Numbers Stations Move From Shortwave To VoIP

IO ERROR writes "For decades, intelligence agencies have been sending secret messages to their agents in the field using shortwave numbers stations broadcasting encrypted messages for all to hear and puzzle over. Now someone is putting numbers stations on VoIP telephone numbers for anyone to call, and posting messages to Craigslist to alert the recipients to the existence of their messages. One of them went up last month and now a second one has appeared. Will there be a third? Who's behind them? And can you crack the code?"

Comments

Doesn't that defeat secrecy?

[gd23ka]

I know it's the first thing that comes to mind but I'm sure They can monitor who calls ("tunes into") that phone number regularily. Broadcasts are anonymous and many people own shortwave radios, VOIP can be traced to a subscriber so what gives?

Re:Doesn't that defeat secrecy?

[EverLurking]

Given that all traffic on the internet can essentially be logged and tracked, why use VOIP at all?

Just put some PGP encrypted data into an image via Steganography and post it to someplace popular like...lets say...CuteOverload.com

Then if your recipient was tracked going there, he would just look like any other Cats In Racks fan looking for cute kittie pictures : )

Eh, ok

[dk.r*nger]

The point of shortwave is that you can listen from anywhere, undetected.
Calling a phone leaves a bunch of traces. There is really no discreet way for our man in Havana to call longdistance and listening to numbers for a few minutes.
You could just put the numbers on a free website somewhere, or use email..

Re:it's a joke

[Yehooti]

Maybe just to get our knickers in a knot.

Re:it's a joke

[abscissa]

there are a lot more sophisticated ways of hiding and transferring secret information. including texts, images, video and audio streams, ssl, gpg, tunneling, etc. why would anyone bother with a child games like that?

Because the publicity and simplicity of the cipher makes it very difficult to determine the intended recipent. Also, it may not be AES, but if it's a one-time-pad it's pretty damn secure.

Re:Silly

[Dachannien]

I'm not sure about that. VoIP has an unrelated and legitimate commercial interest behind it to support it not being blocked solely based on the virtue of it being VoIP, and thus anybody wanting to jam/block particular VoIP calls would have to know ahead of time at what phone number the message was going to appear. And there are a lot of phone numbers ;)

On the other hand, the various intelligence services have some pretty powerful jamming equipment that can render shortwave transmissions at least partially unintelligible. Numbers stations often work based on a schedule (in terms of both time and radio frequency), and once an intelligence service determines this schedule, they can wash out the frequency with crap at the appropriate time. Since the whole reason for numbers stations are that spies in the field are relatively incommunicado with their handlers, figuring out the schedule can have a fairly long-term impact on the spy being able to receive information and orders. In fact, the biggest question is likely where the jammer antenna should be positioned to ensure that the transmission will be jammed.

It's a *code* not a cipher

[crmartin]

Very likely you can't easily crack the code. reason: it's a true code, not a cipher. A real code assigns a symbol like '34187' to a word or phrase arbitrarily. Unlike ciphers, true codes are very difficult to crack without getting the key somehow, because there is very little redundancy to exploit statistically.

Re:It's a *code* not a cipher

[Jurph]

I don't think it is a code. Try dropping the zeroes between each non-zero 2-digit number, allowing "00" to be a legitimate pair and assuming the occasional 3-digit string. You get a sequence that has 45 unique numbers (in the case of Group 617, 41 unique numbers) and a frequency distribution that doesn't seem all that likely for a pad's vocabulary. Add to that the fact that none of the numbers in either group exceeds 112, and the pad you were using would be working with a very small vocabulary. I can't dispute that it could be, for example, an agreed-upon newspaper from the group-date (415 = April 15 edition of the New York Times, front page, main article). The numbers could refer to word-order in that article. But right now, I'm treating it as some form of substitution cipher.

Re:blog comment spam

[AnyoneEB]

Five number/letter blocks are pretty much the traditional way to break up enciphered messages. That does not mean they must be from the same source.

Obvious first step, and some frequency analysis...

[Jurph]

Nobody seems to have pointed out anywhere that these codes, while broken into the canonical 5-digit groupings, are almost certainly composed of three-digit numbers padded with zeroes. The first one, when you strip out the annoying five-character spacing, becomes:

Group 415
13 56 51 12 79 46 65 10 93 00
82 39 13 94 69 12 78 108 17 28 17
69 22 73 38 14 17 15 15 73 04 20
68 12 13 12 51 00 54 04 91 14 13
15 86 22 96 81 66 02 82 55 70 02
00 22 83 29 08 22 12 12 04 71 13
65 27 94 19 29 14 22 08 02 11 83
73 03 26 19 07 86 86

and the second one becomes

Group 617
61 78 02 21 85 06 13 69 06
79 12 15 24 07 06 16 17 69
95 00 17 24 05 14 24 09 87
22 67 89 74 10 82 10 86 78
13 24 04 16 27 73 13 15 06
93 69 112 20 84 00 00 21 03
70 31 76 49 65 23 27 67 00 07 16
12 17

Each one just barely scratches into the low hundreds (once each), and uses "00" several times, occasionally doubled. The first one uses 45 unique numbers ranging as high as 108 with the most common characters in the teens; I haven't done any frequency analysis on the second grouping yet but the teens look popular again. I just happened to start reading David Kahn's "The Codebreakers" this week, so I've got lots of places to start, but I wouldn't mind a little help with this. Holler if you think of something!