Operation Vula
July 29, 2015 1:31 AM   Subscribe

How the ANC sent encrypted messages to one another during the struggle against apartheid. Talking to Vula is a series of six articles by Tim Jenkins about the project from the ANC`s monthly journal Mayibuye from May 1995 to October 1995. (via Schneier)

“Even in those days, 25 years before Edward Snowden, there was talk about ‘backdoors’ in encryption software,” said Jenkin.

The Vula Connection is a documentary on Operation Vula.
posted by jeffburdges (13 comments total) 13 users marked this as a favorite
 
As always with any SA website: Do not read the comments.
posted by PenDevil at 2:26 AM on July 29, 2015 [3 favorites]


Operation Vula is... interesting. It has had a major impacts on modern day SA.

After negotiations with the National Party led SA government began in the late 80's many of the top ANC leaders, Thabo Mbeki included, were unaware of Vula's existence. It had also transformed into an operation smuggling weapons into SA in case negotiations broke down for good and the armed resistance was to restart.

The leaders of Vula were the ANC Intelligence operatives based out of KwaZulu-Natal, among it's members were Jacob Zuma.

When the National Party found out about Vula, Mbeki was livid and embarrassed he was left out the loop. It opened a major rift between Mbeki and his faction of "intellectuals-in-exile" and the KZN based intelligence operatives that continued to fester. That all culminated in the arms deal for new naval vessels and planes in the late 90's that seemingly benefited many of the leaders of the Vula network. Again Mbeki was embarassed by Zuma and his attempts to rid himself of Zuma only led to his early ouster as President and Zuma's ascendancy.
posted by PenDevil at 3:21 AM on July 29, 2015


Which would not be the first time that the intelligence/counter-intelligence operatives take over a revolutionary movement after the revolution has come to pass. It’s almost as if it’s a rule or something.
posted by pharm at 3:43 AM on July 29, 2015 [1 favorite]


Maybe I'm missing something, but this seems weird:
The next step was to make it even harder for potential eavesdroppers to bug our phone lines. Already there were three regular lines coming into my flat and two cellular phone links. The latter operated through a marine antenna on top of the roof to give the very best reception for computer communications. In Holland I acquired a cordless phone unit with a range of 15 kilometres. A sympathetic Briton a few kilometres away had a phone line installed and to this I connected the main unit of the cordless phone. At home the handset was connected to another antenna on the roof. This radio phone provided a secure line that worked perfectly with our computers.
So instead of GCHQ bugging his phone, anyone with a radio in line-of-sight could now bug it? And it's not like it would be hard to find out which line the handset was connected to, either.
posted by Joe in Australia at 4:13 AM on July 29, 2015 [2 favorites]


it's strange that no-one mentions this on schneier's site, but if i understand correctly then the BASIC code way down in that article was the encryption they used later. and to my interested-but-not-expert eye, it's incredibly bad. it looks like it's a linear congruential PRNG.
posted by andrewcooke at 4:16 AM on July 29, 2015 [4 favorites]


>Maybe I'm missing something, but this seems weird:

You're not just whistling Bell 212A. None of that makes a lot of sense. Marine band antennas are on a totally different frequency to cellular systems of the time, and wouldn't work. As for the 'long range phone from Hong Kong'; they certainly existed and did work. But they were also totally illegal in the UK and used low-band VHF channels that were thoroughly illegal for public use in the UK, allocated to other users, and did cause a number of problems. Running a (relatively) high power transmitter at those frequencies from a fixed point in London would absolutely get you noticed in short order by fhe radio interference people, who would quickly work out it where you were. And because it would be clear from the outset that it was encrypted datacomms (not by any means common back then, and certainly not running out of domestic premises) it would be passed on to the spooks. Probably all within a week of the link being established for the first time.

The spooks would probably know all about it anyway, because all international telephone circuits were routinely monitored. Encrypted datacomms there would also be a high priority to investigate.

There is absolutely no way that British intelligence wouldn't have been completely aware of what was going on here. More than that - were they complicit? Did they have ways to read the messages? - who can say. Hopefully, someone. But not I. (At the time, I was living in London and had a pretty good idea of what was going on using a £300 scanner; I certainly knew of some users of those long-range cordless phones. Discreet, they were not.)

(I also notice some HF and VHF ham equipment in those photographs, and mentions of using them to talk to others in London. If you ever want a large audience of interested third parties, do it on the ham bands...)
posted by Devonian at 6:46 AM on July 29, 2015 [3 favorites]


(I should say 'had a pretty good idea of what was going on on the airwaves', not of underground terrorist radiocomms!)
posted by Devonian at 7:36 AM on July 29, 2015 [1 favorite]


andrewcooke: "to my interested-but-not-expert eye, it's incredibly bad. it looks like it's a linear congruential PRNG"
     SUB EncVerFD(MSG$,SNUM$,SALF()) 
        LOCAL BM,CP 
        LENMSG=LEN(MSG$) 
        $EVENT OFF 
        FOR ENC=1 TO LENMSG 
            RL=ASC(MID$(SNUM$,CP+1,1)) 
            CH=SALF(ASC(MID$(MSG$,ENC,1))) 
            CD=(RL XOR CH) MOD 128 
            MOUT$=CHR$(CD+32) 
            IF (ENC+2) MOD 10=0 THEN MOUT$=MOUT$+CHR$(160+BM):INCR BM 
            PUT$ #1,MOUT$ 
            CP=ENC MOD 3000:BM=BM MOD 15 
        NEXT ENC 
        $EVENT ON 
        EM$=STRING$(5,175) 
        PUT$ #1,EM$ 
    END SUB
IANAC (I Am Not a Cryptographer, or a BASIC programmer), but I don't think it is doing any random number generation in this function, just applying the one-time-pad scheme, using the NUM argument as a one-time pad. The FOR loop is looping over each character in MSG: on each iteration, it increments ENC, and XORs the ENC'th character of the message with the (CP+1)'th character of NUM. CP is then set to (ENC mod 3000).

There are a few things that do look a bit weird:
* if a message is longer than 3000 characters, they resort to reusing the first part of the one-time pad
* the if-statement means that the cipher-text for every 10th character has an extra value added to it. This seems like someone just trying to do something non-standard in the hope of obtaining some security through obscurity, but I don't seem how it would actually help in any meaningful way: if an attacker doesn't have the one-time pad, they can't decrypt the message at all; if they have the pad but not the code, obscuring 10% of the characters probably won't make the messages much harder to interpret.
* I think SALF is a function they're written, rather than a builtin PowerBASIC function, and I have no idea what it does
posted by James Scott-Brown at 9:48 AM on July 29, 2015 [3 favorites]


Fantastic. Thank you.
posted by wuwei at 11:19 AM on July 29, 2015


ah, ok, cool. i really just looked at the CP=ENC MOD 3000:BM=BM MOD 15 part. but 3000 doesn't make sense anyway. sorry.

i wonder how they generated the pads?
posted by andrewcooke at 11:39 AM on July 29, 2015


If they reused the pads after 3000 chars they were very, very stupid. This is the big flaw in One Time Pads: if you have a "crib" (a piece of known text) and a reused pad, then you can often decrypt the entire message.

E.g: Suppose you know that the pads they use are 3,000 chars long, and wrap around after that. Your intercepted encoded message is 4,000 chars, so there's 1,000 chars of overlap. You guess that a message will likely contain the word "this" (or any other common word). You take the last 1,000 chars and the first 1,000 chars, and moving one character at a time alternately assume that four characters from the first or last 1,000 chars decrypts to the crib. Use that decryption for the other four characters - is it a "sensible" combination of characters (e.g., "ight" or "expl" or some other part of a word or words) or is it nonsense like "xjuq"? If it's sensible then start guessing the surrounding words, and build up your decryption by applying them to the characters from the other part of the message. In this case you'd be able to decode half of the encrypted message, just from the knowledge that they reused part of their pad.
posted by Joe in Australia at 6:23 PM on July 29, 2015 [2 favorites]


While the technical side of the history seems a little flawed, it's pleasant to know that international secret messages really were being sent via home computer and some electronics knowledge likely picked up from Bernard Babani books (= the UK equivalent publisher of hobbyist electronics books, like Radio Shack was for Forrest Mims).
posted by scruss at 9:07 AM on July 30, 2015


I suppose it's unfair for me to criticise their LEET CRYPTO SKILLZ: I think the revelations about British WW2 cryptography mostly came after this, and of course there's all the modern cryptography stuff you read about in relation to the Internet. None the less, their techniques are so blatant that you really have to think MI5 knew what they were up to. Perhaps it was British policy to not interfere?
posted by Joe in Australia at 7:08 PM on July 30, 2015


« Older "We are the megadead."   |   Fitted Newer »


This thread has been archived and is closed to new comments