Webdriver Torso - Secret messages or nonsense?

Thread Starter

DrRich

Joined Apr 21, 2014
5
Just wondering if everyone on here has heard about Webdriver Torso.

Essentially there are almost 80,000 videos on this YouTube channel, which consist of randomly sized boxes and tones:
http://www.youtube.com/channel/UCsLiV4WJfkTEHH0b9PmRklw

The format of the videos is fixed and they appear random. Guessing it is either some sort of stunt, reserchers using youtube or it is a way of sending secret messages. More inclined on the latter two.

Anyone got any ideas, views or comments?
 

Sparky49

Joined Jul 16, 2011
833
Yeah, it's quite interesting.

I believe that the theory that it is a modern day numbers station is rubbish - unlike radio reception, IPs can be traced. A police force in 'Kasboloistahn' sees one IP visiting this youtube channel twice a day... Rings alarm bells.

My guess is either some secret 'cicada 3301-esq' competition, or just some random videos put up to cause confusion. :p
 

Thread Starter

DrRich

Joined Apr 21, 2014
5
I thought about cicada as well, but it seems a little less direct than what I know about them.

As a one way system with some form of one time pad, youtube may be acceptable. If an operative was hopping about the country, any broadcast method will do and the IPs will seem pretty random anyway. Apart from all the state monitoring that goes on, I guess only youtube could provide a list of IPs accessing particular videos. I remember reading about the authorities seizing a lot of servers offering pirate software and other nasty stuff, mainly for the list of IPs that had accessed it. I maybe wrong on this, I am an hardware engineer. I have worked on cryptography systems and find them incredibly interesting.

Not entirely sure what to think about this though.
 
Last edited:

Sparky49

Joined Jul 16, 2011
833
True, I am more hardware orientated (mainly into RF stuff), but the more I learn about software and various network models, the more it seems that anything can be done. Think that you've got a secure link? Give it a week and someone will have broken it. Give it another week and it might just be posted all over the internet. :p

Of course, looking back at the espionage aspect, what benefit does the Internet have over shortwave radio? Radios are not uncommon, so would not arouse suspicion (even on mobile phones), and they are also very mobile. Radio can be received just about anywhere and is untraceable, so you could have an operative in one location for months, years at a time.

Whereas with the internet approach you first need an internet connection. This straight away limits you to within a few meters of a port, or within wifi range. You've also got to pay for these, which leaves behind a trace. Then there's the traceability of the data, which although it may be hard, is certainly possible. Very likely too, if you want your operative to be in one location for extended amounts of time. Then there's the technology involved. It may not be such a problem in better-off countries, but if you were to walk around some less well-off areas, having something like a laptop or ipad would drag attention. A radio would be much less conspicuous.

Assuming you were arrested, there would be no trace of what you listened to on your radio, but easy information on your hard drive.


I guess, it is a possibility, but I would be extremely surprised if it were. There doesn't seem to be a good reason to change from the old to the new, although that doesn't stop it from happening. :p

Good thread.

Sparky
 

Thread Starter

DrRich

Joined Apr 21, 2014
5
Never thought about paying for the internet :rolleyes:. Where I am based there are many free access points in restaurants, coffee/sandwich places etc. You are correct on traceability though, it has cast doubt in my thinking.

Then again, there are rumours that ebay listing images have been used for secret messages as well - I guess commonly visited sites do not draw the attention of such monitoring. Still have the problem of leaving traces of info on the hard drive. I guess someone more software "savvy" could make sure it is not a problem.

A well implemented one time pad is perfectly secure, look it up. Essentially you take the XOR (addition modulo 2) between your message bit and a key bit. The key must be perfectly random and the same length as the message. The security of the channel does not matter with a OTP (you must have some way of authenticating either party mind, but this can also be done on a public channel. Typical methods for this are to send a little bit of the key to either user, verify it matches and discard).

Very cool and painfully simple. There is of course the key distribution problem (i.e. how to get those key bits shared between us securely without meeting in person), which Quantum Key Distribution solves. It maybe usable by everyone if the technology ever gets cheap and small enough for everyone to use and its use does not get banned by statute :p

n.b. I guess I am in some sort of limited newby forum status. No posts yet, and no option to like other posts...
 

nsaspook

Joined Aug 27, 2009
13,079
It's pretty low data rate if it is a coded transmission. The possible code permutations could be fairly high so a digital code book system could be used. We normally used short sequences of character code groups for authentication/receipt messages because most people could memorize and send the sequences easily. So my WAG if it's a communications channel that it might an acknowledgement signal for some other one-way broadcast or it might just be nonsense.
 

WBahn

Joined Mar 31, 2012
29,976
I'm guessing that it is either some research project or some quality control process. Imagine that you make video processing software that you expect will be used in situations in which the output will be uploaded to YouTube or undergo other significant image compression. So you get a YouTube account, generate a bunch of random video, upload them, then download them and compare them to the original and sort them according to the amount of distortion. You can then use that to identify the kinds of signals that get distorted the most and tweak your algorithms so as to produce video streams that get distorted less.
 

tracecom

Joined Apr 16, 2010
3,944
Or it could be some 13-year-old who figured out how to program his computer to generate vids and automatically upload them to You Tube in order to amuse and amaze his/her friends.
 

Sparky49

Joined Jul 16, 2011
833
Yeah, but to upload 77'000 videos? That is some impressive bandwidth. I would ground my kid to next century if they did that. :p
 

Thread Starter

DrRich

Joined Apr 21, 2014
5
Haha :D, reminds me of this http://hugelolcdn.com/i700/60160.jpg

Still puzzled as to who would go to this effort. The research idea seems less likely in my opinion now, surely if you was going to automatically test parameters for optimum youtube video quality, you would use close to real life video inputs, rather than blocks and tones.

It is weird none the less.
 

poopscoop

Joined Dec 12, 2012
140
It's not covert communication, put your tinfoil hats down. That'd be very crappy tradecraft. If it looks out of place to a casual person, it's breaking every rule.
 

poopscoop

Joined Dec 12, 2012
140
I'm guessing that it is either some research project or some quality control process. Imagine that you make video processing software that you expect will be used in situations in which the output will be uploaded to YouTube or undergo other significant image compression. So you get a YouTube account, generate a bunch of random video, upload them, then download them and compare them to the original and sort them according to the amount of distortion. You can then use that to identify the kinds of signals that get distorted the most and tweak your algorithms so as to produce video streams that get distorted less.
When you hear hooves, think horse, not zebra. WBahn gets it.
 

WBahn

Joined Mar 31, 2012
29,976
Haha :D, reminds me of this http://hugelolcdn.com/i700/60160.jpg

Still puzzled as to who would go to this effort. The research idea seems less likely in my opinion now, surely if you was going to automatically test parameters for optimum youtube video quality, you would use close to real life video inputs, rather than blocks and tones.
Not necessarily. Using real video inputs means that your test signals are far too complex to analyze well. What is it about these five images that resulted in greater distortion than those seven images? When we detect distortion, what is it that is being distorted when we have a complex image. With lots and lots of simple images you have the opportunity to identify building blocks that lead to distortion much more unambiguously.

Not saying that this is what it is, just that it seems plausible to me.
 

Thread Starter

DrRich

Joined Apr 21, 2014
5
Modern day video compression is far more complex than just static objects. For example, motion estimation has been included in video codecs for ages.

There are numerous software tooilsuites and techniques out there to measure video quality objectively rather than subjectively on more realistic video sequences (SSIM, PSNR, VQM etc) so the testing route may be true, however, testing video quality on such a small subset of possible conditions (essentially a very limited "test card" test) on 77,000 videos is a bit "dedicated".
 

Sparky49

Joined Jul 16, 2011
833
77,000 videos is a bit "dedicated".
This.

I guess we don't know alot for certain, but we do know that it is most probably very dedicated.

There's also the mysterious video of the Eiffel tower in there. :\

Also, what's so special about "tmpdKHvbS" that made WT 'like' it?

Sparky
 
Top