Anti - censorship

A new way of beating the web’s censors
.
Steganography, the art of hiding things in plain sight, is a trick as old as espionage.
.
Unlike its cousin, cryptography, which makes no attempt to disguise the existence of a message, but rather hides its meaning, a steganographic message need not be encyphered.
.
What it does need to be is invisible—at least to those who are not the intended recipients.
.
And that, in the modern world of the internet, is a crucial distinction.
.
A censor can block a message he mistrusts, even if he cannot read it, thus putting the onus on the recipient to justify both the message and the fact it is encrypted.
.
A well-crafted steganographic message, though, will never come to the censors’ attention in the first place.
.
Which is the purpose of Collage, a system devised by Nick Feamster and his colleagues at the Georgia Institute of Technology.
.
Traditional steganography hides its message as, say, every 20th word in a letter, or as the colour of every hundredth pixel in an electronic image.
.
Sophisticated analysis of such things might, though, notice something odd and thus flag a document for closer examination.
.
Collage escapes notice by dividing the message into pieces, and then hiding these in electronic files posted to public websites, such as Flickr, Twitter and YouTube.
.
Scattering the message among multiple files and websites offers a number of advantages.
.
For instance, the small amount of data in each file makes it difficult for a censor to notice anything odd unless all traffic on the network is subjected to advanced analysis techniques.
.
Though possible in theory, the cost and effort of doing so makes this unfeasible in practice.
.
More importantly, Collage’s design allows reconstruction of the original message even if only 60-80% of the files hiding it are recovered.
.
Thus, even if a censor manages to block some of the files, users are still able to communicate.
.
To pull a message together requires the execution of what Dr Feamster refers to as a “task sequence”—a series of actions, chosen not to arouse suspicion, that must be performed to locate the files hiding the message.
.
For example, if images of the Himalayas are used as cover media, the associated task sequence might be to search specific sites for images tagged with the words “mountain” and “glacier” and download a defined number of images in a set amount of time.
.
The weak point of the system is the transmission of the initial task sequence.
.
That is probably best done off-line.
.
Once the system has been “bootstrapped” into existence, though, subsequent task sequences can be passed on as part of the hidden messages.
.
Although much of Collage’s design focuses on avoiding detection, it also provides a measure of deniability should detection occur.
.
The cover media used to hide data and the actions performed to locate and recover messages mimic those of innocent users, allowing Collage users to argue that they were unaware of the presence of hidden data.
.
Collage also aims to mimic the traffic patterns of legitimate users, so that digital fingerprints which may suggest illicit behaviour are not produced.
.
Finally, hiding messages widely amid otherwise-legitimate content makes it difficult for a censor to block communication effectively without also causing significant disruption to innocuous users.
.
For governments trying to walk the fine line between access and censorship, this reluctance to block large quantities of legitimate content in the hope of also blocking prohibited content can be crucial.
.
Even if the hidden messages are known to be present, there may be little the censor can do about it.
.
Wired