Historic Authentication Schemes Before Computers

Question

The Bulletin of the Atomic Scientists recently ran an article about an alleged near miss with rockets in 1962, which raises some interesting cryptographic questions:

After the usual time-check and weather update came the usual string of code. Normally the first portion of the string did not match the numbers the crew had. But on this occasion, the alphanumeric code matched, signaling that a special instruction was to follow.

Bordne’s crew was instantly alarmed and, indeed, the second part, for the first time ever, also matched.

At this point, the launch officer of Bordne's crew, Capt. William Bassett, had clearance, to open his pouch. If the code in the pouch matched the third part of the code that had been radioed, the captain was instructed to open an envelope in the pouch that contained targeting information and launch keys.

To me it is doubtful that this event ever happened, but it made me wonder how messages were authenticated in pre-computer days during the time of mechanical calculators and long before the advent of public key cryptography.

At this time, the standard was a codebook + superencryption and electromechanical rotor machines. In this case they probably used a code book. Neither scheme provides integrity or authentication.

CBC-MAC as one of the easiest MACs is (barely) feasible to do by hand and doesn't fit the description 'the code in the pouch matched the third part of the message'

Another issue with all of these schemes is that every potential recipient can trivially originate a fraudulent message himself. There is nothing to keep a rogue crew from opening the secret pouch, constructing a message and sending it out to all other sites. At that point, all of them will open their pouches, so the seal is useless to find the guilty after the fact.

How were message authentication and integrity handled during the time of codebooks and one time pads? Were there any schemes that provided any non-repudiation capability?

Edit: I found an article that claims:

When I served as a Minuteman launch officer in the 1970s, I and another crew member could have reverse engineered this entire business to create and transmit to the entire U.S. land- and sea-based strategic force a fully valid and authentic order that would have been carried out by crews whose only validation requirement was to match their SAS codes with the order’s SAS codes. (We didn’t use unlock codes at the time, even though Defense Secretary McNamara thought otherwise). The source of the message was irrelevant; all crews were trained to carry out any launch message that validated regardless of the source.

But, while somewhat morbidly fascinating, I don't want to dwell solely on the issue of nuclear authentication. There is a huge body of literature about historical military encryption, but only very little about authentication and integrity. This is strange, because messages were transmitted in error prone Morse code via HF links. I guess, the famous 'THE WORLD WONDERS' message during the battle of Leyte Gulf shows that at least the Navy did not verify message integrity.

Answer 1

As explained by Cort Ammon, they mostly relied on physical, rather than mathematical processes.

Here is a video on Youtube titled "How to launch a nuclear missile", taken in the Titan Missile Museum mentioned by Cort Ammon.

The message contains a seven character string. The personnel in the shelter have a set of envelopes with the first two letters on the outside. If the inside of the envelope matches the message, they launch.

Here is another one from a British sub, starting at 23:00. Decryption is done behind a curtain.

It is all very manual, no computer is visible. Note the wooden club used to protect the codes (24:17).

Answer 2

In the past, they relied far more on human techniques than cryptographic ones.

The Trident example you linked is interesting to me, because it is conflicting with my experience receiving a tour in the Titan Missile Museum in Arizona. I won't claim my tour qualifies as a reliable source, but it's what I was given.

In the tour, they explained the process with the SAS codes, just like in the Trident example. However, there was one final unlock code for the Titan missile which was never housed in the silo. It was a special crypto lock on the oxidizer lines going down to the nozzle below. Unless it was unlocked, the oxidizer would not flow, and the missile was going nowhere. When the launch authorizations were sent, the president would also send the codes to unlock the oxidizer lines. Thus, you could use your SAS cookie to forge a launch message, but you could not give anyone the correct code to actually fly the missiles. That number was in the president's hands alone (or in the hands of their aides, if the page you linked is to be believed).

In theory the oxidizer manifold had the magic key they needed to forge a nuclear attack that would actually do something. However, accessing this would require far more effort than merely getting people to agree to unlock a safe and access the SAS cookie. It would require active maintenance on the missile in a known sensitive area. There would most likely be additional security procedures to make it difficult for anyone to get into the crypto lock without tripping either a tamper resistent layer or getting caught by one of the observers.

There may have been a reason why the people entrusted with the Titan missiles were selected for their trustworthyness and support of America!