WikiLeaks Posts Mysterious 'Insurance' File

Originally posted by JBA2848
Here a web site that shows his charges and he illegally used secret internet protocal router network (SIPRNET) To gain access to a secret microsoft office power point presentation. That could be the excel streaming file.

I don't see it that way. What it says is that he "intentionally exceed his authorized access on a Secret Internet Protocol Router network computer", and to me it means that he took the files from that computer. So, if I'm right, he is not being accused of accessing the network, he his being accused of accessing files he wasn't authorized to access on a computer for which he already had access.

I guess the streaming file is all excel files changed to binary and sent out. Since the computers using this excel stream know what the files are there using the normal headers of the files for other data.

What streaming file? In fact, what is a "streaming file"? And what is an "excel stream"?

I guess most devices that are Posix are just compliant and not true Posix systems. I think Windows ME is the only Posix proven Windows.

I don't think that Windows ME was Posix compliant, that Posix subsystem is (or was) based on Windows NT code, meaning that only Windows NT or 2000 had it. Windows Vista and 7 have it already built in.

It's easy to get a full Posix operating system just by downloading it.

I think Assange has gained access to this data stream only on the lower PKI area and not deep enough to get to any thing really secret.

It's much more difficult to get access to data streams than to files.

PS: could you please answer the questions from my previous post? Thanks in advance.

reply to post by misinformational


I exported my SQL database to an Excel file, and the 76,911 records resulted in an (uncompressed) 44,5 MB (46.743.396 bytes) Excel file.

That means an average 608 bytes per record, so the remaining 15,000 reports would be something like 8.7 MB and the 250,000 something like 145 MB.

1.4 GB is a lot of data.

Originally posted by ArMaP
1.4 GB is a lot of data.

Yes it is and considering it could be a 7zip archive make it feel so encyclopedic!

I supposed many things about this file, for example a TrueCrypt volume or a n-times-encrypted URL in a single text file.
It may be everything, although i had to rethink about it a bit: Julian said in some last interview he just need to tell the password, so i presume he want anyone to access it with that password, and not just the few capable of performing exotic magic with it?

I mean, he could have raised the number of rounds performed during encryption, from 14 to 20 just to be sure no one could ever crack it, but then only coders could have access to it at first, so this wouldn't mean exactly access to anyone

So this thing should be probably accessible by anyone, once the passphrase/key+iv is/are given: just wondering..

I hear they are trying to find Chuck Norris to break this file, you dont break chuck he breaks you.

Sillyness aside, will this make any real difference to anybody, they have the power and we are all just too lazy to care. Thats what makes us human.

reply to post by ArMaP


You make the assumption that all the data from each row would be taken from the excel file as-is.
It might have been stripped from some columns, optimized for size. And most likely would have been. As storing this data in an excel file is ridiculous in the first place, most likely the row format in which it was done was even worse.
1.4 GB would be about 15 KB per row, which is not that outlandish.

Originally posted by the.krio
You make the assumption that all the data from each row would be taken from the excel file as-is.

Obviously, it's the only thing I have.

It might have been stripped from some columns, optimized for size. And most likely would have been.

Why, are you thinking about a specific format?

I don't know if you know the structure of the file that was published (the one with 76,000 records), but if you don't (and for all those that don't know it and may be interested), here are the columns for each record.

ReportKey varchar(255)
Date varchar(255)
Type varchar(255)
Category varchar(255)
TrackingNumber varchar(255)
Title text
Summary text
Region varchar(255)
AttackOn varchar(255)
ComplexAttack varchar(255)
ReportingUnit varchar(255)
UnitName varchar(255)
TypeOfUnit varchar(255)
FriendlyWIA varchar(255)
FriendlyKIA varchar(255)
HostNationWIA varchar(255)
HostNationKIA varchar(255)
CivilianWIA varchar(255)
CivilianKIA varchar(255)
EnemyWIA varchar(255)
EnemyKIA varchar(255)
EnemyDetained varchar(255)
MGRS varchar(255)
Latitude varchar(255)
Longitude varchar(255)
OriginatorGroup varchar(255)
UpdatedByGroup varchar(255)
CCIR varchar(255)
Sigact varchar(255)
Affiliation varchar(255)
DColor varchar(255)
Classification varchar(255)

As storing this data in an excel file is ridiculous in the first place, most likely the row format in which it was done was even worse.

Why? What are you assuming to say something like that?

1.4 GB would be about 15 KB per row, which is not that outlandish.

Uncompressed, right?

Text compresses to something between 10% and 15% of the original size, so a compressed 1.4 GB file (and I think that assuming that the file is compress it's not an outlandish assumption) would be something like an original, uncompressed, 9 or 10 GB file or files.

Originally posted by ArMaP
Uncompressed, right?

Yes, point is, don't underestimate the idiocracy; if what Wikileaks got was an excel file, it could have been 1.4GB originally. Uncompressed
Assange however is supposedly smarter than that, so it might make some tiny bit of sense only if they've decided to upload the original file they've got as-is, just like they received it.

Anyhow, I'm currently looking into breaking AES simply with neural nets.
Surprisingly, breaking any block cipher like that is not a well researched area..
But here is a tiny article from 2009 on Feistel in general.
The problem with most symmetric-key block ciphers IMO is that their security is not based on well defined math problems, but rather on complexity which yields inability to define how to extract the key. Which leads to expiration dates on them, if the tech doesn't catch up to make it feasible to bruteforce, sooner or later someone finds a way to narrow the search space significantly.
Meanwhile we have neural networks, genetic algorithms, state vector machines and even some nifty unresearched probabilistic hypergraphs and all kinds of combinations of them which supposedly should be able to solve problems based on complexity and lack of our knowledge.
Prove me wrong before I start writing code.

[edit on 16-8-2010 by the.krio]

Does this make any sense to anyone?

Interesting huh?!

reply to post by qualitygossip


I found the above link here...

reply to post by qualitygossip

Yeah, it's another distraction.
The whole forum apparently is installed for 2 days.
Ignoring this until someone posts the unencrypted archive.

Edit:
Aaand the domain is 3 days old :
------
Domain Name: PGPBOARD.COM
Registrar: PLANETDOMAIN PTY LTD.
Whois Server: whois.planetdomain.com
Referral URL: www.planetdomain.com...
Name Server: NS1.WISERHOSTING.CO.UK
Name Server: NS2.WISERHOSTING.CO.UK
Status: ok
Updated Date: 13-aug-2010
Creation Date: 13-aug-2010
Expiration Date: 13-aug-2011



Edit 2: The yahoo mail list on the other hand looks interesting but is dead end imo. Apparently there was a pgpboard which was indeed operated by Alan Taylor from Philippines, see here. But I don't see a solid link between those folks and the pgpboard.com other than the cryptome. Why aren't they signing their messages with PGP?
Strangely, this is starting to remind me project Serpo BS back in the days..

*Still waiting for a torrent of the unencrypted archive.... *

[edit on 16-8-2010 by the.krio]

[edit on 16-8-2010 by the.krio]

Ok, Assange seems to be off the grid since Aug 13
And someone is apparently phishing IP addresses off this thread and others
This is getting fun, love it
May I see some spaceships now?

Originally posted by misinformational
So if Manning only had access to the SIPRNET, he only had access to data classified Secret. And since we haven't heard of any leaked TS or SCI data, then we should be inclined to believe that this was the case.

I forgot to answer this post, but not completely, as I am answering now.

According to the Wikileaks War Diary file, the cables (or whatever is in that file) have a column for "Classification", and these are the 16 unique classifications.

(empty)
CONFIDENTIAL
NATO CONFIDENTIAL
NATO CONFIDENTIAL Releasable to GCTF
NATO RESTRICTED Releasable to GCTF
NATO SECRET
NATO SECRET Releasable to GCTF
NATO/ISAF CONFIDENTIAL
NATO/ISAF CONFIDENTIAL REL GCTF
NATO/ISAF CONFIDENTIAL REL GCTF, GIRoA
NATO/ISAF RESTRICTED REL GCTF, GIRoA
NATO/ISAF SECRET
NATO/ISAF SECRET REL GIRoA
None Selected
SECRET
UNCLASSIFIED

No Top Secret classification.

reply to post by ArMaP


You are correct. That data would be rightly stored within the SIRPNET, not JWICS - so the charges all add up so far.

This waiting game isn't very exciting

[edit on 17-8-2010 by misinformational]

Originally posted by misinformational
reply to post by ArMaP

 

This waiting game isn't very exciting
[edit on 17-8-2010 by misinformational]

I disagree
It's getting really entertaining if you know where to look.
For example, note that the twitter feed is dead atm: twitter.com...

reply to post by the.krio


Don't underestimate the difficulty of implementing the necessary algorithms in a way that actually performs well and accomplishes your goal. Ever written any actual AI code before? Constraint satisfaction code? If not, you've got a giant hill to climb and I still believe that with the current state of technology, leveraging a neural net isn't ultimately going to result in you cracking this file open. Try writing some code to solve a simple map coloring problem. If you can't do that, you can't do what you're proposing, which will require more research and coding time than I think you realize.

Take a look at the four color theorem, then write code that understands how to shade a map of the United States with no more than four colors, such that no single color is adjacent to itself. Write it such that the code can solve the problem from any starting point. In other words, write it such that a user could pick to start shading in Florida, or Idaho, or whatever. Don't write it such that it always shades from the same places outward. If you don't want to get deep into the need to provide a GUI, simply output the name of the state and the name of the color and verify against a map that the output would solve the problem.

If you're still feeling confident after that, by all means give it a shot. I'm all for personal code growth. You may not crack the file, but certainly you'll learn something along the way, which has its value.


[edit on 17-8-2010 by bikeshedding]

reply to post by bikeshedding


I do have experience with nlp (pos tagging, np chunking, ner and such), svr for example..
Not neural nets in particular yet though
And sure, I'm not going to write anything from scratch, the saying goes - the less code you write the smarter you are Offshoring some of the stuff to GPU would be lovely as well.. Cuda, GLslang, cg are in the experience list too.
Also, I don't want to waste my time debugging c code for this little endeavor.
So, I'm currently choosing the right python packages for the job, any specific recommendations?

reply to post by the.krio


I did find that feed a while back and subscribed.. And its true that there is some degree of fun in trying to put cryptic meaning to those tweets.

But still - No conclusions may be made, at least not to the questions that never seem to get answered.

IMO, Wikileaks should have never cast itself into the limelight and should have relied on the data to speak for itself... Their decision to do such, seems to be removing focus from the leaks and focusing it on the messenger - I believe WL needs to get back to leaking information and get out of the press conferences.

They would be much better served by never commenting on anything. Commentary is not necessary when you have access to the type of information they do.

reply to post by misinformational


Yeah, I'm not talking about their views or extracting meaning from cryptic messages. They aren't cryptic.
Look at the dates.

reply to post by the.krio


Sorry, never bothered with Python in depth, so I can't be of any assistance there.

Good luck.