HFQ Official Snippet #6 (oopsie!)

Yeah, it's why I find the 'Nynian totally has a secret way that Merlin can't decode!' meme annoys me. People way underestimate encryption, and at the same time underestimate it.

The best way to keep a secret is for no-one to realise there is a secret. Given the volume of traffic over the semaphore, I just don't think it is possible to subject all the messages to the kind of highly-skilled time consuming analysis many are implying. If Nynian is using it, she will be inserting her own messages into highly routine messages being sent by people who have been regular and trusted users of the semaphore for years. They will avoid anything that looks at all out of the ordinary. She will use a variety of routes so that none of them is used terribly often.

I haven't read 'the codebreakers' but I suspect many of these instances refer to messages already known to be encoded or written by or to people already under suspicion.

Randomiser wrote:The best way to keep a secret is for no-one to realise there is a secret. Given the volume of traffic over the semaphore, I just don't think it is possible to subject all the messages to the kind of highly-skilled time consuming analysis many are implying. If Nynian is using it, she will be inserting her own messages into highly routine messages being sent by people who have been regular and trusted users of the semaphore for years. They will avoid anything that looks at all out of the ordinary. She will use a variety of routes so that none of them is used terribly often.

I haven't read 'the codebreakers' but I suspect many of these instances refer to messages already known to be encoded or written by or to people already under suspicion.

Forgive me, but you suspect wrong. For example transAtlantic mail during WW2 was (1) subject to very strict limitations and (2) dropped off in Bermuda and perused on a piece-by-piece basis. Most pieces of course received very cursory examination. But records were kept that could be linked as necessary over time. And for a channel to be useful it must carry a goodly amount of traffic.

The manpower to do this over commercial semaphore chains would be trivial in comparison. Remember it's not that any one msg can get through unnoticed. Yes that's reasonably possible. But establishing and using a channel to control a large organization without attracting any notice whatever over time? A MUCH harder problem. Basically impossible even with the technology of that time and place.

Even traffic analysis begins to show suspicious patterns over time. Why does company A reliably take 25% more bandwidth than company B to conduct its business? Once noticed, investigators will go deeper and deeper.

jgnfld wrote:

Randomiser wrote:The best way to keep a secret is for no-one to realise there is a secret. Given the volume of traffic over the semaphore, I just don't think it is possible to subject all the messages to the kind of highly-skilled time consuming analysis many are implying. If Nynian is using it, she will be inserting her own messages into highly routine messages being sent by people who have been regular and trusted users of the semaphore for years. They will avoid anything that looks at all out of the ordinary. She will use a variety of routes so that none of them is used terribly often.

I haven't read 'the codebreakers' but I suspect many of these instances refer to messages already known to be encoded or written by or to people already under suspicion.

Forgive me, but you suspect wrong. For example transAtlantic mail during WW2 was (1) subject to very strict limitations and (2) dropped off in Bermuda and perused on a piece-by-piece basis. Most pieces of course received very cursory examination. But records were kept that could be linked as necessary over time. And for a channel to be useful it must carry a goodly amount of traffic.

The manpower to do this over commercial semaphore chains would be trivial in comparison. Remember it's not that any one msg can get through unnoticed. Yes that's reasonably possible. But establishing and using a channel to control a large organization without attracting any notice whatever over time? A MUCH harder problem. Basically impossible even with the technology of that time and place.

Even traffic analysis begins to show suspicious patterns over time. Why does company A reliably take 25% more bandwidth than company B to conduct its business? Once noticed, investigators will go deeper and deeper.

It's quite easy to send messages that can't be decrypted, even today. It's called one-time pads or one-time use keywords. The semaphore network on Safehold is used quite extensively by businesses and it's not that difficult to bury codes in the traffic. For someone to pick up on for example misplaced words, financial data where a comma is in the wrong place or inventory lists that's listed in a specific order is very very remote. It quickly adds up to pretty big amounts of combinations that can carry a lot of information with few symbols.

What you tend to forget is that before the war SSK didn't have a need for high-speed communications, it seems they mostly collated information on what the church was up to or managing their financial interests when they weren't recruiting rebellious girls. My guess is that most of this information was transferred by vetted persons and still is.

In retrospect on what we now know about SSK it seems that they probably have prepared contingency-plans that probably go into action by just mentioning prearranged combinations of names, words or places. That's more or less impossible to pick up on.

For example, if someone in SSK places a legitimate sale order it can mean to operatives for the SSK that the Mother Abbess has to report to Zion ASAP. It's kinda hard to find that out by just seeing a sale order.

jgnfld wrote:

Randomiser wrote:...
The kind of code to use on this system is not the kind which scrambles every word but one which uses indicators or keywords to pass the Message. when they sent Earl Coris the wyverns from Safehold they were accompanied by a message containing references to the Writ. The chapter and verse numbers were pointers to the page and word numbers, in a previously agreed novel, of the words of the actual message. That kind of code is unbreakable without the 'decoding pad' and is not readily detectable if used wisely.

In our real world these codes were tried a lot. Friedman's wife (Freidman led the effort to break Purple) broke such book codes by hand in the rum running/anarchist era of the 20s and 30s only identifying the book after the fact! And again, simply sending such thing brings attention to yourself as the first measure of the authorities is to limit users to one or a few code books.

These codes are only breakable if the messages are both long or frequent enough and can be identified as such.

the thing is that if you send routinely perfectly genuine financials messages between 2 entities and only send crypted ones once in a while, the odds of interception are extremely low. Of course, ideally the information that a message is crypted is not in the message itself.

Once a message has been identified, deciphering it if you dont knw the code book is still very hard without profound knowledge of letters/words frequencies (and a word based map space is very large). Considering the state of maths in Safehold, it is unlikely the church know how to do it.

Once deciphered, you still ideally a message with only non sensical pre arranged codes.

This is equivalent to a one time pad, so no repetition patterns to work with either

Joat42 wrote:...

It's quite easy to send messages that can't be decrypted, even today. It's called one-time pads or one-time use keywords. The semaphore network on Safehold is used quite extensively by businesses and it's not that difficult to bury codes in the traffic. For someone to pick up on for example misplaced words, financial data where a comma is in the wrong place or inventory lists that's listed in a specific order is very very remote. It quickly adds up to pretty big amounts of combinations that can carry a lot of information with few symbols.

What you tend to forget is that before the war SSK didn't have a need for high-speed communications, it seems they mostly collated information on what the church was up to or managing their financial interests when they weren't recruiting rebellious girls. My guess is that most of this information was transferred by vetted persons and still is.

In retrospect on what we now know about SSK it seems that they probably have prepared contingency-plans that probably go into action by just mentioning prearranged combinations of names, words or places. That's more or less impossible to pick up on.

For example, if someone in SSK places a legitimate sale order it can mean to operatives for the SSK that the Mother Abbess has to report to Zion ASAP. It's kinda hard to find that out by just seeing a sale order.

It is impossible to use one-time pads over the semaphore as the first elementary precaution against such the controlling authority would take (and always has on Earth) is to outlaw coded messages that are not in one of a very few codebooks on file with the central authority. This means that unless the Inquisition can read your traffic they simply will not send it. One-time pads are theoretically unbreakable, yes, but there is a reason they are not in general use except in the highest security areas: key distribution is a terrible logistical nightmare and susceptible to a number of attacks even though the cipher system properly used (an oft broken assumption)is not. As well, spy agencies tend to get mighty curious when they find books of random numbers in your possession. Not a problem for the central office but definitely a problem for the message sender in the field. For example, it is my advice that you not carry around megabytes or a gigabyte of random numbers on your laptop if you cross borders much. That would be a true mistake.

Writing "pre-arranged combinations of names, words and places" without making the language stilted every single time is similarly difficult. Names, words, places, and numbers have to mean something. If the analyst checks and finds they do not perfectly make sense, then that implies something else is going on. An analyst will even get suspicious if he should notice that the same place keeps ordering silver candlesticks or some such. And so on. And it is very difficult to memorize an entire, useful codebook that contains enough variation to encompass everything that might be said. Therefore your signs method requires carrying a copy of all the possible signals. Again, spy agencies tend to get mighty curious when they find such materials in your possession. And the techniques to break such word codes were around in late medieval times when every diplomat had a "nomenclator" in his possession precisely for the purpose you envision. (See Kahn.)

Every person who comes up with a cipher system typically ignores the immense practical problems that actually using a coded communications channel entails. Even today, the NSA tends to break implementations of ciphers through all sorts of man-in-the-middle attacks, device attacks, and social engineering attacks not the core ciphers themselves. (It's usually easier to find a way to get the password some way than to brute force a solution.)

Kahn's book is probably overly long but is a book many are unable to put down once started (me--but then as a stats guy who knows some people who went to work for, uh, various DC agencies, that should not be surprising). It will take you from ancient times up to WW2. There is an abridged paperback version that is almost as good. For more modern material you should consult Bruce Schneier's writings. He also has a blog where he discusses such issues a lot, but you need as least some basis in theory probably to make complete sense of what he is saying.

The whole history of codes and codebreaking is that codebreakers typically find ways in through some combination of theoretical, technical, or human means.

With respect to semaphores, the quotation from

http://infodump.thefifthimperium.com/en ... hold/229/1

reading

" At the moment, it hasn't really occurred to any of the Group of Four that their semaphore traffic might be compromised by Charisian agents, since they are operating more or less on the assumption that any "leaks" which occur in the semaphore transmission of messages are occurring because members of the Church's own semaphore service have been bribed by -- or placed in their current positions by -- members of the Church's own hierarchy. In other words, they think that any "taps" on their message traffic are in-house, not put in place by the Church's external enemies (of which there were none of any significance, whatever some members of the Group of Four may have thought, prior to the decision to destroy Charis)."

The statement that there was a lack of external enemies prior to the attack on Charis puts an interesting perspective on the Sisterhood of Saint Khody.

The SSK could have people in the semaphore offices themselves, and be burrying messages in the overhead traffic.

Hi Jgnfld,

I'm surprised, didn't RFC have a post that mentioned one time pads being quite commonly used on the semaphore?

I believe he also stated the church transmitters could hack messages and provide them to their competition for a small fee...

L

jgnfld wrote:

Joat42 wrote:...

It's quite easy to send messages that can't be decrypted, even today. It's called one-time pads or one-time use keywords. The semaphore network on Safehold is used quite extensively by businesses and it's not that difficult to bury codes in the traffic. For someone to pick up on for example misplaced words, financial data where a comma is in the wrong place or inventory lists that's listed in a specific order is very very remote. It quickly adds up to pretty big amounts of combinations that can carry a lot of information with few symbols.

What you tend to forget is that before the war SSK didn't have a need for high-speed communications, it seems they mostly collated information on what the church was up to or managing their financial interests when they weren't recruiting rebellious girls. My guess is that most of this information was transferred by vetted persons and still is.

In retrospect on what we now know about SSK it seems that they probably have prepared contingency-plans that probably go into action by just mentioning prearranged combinations of names, words or places. That's more or less impossible to pick up on.

For example, if someone in SSK places a legitimate sale order it can mean to operatives for the SSK that the Mother Abbess has to report to Zion ASAP. It's kinda hard to find that out by just seeing a sale order.

It is impossible to use one-time pads over the semaphore as the first elementary precaution against such the controlling authority would take (and always has on Earth) is to outlaw coded messages that are not in one of a very few codebooks on file with the central authority. This means that unless the Inquisition can read your traffic they simply will not send it. One-time pads are theoretically unbreakable, yes, but there is a reason they are not in general use except in the highest security areas: key distribution is a terrible logistical nightmare and susceptible to a number of attacks even though the cipher system properly used (an oft broken assumption)is not. As well, spy agencies tend to get mighty curious when they find books of random numbers in your possession. Not a problem for the central office but definitely a problem for the message sender in the field. For example, it is my advice that you not carry around megabytes or a gigabyte of random numbers on your laptop if you cross borders much. That would be a true mistake.

Writing "pre-arranged combinations of names, words and places" without making the language stilted every single time is similarly difficult. Names, words, places, and numbers have to mean something. If the analyst checks and finds they do not perfectly make sense, then that implies something else is going on. An analyst will even get suspicious if he should notice that the same place keeps ordering silver candlesticks or some such. And so on. And it is very difficult to memorize an entire, useful codebook that contains enough variation to encompass everything that might be said. Therefore your signs method requires carrying a copy of all the possible signals. Again, spy agencies tend to get mighty curious when they find such materials in your possession. And the techniques to break such word codes were around in late medieval times when every diplomat had a "nomenclator" in his possession precisely for the purpose you envision. (See Kahn.)

Every person who comes up with a cipher system typically ignores the immense practical problems that actually using a coded communications channel entails. Even today, the NSA tends to break implementations of ciphers through all sorts of man-in-the-middle attacks, device attacks, and social engineering attacks not the core ciphers themselves. (It's usually easier to find a way to get the password some way than to brute force a solution.)

Kahn's book is probably overly long but is a book many are unable to put down once started (me--but then as a stats guy who knows some people who went to work for, uh, various DC agencies, that should not be surprising). It will take you from ancient times up to WW2. There is an abridged paperback version that is almost as good. For more modern material you should consult Bruce Schneier's writings. He also has a blog where he discusses such issues a lot, but you need as least some basis in theory probably to make complete sense of what he is saying.

The whole history of codes and codebreaking is that codebreakers typically find ways in through some combination of theoretical, technical, or human means.

jgnfld wrote:It is impossible to use one-time pads over the semaphore as the first elementary precaution against such the controlling authority would take (and always has on Earth) is to outlaw coded messages that are not in one of a very few codebooks on file with the central authority. This means that unless the Inquisition can read your traffic they simply will not send it. One-time pads are theoretically unbreakable, yes, but there is a reason they are not in general use except in the highest security areas: key distribution is a terrible logistical nightmare and susceptible to a number of attacks even though the cipher system properly used (an oft broken assumption)is not. As well, spy agencies tend to get mighty curious when they find books of random numbers in your possession. Not a problem for the central office but definitely a problem for the message sender in the field. For example, it is my advice that you not carry around megabytes or a gigabyte of random numbers on your laptop if you cross borders much. That would be a true mistake.

But this isn't Earth, the Church has had no reason for institutionalized codebreaking and therefore has no mindset for it and what to look for. Remember, before Charis the Church had no need for this since if they suspected some nation where up to something they just sent in the inquisition to deal with any problems.

jgnfld wrote:Writing "pre-arranged combinations of names, words and places" without making the language stilted every single time is similarly difficult. Names, words, places, and numbers have to mean something. If the analyst checks and finds they do not perfectly make sense, then that implies something else is going on. An analyst will even get suspicious if he should notice that the same place keeps ordering silver candlesticks or some such. And so on. And it is very difficult to memorize an entire, useful codebook that contains enough variation to encompass everything that might be said. Therefore your signs method requires carrying a copy of all the possible signals. Again, spy agencies tend to get mighty curious when they find such materials in your possession. And the techniques to break such word codes were around in late medieval times when every diplomat had a "nomenclator" in his possession precisely for the purpose you envision. (See Kahn.)

Actually, it's quite easy to use "pre-arranged combinations of names, words and places" without anyone having a clue and the messages will not look stilted. For example:
"The buyer will arrive at noon"
"The buyer will meet you at noon"
The first one means everything is ok, the second means "get the heck out!". You use that phrase once, then it's discarded. Tell me how ANYONE would understand that without having knowledge about the pre-arranged code-phrases? You are still confused by the simple fact that we don't need to send complex or huge volumes of messages and the church isn't actively looking for coded messages (yet). And we can also infer by Nynians use of codephrases to pass along simple instructions that SSK have a far better understanding of security than the church has.

jgnfld wrote:Every person who comes up with a cipher system typically ignores the immense practical problems that actually using a coded communications channel entails. Even today, the NSA tends to break implementations of ciphers through all sorts of man-in-the-middle attacks, device attacks, and social engineering attacks not the core ciphers themselves. (It's usually easier to find a way to get the password some way than to brute force a solution.)

Kahn's book is probably overly long but is a book many are unable to put down once started (me--but then as a stats guy who knows some people who went to work for, uh, various DC agencies, that should not be surprising). It will take you from ancient times up to WW2. There is an abridged paperback version that is almost as good. For more modern material you should consult Bruce Schneier's writings. He also has a blog where he discusses such issues a lot, but you need as least some basis in theory probably to make complete sense of what he is saying.

The whole history of codes and codebreaking is that codebreakers typically find ways in through some combination of theoretical, technical, or human means.

The whole history of Safehold after the WatF hasn't had the need to break codes, since the whole of Safehold was united with the Church.

What you are ignoring is that the fact that the Church doesn't have dedicated code-breakers, they don't have an organization that sifts through every message passing through the semaphore system looking for coded messages, which there are plenty of anyway since businesses on Safehold have regularly used the semaphore system to send coded messages using one-time pads. It's a quite common practice.