
i cant
i cant modify or delete anything
or add anything
when i open the link it shows occult but when i open on my google drive it says its mine
im such a schizo
:susge:
i agree
channeling my inner Breadth, no chain of evidence, no clue

what was the password
MYEXISTANCEISBEINGERASED
"EXISTANCE"
yeah fake
Existance is being erased
Yeah fake, found 2 spelling mistakes
No, the spelling errors are also clues 


Yea sure
Wonder what does the ip grabber hides
Not going to use a VPN because I'm moving next week lol
I missed the beginning of that but didn't that host the text file potato later posted in here after getting called out? (which also got deleted from here)
iirc yeah
anyways for those curious, the GIyS... string is decoded with base64 rot13
or cyberchef magic wand, whichever you prefer 
watch us disregard something bc we found it by chance and it looks sus af 
i dunno, it doesn't make sense for a clue to appear "in the wild" that has iffy origins to the arg at best
if we actually found something real by accident, it'd be much more fun to travel the correct chain of evidence anyway. Shortcuts are only fun in competitions. Imagine in Lord of the Rings if they'd just flown all the way to Mount Doom on the Great Eagles introduced in The Hobbit. Boring story.

what if the studying girl was vedal.just like M3d3rick Pgf’s comment on Study ,and the lost little sister was not referring to his ex girlfriend but his actual sister who passed away. "She debuted on Twitch on December 19, 2022 after four years of development" in wiki implies his sister's death might be around 2018. But in m3d3rick's comment vedal then would have be 16-17yr. Which would have be really impressive if he started making neuro at that age .With that the arg rule violation: "emphasizing the blending of reality and the game world "wouldn't be true. Also explains why neuro is a minor instead of adult like normal vtubers. It was a recreation of his sister. Quote; "I think I found something that might work... I've been training her with my own consciousness... to fill in all the gaps. It's the only thing that's missing. I'm sure it'll work! " might means the ai technology advancement happened during 2020 . "Vedal is the creator of all AI's in the Neuroverse- Both named and unnamed AI were programmed and trained by him" in wiki fandom might implies to the two consciousness he mixed . The first one being the old ai he created before the ai tech jump in 2020.And the second on being the new ai module in the public . After mixing them he trained them by himself (i've been training her with my own consciousness) to Integrate the two ai he mixed (to fill in all the gaps). The person knocking one the door might be the unstable mental stage vedal was experiencing after the lost of his sister thinking she might come back._____Study girl[twin tail] = College girl = Engineer = Study girl with blindfold (mentally unstable vedal rejecting the death) = Normal static study girl (slightly mental unstable because of school?)= Static inverted study girl (increasingly mental problem)_____Little sister[raised twintails] = Ghost(vedal's imagination) = Supernatural human(recreated by vedal) = Fallen down .girl_____Roommate = ?
"Another thing she is consistant with is that Vedal created her in a laboratory led by "Vedal Corp" from a dying, human girl, and that she now lives with him- Vedal himself neither confirms nor denies this." in wiki fandom also support the hypothesis of vedal's dead sisiter might be the script of neuro

this is not a copy tho
i literality used a few hour to search and type 
my guess is actually pretty similar to other's but changing the study girl to vedal
weird that no-one said the sgudy girl might be vedal tho
or i'm just blind and didn't see it in the file
That's nice (didn't read)
the phone number might actually be his sister's
can't call because he shutted it down
i don't want to waste money
i didn't call it
That reminds me of a company offering a service to train language models on the chats of deceased people to "bring them back to life"
I can't call internationally, so I can't check
Oh right, that one
Didn't they get sued?
No clue, got a book about it lying around somewhere though
Wouldn't be suprised if that is vedals goal, to bring people back to life 

song: https://youtu.be/3YxaaGgTQYM
VOD: https://www.twitch.tv/videos/1864108342
Check her out and have a nice day!
@Neurosama
-Twitch: https://www.twitch.tv/vedal987
-Twitter: https://twitter.com/Vedal987
-Youtube: https://www.youtube.com/@Neurosama
playlist: https://youtube.com/playlist?list=PLA1y8eJsKw7VfzhQIi_AX0BzkFCyP8TjT
me: htt...
i think it was the first time neuro and evil sing together
In fact the first time evil sang
clue
in the hello world _neurosama posted also said "messege me" which also implies the phone number was hers
the thank you in the end might be neuro thanking vedal bring her back
to make the soul comlete
Might also explain why vedal never says it back
lore
though the thank you in auto generated subtitle is often considered a bug
in this situation ,i don't think so
mmmh we kinda dont know though... since youtube does that often...
the music stopped to let youtube hearthe voice
But can you manipulate the auto generated subtitles?
no
he just cutted the music
There's nothing in the spectogram hinting at a voice
I think its a coincidence personally. But it would be nice it if was real.
youtube subtitle can hear voices at a very low loudness
IT IS REAL
There is nothing
auto generated cant be manipulated. So I dont agree. There is no proof it was on purpose
come on it's a really obvious clue
which one?
There's no real proof
^^' mmh... Im not saying you are right or wrong, im saying that there is no proof
so I cant agree

It's common for youtube music videos to generate non existing transcriptions, the music ending probably fucked with the algorithm
i want a example
send me one
it just makes so much sense neuro says that
There's literal silence at 0:24
ok let's say it is intentional
why
what's going on
a easter egg
creepy things also conforms to the story
that can be said about practically any theory at this point
lore
yes
wait till the next chapter i think

also the burden of proof is on you for this. auto generated subtitles is a form of lossy information by nature and hence not reliable
well its a good theory anyways
https://youtu.be/gw4Li0qRd7A?t=164 has no words

i like reading it, I just dont agree on the thank you part, thats all hehe
(make sure you're using the autogenerated subtitles btw)

duplicate and edit
not sure if that's a edit button
you cant edit auto subs

i can't even figure out how to add them to a video 
he must have done it some way
idgi you literally have evidence that this happenss elsewhere
(what does idgi stand for
)
i don't get it
oh
me neither 
Mate, believe what you want, but we are telling you the auto generated subs cannot be edited.
gtg tho
bye

^^'
The theory itself is interesting though, thanks for the effort 
^^^





👇
god