Return to Unfiction unforum
 a.r.g.b.b 
FAQ FAQ   Search Search 
 
Welcome!
New users, PLEASE read these forum guidelines. New posters, SEARCH before posting and read these rules before posting your killer new campaign. New players may also wish to peruse the ARG Player Tutorial.

All users must abide by the Terms of Service.
Website Restoration Project
This archiving project is a collaboration between Unfiction and Sean Stacey (SpaceBass), Brian Enigma (BrianEnigma), and Laura E. Hall (lehall) with
the Center for Immersive Arts.
Announcements
This is a static snapshot of the
Unfiction forums, as of
July 23, 2017.
This site is intended as an archive to chronicle the history of Alternate Reality Games.
  

The time now is Thu Nov 14, 2024 2:03 pm
All times are UTC - 4 (DST in action)		 View posts in this forum since last visit
View unanswered posts in this forum
Calendar
 Forum index » Archive » Archive: MetaCortechs » MetaCortechs: Puzzles
[SOLVED] Marcus's Metadex files (dimitri.zip) [Nov 6]
View previous topicView next topic
Page 7 of 10 [149 Posts]   Goto page: Previous 1, 2, 3, 4, 5, 6, 7, 8, 9, 10  Next
Author Message
Insomniac
Decorated


Joined: 14 Oct 2003
Posts: 234
Location: 01

 		I'd buy a t-shirt as I already have a metacortechs t-shirt I made before I even discovered this forum.

(I wore the shirt when I went to see Revolutions)


[EDIT]
Also check this! http://www.thinkgeek.com/tshirts/generic/5f84/
[/EDIT]
_________________
"I find it kind of funny, I find it kind of sad.
The dreams in which I'm dying are the best I've ever had."

PostPosted: Fri Nov 07, 2003 4:40 pm
Last edited by Insomniac on Sat Nov 08, 2003 9:33 pm; edited 1 time in total
 View user's profile Visit poster's website AIM Address
 Back to top 
bakntime
Unfettered

Joined: 31 Oct 2003
Posts: 462
Location: back in time

 		There hasn't been a whole heck of a lot of talk about pictures #1-#6...

Is anyone else as creeped out as I am by the pictures of Avery (assuming it really is James Avery)? They're weird! What's up with the guy behind him in photo #6? Is he taking a picture of Avery? He appears to be holding something bright in his hand(s)...

I don't know... I don't have much to speculate as to what exactly is happening in these pictures. They just seem pretty disturbing in a way, and I think there's a big part of the story going on here.

PostPosted: Fri Nov 07, 2003 11:19 pm
 View user's profile
 Back to top 
Bort
Kilroy

Joined: 07 Nov 2003
Posts: 2
Location: Newcastle UK

 		Re: [SPEC] Message stuff
[quote="xnbomb"]
chancesend wrote:
Is there a reliable pattern that can be found in that shift so the initials can be predicted with some certainty? If so, we may be able to determine the author of the message, and learn the identity of Avery's son.


My first post, so may be completely off the mark, but perhaps Jesse from the Emerson CD is Avery's son???

PostPosted: Sat Nov 08, 2003 12:19 am
 View user's profile MSN Messenger
 Back to top 
xnbomb
Unfettered


Joined: 13 Oct 2003
Posts: 660
Location: J302B S8JDC

 		Addendum to tapestry
First of all, chancesend and yeahyeah did a phenomenal job on this ... finding the way the message is hidden in this cryptic image is an impressive feat.

I was curious about some of the apparent slight inconsistencies in the substitution cipher, so I wrote some code to test out the solution in an automated fashion. This was easy because of the work already done to figure out how it works. I did it using the ArcView geographic information system. It's designed for spatial analysis, so most of the capabilities to recognize and compare shapes are already in there ... so it actually wasn't a lot of code.

The first thing you might be interested in knowing is that this puzzle seems to have been designed to be solved by hand and eye rather than by using computers to match symbols exactly. The reason this is the case is because there are slight inconsistencies in the exact set of pixels used to make up the 'same' symbol (even taking rotation into account). The pixels near the edges of symbols are often modified to produce the uniform patterned appearance of the tapestry. Thus, two symbols that are the 'same' to the eye are not identical from a pixel-based pattern recognition point of view. For example, compare the symbols for the question marks after KNOW and HEARD ... the 'bite' taken out of them to get them to make them match up with the adjacent symbols is not quite the same:





This is true for many of the symbols. If you take into account the small differences introduced this way, there are 94 unique symbols, whereas if you ignored those differences (as chancesend and yeahyeah rightly did to get a sensible looking plaintext), there are only 28 or so as identified (some of which are punctuation).

There are a few typos which appear to be in the encrypted text ... the plaintext interpretation seems strong enough (repetitions of same letters elsewhere in a fashion that makes sense) that it is unlikely that they've been mis-substituted (YIUR, LNNG, THUY). Here's how the text reads without correction (in the original 18x18 grid):

Code:
I_KNOW_THAT_MY_DAD
_TRUSTED_YOU_AND_I
_COULD_USE_YIUR_HE
LP._I_DON'T_KNOW_W
HERE_ELSE_TO_TURN.
_IT_HAS_BEEN_TOO_L
NNG*_I_SHOULD_HAVE
_HEARD_BY_NOW._NON
E_OF_OUR_PLANS_HAV
E_WORKED_OUT._I_KN
OW_THAT_THUY_ARE_S
TILL_OUT_THERE!_BU
T_WHERE?_DO_YOU_KN
OW?_HAVE_YOU_HEARD
?_CAN_YOU_HELP?_TH
IS_IS_ALL_MY_FAULT
._I_WISH_I_UNDERST
OOD_WHAT_I_DID._**


I've used * for characters whose identity is ambiguous. The * at the end of LNNG* could be a colon. The double ** at the end of the message is interesting. As yeahyeah noted, those symbols are unlike any others used elsewhere, so they are either filler, or potentially are letters not used elsewhere. It's worth noting that the letters that do not appear anywhere in the plaintext are J, Q, X, and Z. Therefore, it is possible that those two last characters could be any of those letters, and if they are indeed initials it could be a hint as to the name of the author of the message.

[EDIT: No error in HELP ... my mistake ... found by redoing the analysis by computer Very Happy ]

(EDIT: Moved image to attachment)
qmarks.gif
 Description   
 Filesize   4.55KB
 Viewed   859 Time(s)

qmarks.gif
_________________
My location is a little tricky, but sooner or later, you'll get the knack.

{J302B S8JDC, 8996N M8L4W, 92D40 Q1JX5, 4PPRN R2B97, 8DC7C NZJNV, 8CH7V Q891H}

PostPosted: Sat Nov 08, 2003 1:49 pm
Last edited by xnbomb on Wed Nov 26, 2003 7:59 pm; edited 3 times in total
 View user's profile
 Back to top 
BrianEnigma
Entrenched


Joined: 05 Oct 2003
Posts: 1199
Location: Pacific Northwest

 		Re: Addendum to tapestry
xnbomb wrote:
The first thing you might be interested in knowing is that this puzzle seems to have been designed to be solved by hand and eye rather than by using computers to match symbols exactly.


It can still be solved computationally. If the image parsing is done in the computer, you can do a "fuzzy" match as the symbols are being lexed. When scanning tiles, you compare a tile to all previous ones using slightly fuzzy heuristics. If you get a (to pick a random threshold) 90% or higher pixel match, you can assume they are the same.

Alternately, you can hand-translate the symbols into numbers or ASCII characters. At that point you just bypassed the lexing and can simply do automated work against the Caesar cypher directly. (I have code up on SourceForge that was written several years back to do this very same thing because I was addicted to the cryptogram section in the Sunday paper.)

xnbomb wrote:
The double ** at the end of the message is interesting. As yeahyeah noted, those symbols are unlike any others used elsewhere, so they are either filler, or potentially are letters not used elsewhere. It's worth noting that the letters that do not appear anywhere in the plaintext are J, Q, X, and Z. Therefore, it is possible that those two last characters could be any of those letters, and if they are indeed initials it could be a hint as to the name of the author of the message.


Often in block cyphers, you add extra padding to the end of a message so that you have enough data to calculate the final block in full. While this is not quite the same sort of block cypher as you usually run across in crypto, you still need padding at the end because if the geometric shape. Imagine if the last two squares were plain white or plain black. We would instantly know the size of a "tile" (as opposed to stumbling around with tiny 6x6 squares or whatever). I think it is kind of cool that they used symbols outside of the "vocabulary." It simultaneously acts as padding AND misdirection.

Can you tell I'm having fun playing with crypto in the context of a game, as opposed to the secure B2B eCommerce transport layers I deal with most days? Wink Although, if the PM's start giving us PKCS7 documents as part of the game, I will personally murder all of them.
_________________
Y0 Resources / VP Wiki / PXC Catalog / Metacortex

PostPosted: Sat Nov 08, 2003 2:21 pm
 View user's profile Visit poster's website
 Back to top 
xnbomb
Unfettered


Joined: 13 Oct 2003
Posts: 660
Location: J302B S8JDC

 		Re: Addendum to tapestry
BriEnigma wrote:
It can still be solved computationally. If the image parsing is done in the computer, you can do a "fuzzy" match as the symbols are being lexed. When scanning tiles, you compare a tile to all previous ones using slightly fuzzy heuristics. If you get a (to pick a random threshold) 90% or higher pixel match, you can assume they are the same.

Good point ... I was implying that it wasn't easy to do ... but that's not true either: My code is comparing on the basis of exact match. It would be just as easy for me to compare the 'degree of similarity', based on how many pixels are not the same, and then I'd just have to threshold the criterion of how many need to be the same before it is a match (just as you've described). I may just try it ... if I find a threshold that gives the same set as the manual interpretation, that would be awesome. Very Happy

BriEnigma wrote:
Alternately, you can hand-translate the symbols into numbers or ASCII characters. At that point you just bypassed the lexing and can simply do automated work against the Caesar cypher directly.

For it to work as a Caesar cipher or rot-shift, you'd have to assign letters or numbers to the symbols in the right order, which you wouldn't know until you solved it. I guess what I'm saying is that because the symbols don't have an apparent ordering, it's hard to think of that way ... maybe there is one though (that is a way to arrange the symbols such that they follow some logical progression of shapes, the kind of thing you find on IQ tests etc.)

BriEnigma wrote:
Often in block cyphers, you add extra padding to the end of a message so that you have enough data to calculate the final block in full. While this is not quite the same sort of block cypher as you usually run across in crypto, you still need padding at the end because if the geometric shape. Imagine if the last two squares were plain white or plain black. We would instantly know the size of a "tile" (as opposed to stumbling around with tiny 6x6 squares or whatever). I think it is kind of cool that they used symbols outside of the "vocabulary." It simultaneously acts as padding AND misdirection.

Yep, it could be just padding, or they may be characters unused elsewhere in the message ... no way to know without another tapestry sample that uses those symbols (EDIT: or by discovering some structure in how symbols are assigned to letters and punctuation marks ... some kind of order or logical progression in symbols). It is a very elegant piece of work ... I especially like how the symbols are modified so their edges match up, giving that uniform appearance. I'd love to see the code that created the tapestry. Very Happy
_________________
My location is a little tricky, but sooner or later, you'll get the knack.

{J302B S8JDC, 8996N M8L4W, 92D40 Q1JX5, 4PPRN R2B97, 8DC7C NZJNV, 8CH7V Q891H}

PostPosted: Sat Nov 08, 2003 2:34 pm
Last edited by xnbomb on Sun Nov 09, 2003 7:35 pm; edited 1 time in total
 View user's profile
 Back to top 
Giskard
Sassypants


Joined: 07 Oct 2003
Posts: 2066
Location: Chicago
xnbomb wrote:

For example, compare the symbols for the question marks after KNOW and HEARD ... the 'bite' taken out of them to get them to make them match up with the adjacent symbols is not quite the same:


Could it be that the original image was "perfect" in the sense that the symbols were all the same, but that in converting it to a JPEG, stuff got lost in compression? If you zoom in on the white parts of tapestry.jpg, there are a lot of 'dirty' pixels in there, which also look like a remnant of JPEG compression.

Furthermore, I just wanted to say that the amount of depth you guys put in your analysis amazes and impresses me a lot...I feel really humbled by the puzzling abilities of the die-hard unFiction people at times Very Happy
_________________
"They never tell you truth is subjective, they only tell you not to lie." -- Gary Jules

PostPosted: Sat Nov 08, 2003 2:45 pm
Last edited by Giskard on Sat Nov 08, 2003 2:49 pm; edited 1 time in total
 View user's profile Visit poster's website
 ICQ Number 
 Back to top 
bakntime
Unfettered

Joined: 31 Oct 2003
Posts: 462
Location: back in time
Giskard wrote:

Could it be that the original image was "perfect" in the sense that the symbols were all the same, but that in converting it to a JPEG, stuff got lost in compression? If you zoom in on the white parts of tapestry.jpg, there are a lot of 'dirty' pixels in there, which also look like a remnant of JPEG compression.


That's not characteristic of JPEG compression. JPEG might add some "noise", but it's rare (impossible?) for it to turn a while pixel black like that. Plus, the compression is low enough that the artifacts that are seen need to be looked at really closely.

PostPosted: Sat Nov 08, 2003 2:48 pm
 View user's profile
 Back to top 
Transmission3000
Boot

Joined: 19 Oct 2003
Posts: 13
bakntime wrote:
Giskard wrote:

Could it be that the original image was "perfect" in the sense that the symbols were all the same, but that in converting it to a JPEG, stuff got lost in compression? If you zoom in on the white parts of tapestry.jpg, there are a lot of 'dirty' pixels in there, which also look like a remnant of JPEG compression.


That's not characteristic of JPEG compression. JPEG might add some "noise", but it's rare (impossible?) for it to turn a while pixel black like that. Plus, the compression is low enough that the artifacts that are seen need to be looked at really closely.


Looks to me like they scanned whatever it was in as 2 color line art. You get a lot of that jaggedness and errant pixeling that way, some times.

PostPosted: Sat Nov 08, 2003 4:23 pm
 View user's profile
 Back to top 
bakntime
Unfettered

Joined: 31 Oct 2003
Posts: 462
Location: back in time
Transmission3000 wrote:
Looks to me like they scanned whatever it was in as 2 color line art. You get a lot of that jaggedness and errant pixeling that way, some times.


Interesting point.

If this was scan of a hand-drawn picture, that would explain the imperfections in the scan.

PostPosted: Sat Nov 08, 2003 4:48 pm
 View user's profile
 Back to top 
Transmission3000
Boot

Joined: 19 Oct 2003
Posts: 13
bakntime wrote:
Transmission3000 wrote:
Looks to me like they scanned whatever it was in as 2 color line art. You get a lot of that jaggedness and errant pixeling that way, some times.


Interesting point.

If this was scan of a hand-drawn picture, that would explain the imperfections in the scan.


Wouldn't even have to be a hand drawn image, really. You could scan just about anything on your desk into photoshop using the 2 color line art setting and get similar results.

I knew that two years of working at Kinko's would come in useful some day...

PostPosted: Sat Nov 08, 2003 4:51 pm
 View user's profile
 Back to top 
chancesend
Veteran


Joined: 08 Oct 2003
Posts: 128
Location: Colorado, USA

 		Regarding finding the two "missing" characters at the end:

I have been thinking - what if this wasn't a substitution cipher after all? What if these symbols were actually characters just scrambled up? Seems to me that if we found how the descrambling algorithm (which would have to be the same for all characters), we could thus find what the last characters are.

What I'm thinking of would be similar to dividing the individual characters into a 3x3 grid, and then switching the placement of these smaller blocks. So by moving say block [1,3] to [2,1], and [2,1] to [1,1], etc., we could "decode" this exactly.

Just a thought. The fact that those two end characters are totally different has been bugging me...
_________________
--
(Chance's End )
Skyway 12" now on sale! (MP3 of Radio Mix)

PostPosted: Sat Nov 08, 2003 6:36 pm
 View user's profile Visit poster's website
 Back to top 
yanka
Fickle


Joined: 06 Oct 2003
Posts: 1214
Location: undesirable

 		Just wanted to say to chacesend (would have said it earlier, but it took me a while to get caught up): I am kind of in awe of what you did. I think you solved the most convoluted and obscure puzzle here (and don't anybody bring up the evil Oct. 1 picture up Smile ), and I couldn't be more impressed. Thanks!
_________________
Annushka has already bought the sunflower oil, and has not only bought it, but has already spilled it.

PostPosted: Sat Nov 08, 2003 6:41 pm
 View user's profile
 Back to top 
stavro510
Greenhorn

Joined: 23 Oct 2003
Posts: 5

 		My take:

I KNOW THAT MY DAD TRUSTED YOU AND I COULD USE YOUR HELP.
Caesar is saying that his dad (Ethan?) trusted Dmitri. [I say this because so far, Caesar is the only character who mentions parents]

I DON'T KNOW WHERE ELSE TO TURN.
This is his last resort

IT HAS BEEN TOO LONG, I SHOULD HAVE HEARD BY NOW.
No one replied to the NOT files.

NONE OF OUR PLANS HAVE WORKED OUT.
The Paintover/Tavern.Pointe system has failed.

I KNOW THAT THEY ARE STILL OUT THERE BUT WHERE?
What happened to the other hackers?

DO YOU KNOW? HAVE YOU HEARD? CAN YOU HELP?
Self-explanatory.

THIS IS ALL MY FAULT.
I WISH I UNDERSTOOD WHAT I DID.
He is unsure why the other hackers (Scratch et al) abondoned the Taverne.Pointe. Caesar (Jesse?) still doesn't understand why Scratch got so mad.




And kudos to chancesend for solving this difficult puzzle.

PostPosted: Sun Nov 09, 2003 1:49 am
 View user's profile
 Back to top 
xnbomb
Unfettered


Joined: 13 Oct 2003
Posts: 660
Location: J302B S8JDC

 		Addendum II
BriEnigma wrote:
Can you tell I'm having fun playing with crypto in the context of a game

Me too ... I find the tapestry puzzle fascinating ... that's why I'm still analyzing it. If you (meaning any of you unforum readers out there) don't find it (and its properties) interesting, please don't read this post, you'll find it boring and unimportant.

One reason I've been pursuing understanding some of its properties is that I'm a little obsessed with the idea that one can create puzzles that are easier for a person to solve manually than with a computer, because in the context of this game, that has an implication for why that sort of puzzle or code was potentially chosen to hide information. At first, I was convinced the tapestry code was such a puzzle. I still believe that it is easier to solve by hand than by computation. Even knowing the answer now, it's not as easy (or nearly as quick) to conceive software to do what chancesend did with his eyes and brain, namely identifying which symbols are the 'same'.

However, it's not as hard as I initially thought either. My initial approach was that symbols should be pixel-for-pixel identical to be the same. This proved to be a bad idea, especially since it is clear the symbols are each modified slightly to make the edges of symbols match up better so it is less obvious it is a code, and looks more like a meaningless holistic pattern (although, even if you use the strict definition of pixel-for-pixel identical identifying each kind of symbol, which gives 94 symbols, it's still solvable, you just have a much less bounded substitution problem that could still be solved with sufficient brute force, although there might be multiple solutions that are equally parsimonious). With that in mind, it's clear that small differences between multiple examples of the 'same' symbol need to be allowed. BriEnigma intelligently suggested that this did not preclude a computational solution: One just needs to use a more flexible evaluation of 'sameness'.

The easiest evaluation of sameness between two symbols (made up of binary colored pixels) that I can think of is to determine how many pixels differ between them. That is, one superimposes one symbol on top of the other and counts how many pixels are not the same color (this is made a little more tedious because symbols can be rotated and need to be checked rotated as well, but that's not a conceptual problem, it just creates 4 times as many comparisons Very Happy ). Because I already have the solution (thanks chancesend!) as to which symbols are the same (letters assigned so that each group of 'same' symbols is already defined), it was fairly easy for me to write code to evaluate this on a per letter basis. The two quantities I calculated are the maximum difference in pixels between all examples of each letter, and the minimum difference in pixels between any example of each letter and all examples of all other letters (effectively the upper limit difference of a positive match and the lower limit difference of a negative match). If the first quantity is much less than the second in all cases, then code that compares symbols on the basis of how many pixels they differ by, in combination with a properly chosen threshold difference in pixels could successfully match the 'same' symbols.

Except for the presence of two unusual characters in the tapestry (no, not the ones you think), this method would work easily for the set of symbols in the tapestry puzzle. If one ignores these two characters for a moment, the maximum difference within any letter is 14 pixels, and the minimum difference between a letter and all other letters is 34 pixels, so any threshold in between would work just fine (E and F use very similar symbols to produce this smallest difference of 34 ... is this a hint symbols are assigned in some alphabetical or letter shape oriented fashion? ... ignore their similarity and this latter figure jumps up to 49 pixels). Incidentally, since these symbols are 32x32=1024 pixels, we're talking about very small differences that would be imperceptible or easily ignorable when working by eye.

The two troublesome characters are one example of the letter F and one example of the letter I. These two characters are unusual because instead of having the occasional pixel different from their same letter symbols, they appear to have the whole symbol shifted a pixel to one side, which causes a much larger number of different pixels (58 between the shifted F and the other example of F, 66 between the shifted I and all other examples of I). It seems likely to me that those symbols ended up shifted to the side in error ... the 322 other symbols are each much more similar within each letter.

It's worth noting that a more sophistication evaluation of 'sameness' based on the detection and comparison of shape (which the human eye does easily and ever so fast) would have no trouble with the problem that these two characters pose. However, writing code to do that would be a fair bit of work because the definition of 'sameness' needs to be even more flexible, and more possibilities need to be considered. As it is, an evaluation based on the number of different pixels with a threshold in the range described would work for over 99% of the characters ... you'd have to fix just those two.

So, because we have a manual solution in hand, it was easy for me to evaluate how effectively this automated solution could have been applied. It still would have been a real pain to find an effective pixel difference threshold flying blind, because you still really would have had to evaluate the results by eye to tune the threshold, so you still really would need some manual input to solve this effectively. Even an approach that uses morphological analysis and some fuzzy logic still requires someone to make a decision on the appropriate threshold of 'sameness'. This is all before the substitution cipher part of the solution, which is the real proof if you've identified which symbols are the 'same' correctly.

Without some subjective judgement of appropriate 'sameness' of symbols (because they are not identical pixel-for-pixel), it would take considerable computation cycles to solve this puzzle ... you'd have to try a range of thresholds of 'sameness' to produce sets of same symbols, then solve the substitution problem on those sets, and based on the level of success of the result of the two steps together (how reasonable the plaintext looks ... complicated by the few typos in the ciphertext), deciding which is the best solution that can be found (or as suggested near the top, ignore the symbol set problem using the strictest definition of same symbol and attack a nastier substitution problem).

Incidentally, simply trying to identify characters by counting the number of black pixels they contain almost works. That's even simpler, and it ignores the arrangement of the pixels entirely. There are a few letters that have nearly the same black pixel counts, but not many.

P.S. There was no typo in the first instance of HELP. ... one nice side effect of running this analysis was that I found my own mistake ... statistics of symbol similarity showed me my error Very Happy
_________________
My location is a little tricky, but sooner or later, you'll get the knack.

{J302B S8JDC, 8996N M8L4W, 92D40 Q1JX5, 4PPRN R2B97, 8DC7C NZJNV, 8CH7V Q891H}

PostPosted: Sun Nov 09, 2003 2:20 am
Last edited by xnbomb on Sun Nov 09, 2003 2:30 am; edited 1 time in total
 View user's profile
 Back to top 
Display posts from previous:   Sort by:   
Page 7 of 10 [149 Posts]   Goto page: Previous 1, 2, 3, 4, 5, 6, 7, 8, 9, 10  Next
View previous topicView next topic
 Forum index » Archive » Archive: MetaCortechs » MetaCortechs: Puzzles
Jump to:  


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum
You cannot post calendar events in this forum



Powered by phpBB © 2001, 2005 phpBB Group