[409] in Privacy_Forum
[ PRIVACY Forum ] Sending Secret Messages Via Google's SearchWiki
daemon@ATHENA.MIT.EDU (privacy@vortex.com)
Tue Nov 25 21:40:39 2008
Message-Id: <200811260215.mAQ2F5HT028527@chrome.vortex.com>
To: privacy-list@vortex.com
Date: Tue, 25 Nov 2008 18:15:05 -0800
From: privacy@vortex.com
Reply-To: PRIVACY Forum Digest mailing list <privacy@vortex.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Errors-To: privacy-bounces+privacy-forum=mit.edu@vortex.com
Content-Transfer-Encoding: 8bit
Sending Secret Messages Via Google's SearchWiki System
http://lauren.vortex.com/archive/000476.html
Greetings. I've previously written of some concerns associated with
Google's new "SearchWiki" feature, a concept with considerable merit
in theory, but as currently implemented appears to be easily abused.
( http://lauren.vortex.com/archive/000474.html ).
Up to now I've noted the issues of "electronic graffiti" and similar
problems associated with the current unmoderated SearchWiki
environment, which also lacks site opt-outs, simple user opt-outs,
or useful comment control mechanisms of the sort that do exist for
Google's YouTube ( http://lauren.vortex.com/archive/000473.html ).
It appears that SearchWiki also provides an interesting platform for
the global distribution of secret messages. By allowing users to
attach publicly viewable, arbitrary comments to virtually any URLs
returned in Google search results, SearchWiki can be used as a
gigantic "spread-spectrum" secret message transmission medium.
Users need only associate individually innocuous message fragments
(which may themselves be encoded in a wide variety of ways) as
comments to the limited set of URLs returned by obscure Google
search queries.
Without knowing the exact search query used to "aggregate" the
comment set in any specific case, outside parties who might stumble
across individual message fragments (as comments on arbitrary URLs)
would be extremely unlikely to recognize them as parts of a coherent
message, and would have no simple technique to locate the other
parts of the message in any case. This is a key attribute of the
described technique -- a message that is dispersed in this manner is
unlikely to even be recognized as a message worthy of attention or
log-based analysis.
Before I present an example, a few additional points of note.
Users must be logged in to Google with their Google accounts to
attach or see SearchWiki comments. While for this example I created
all message fragments with the generic nickname "Searcher" from a
single account, in practical use it is likely that users would
submit fragments from different accounts, and probably from different
IP addresses. Google accounts can be easily created in just a few
minutes, and using different accounts would not only help to defeat
algorithmic methods that might be aimed at slowing comment
submissions, but also could complicate associated forensic analysis
of activity logs. By staging message fragment submissions in
various ways over time, algorithmic detection of message creation
patterns could also be made more difficult.
While Web pages and search results are typically ephemeral in nature,
initial experiments suggest that they are stable enough over periods
of time that would be useful for the sending of messages as described.
For this trivial example, I have created ten simple plaintext
sentences that are attached to ten URLs. I cannot guarantee how
long these comments will be present, and other comments may be added,
since this query will now be widely publicly known (which wouldn't be
the case in a practical use of this technique).
Each comment has been sequence numbered in an obvious fashion for
demonstration purposes. Comments were individually created then the
"promote" up-arrow was selected for each associated URL (this
"promotion" step appears important to enable the rapid appearance of
comments to other Google users).
In practical use, sequence information would most likely be kept
more obscure. Keeping the actual comments in some form of plaintext
would help avoid possibly being deleted as "unintelligible" text,
though of course the actual meanings of the text words could be
subject to code-based obscuring techniques. With a bit of effort,
any data, including images, could probably be transmitted in this
manner.
Only *one word* in each of the ten comment sentences for this basic
example is part of the actual secret message. Your mission is to
derive the complete secret message, which is currently dispersed
across the space of Google's search database.
To aggregate the message fragments, you need to enter the
appropriate search query into Google, while logged into a Google
account. Then select the "See all notes for this SearchWiki" link
at the bottom of the page. You should then be able to see comment
links associated with a number of the associated URLs. Click on the
links as necessary to view the actual comment texts, remember the
encoding procedure I outlined above, and the secret message is yours.
One final thought before we begin. If I can use this technique, we
can be sure that other parties can use it as well, including entities
who would figure it out on their own without ever seeing this blog item.
Forewarned is forearmed.
Special thanks to Lou Katz of Metron Computerware for his assistance
in the testing of this technique.
OK, let's start. The secret message keyword search string is:
orthogonal terwilliger accordion
(Click http://lauren.vortex.com/searchwiki-secret to visit the
appropriate Google search results page for this query.)
Be seeing you.
--Lauren--
Lauren Weinstein
lauren@vortex.com or lauren@pfir.org
Tel: +1 (818) 225-2800
http://www.pfir.org/lauren
Co-Founder, PFIR
- People For Internet Responsibility - http://www.pfir.org
Co-Founder, NNSquad
- Network Neutrality Squad - http://www.nnsquad.org
Founder, PRIVACY Forum - http://www.vortex.com
Member, ACM Committee on Computers and Public Policy
Lauren's Blog: http://lauren.vortex.com
_______________________________________________
privacy mailing list
http://lists.vortex.com/mailman/listinfo/privacy
