An excerpt from Secret Key Cryptography by Frank Rubin

This article covers:

       Basic terms used in cryptography

       What is an unbreakable cipher?

       What are the different types of cryptography?

Read it if you’re interested in cryptography.

Take 25% off Secret Key Cryptography by entering fccrubin into the discount code box at checkout at

Cryptography is often called “The Art of Secret Writing.”  It is more than that.  It encompasses everything from invisible inks to transmitting messages by quantum entanglement of photons.  In particular, cryptography includes the making and breaking of codes and ciphers.

Different authors use cryptographic terminology in inconsistent ways, so let us begin by agreeing on some basic terms.

Plaintext or cleartext is the message or document that you wish to keep secret.  In traditional cryptography the message would be text written in some language known to both sender and receiver.  In a computer setting, this could be any type of file such as a PDF (text), JPG (image), MP3 (audio) or AVI (multimedia).

A cipher is a method, or algorithm, for garbling a message to make it unreadable, for example by changing the order of the characters or by replacing some characters by different characters.  In general, ciphers operate on individual characters or groups of characters in the text without regard for their meaning.

A key is a secret piece of information known only to the sender and the legitimate receiver(s) that selects which transformation is used for each message.  For example, if the cipher (method) is to change the order of the letters in a message, the key might specify which order to use for that day’s messages.  A key can be a letter, a word or phrase, a number, or a sequence of letters, words and numbers.  The strength of a cipher is highly dependent on the total size of the keys it uses.

A keyword or keyphrase is a word or phrase used as a key.

Encryption or encipherment is the process of changing the plaintext into an unreadable garble by the legitimate sender who knows the key.

Ciphertext is the resulting garbled unreadable message or document which will be transmitted or stored.

Decryption or decipherment is the process that the legitimate receiver, who knows the method and the key, uses to turn the garbled ciphertext back into the original plaintext message.

A code is also a method for garbling a message to make it unreadable.  By contrast to a cipher, a code normally operates on words or phrases in a message.  A typical code would replace words or phrases with groups of digits or letters.  (Confusingly, the word code is also used to mean a standardized representation for letters, such as Morse code.  Hopefully the meaning will be clear from the context.)

Cryptology is the formal study of cryptography, the mathematics and methodologies used for constructing and solving ciphers.  Scholars study cryptology; code-breakers study cryptanalysis.

Cryptanalysis is the study of codes and ciphers for the specific purpose of identifying weaknesses and finding ways to break them or, conversely, ways to strengthen them.

Code-breaking is the process of solving encrypted messages by third parties (enemies or opponents) who do not have the key, and may not even know the method.  This can be done by mathematical methods, or by the patient amassing and collating of intercepts, but in practice often comes down to the three B’s:  bribery, blackmail, and break-ins.

Unbreakable Ciphers

Now that we have some common language let me address the main issue.  What exactly do I mean by “unbreakable?”  First, I mean that a cipher cannot be broken by cryptographic means.  This excludes break-ins, bribery, coercion, defections, extortion, honeytraps, and similar means.  Those lie outside our scope.  Second, I mean that the cipher cannot be broken in a practical sense.  Any opponent has finite resources and finite time to devote to the code-breaking task.  When choosing a cipher, you need to have some idea of how much manpower and computer power your potential opponent(s) can expend on breaking your cipher.  Make a conservative guess, allow for improvements in computers, add a margin of safety, and pick a number.  Then, when you choose a cipher, you have a target to aim for.  Reach that target and your cipher is effectively unbreakable.

Remember that many messages have a limited lifespan.  If your message is ATTACK AT DAWN, and your enemy reads your message at noon, it’s too late.  You have already attacked.  A cipher that can be broken in 12 hours is effectively unbreakable when your opponent does not have 12 hours.

Just to make this concept doubly clear, when I say that a cipher has been broken, I mean that an opponent can read messages sent using that cipher.  Even if the opponent can read only 1% or .01% of the messages, the cipher is broken.  But there is a cutoff somewhere.  If the opponent can read a message only if they have intercepted many messages of the same length enciphered with the same key, or when the key has 63 bits out of 64 being zero, then the cipher is still unbroken.  The opponent has no a priori way of telling which messages used which key, or which keys are nearly all-zero.  It may never happen that you send two messages with the same length and same key, or where 63 key bits out of 64 are zero.

If your cipher uses a 256-bit key, and an enemy cryptanalyst finds a mathematical or computational method to reduce this to 200 bits or even 150 bits, that cipher may be weakened, but it is still unbroken if your chosen level of security is 128 bits.  Using a 256-bit key to achieve a security level of 128 bits provides a huge margin of safety.

When the government decided that the old Data Encryption Standard was no longer safe, it held an international competition for a new cipher.  Proposals were solicited worldwide.  Dozens of ciphers were submitted.  Hundreds of cryptographers evaluated these candidate ciphers for security and speed.  There were three rounds of winnowing lasting from 1997 to April 2000 until a winner was chosen.  That’s what you need to do when your cipher is going to be a worldwide standard for governments, banking, industry and the military.  If you decide to enter the next competition, this book will help prepare you.

Most readers, however, will not attempt that.  Their ciphers will have more limited scope.  They may trust their own judgment, or whatever verification process they devise, for evaluating their ciphers.

 Types of Cryptography

There are many different types of cryptography.  Some types used in the past were:

  • Hidden message, for example the messenger could swallow the message, or hide it in his boot heel or saddle, or simply memorize it.  It was common in ancient times to have a messenger memorize a message in a language he did not understand.
  • Secret method, for example the Caesar Cipher where each letter of the alphabet is replaced by the letter 3 places later.  That is, A becomes D, B becomes E, C becomes F, and so forth.
  • Disguised message, where the message is made to look like something else such as a design in the messenger’s garments.
  • Invisible message, such as microdots, or invisible inks which become visible when heated or exposed to acid.
  • Misdirection, for example where the signature or the shape and color of the paper are the true message, and everything else is distraction or disinformation.

Collectively, all methods for hiding a message are called steganography, first described in the 1499 book Steganographia by Benedictine abbot Johannes Trithemius, born Johannes Heidenberg.  Trithemius’s book is itself a form of steganography, since it is disguised as a book of magic.

Some of these steganographic methods have modern counterparts.  For example, a message can be concealed in a JPEG image file by using only the low-order bits of each pixel.  Another example is to use a random number generator to pick certain bits in each byte of a file.  The chosen bits contain the message, and the remaining bits can be random gibberish.

Before describing modern ciphers, let me introduce a useful shorthand.  A message is sent from a sender to a receiver, and the purpose of encryption is to keep some enemy from reading the message.  For brevity I call the sender “Sandra,” the intended receiver “Riva,” and the enemy “Emily.”  Now that is more natural than Alice, Bob and Carol, isn’t it?

Usually Sandra enciphers the message at her location before sending it to Riva.  The message may be sent by any means, letter, telephone, Internet, shortwave radio, Aldis lamp, microburst, telegraph, fiberoptic cable, semaphore, quantum entanglement, or even smoke signals if there is a direct line of sight.  To make this picture more complete, the cipher may require a key as well as the plaintext, and there may an enemy listening in.  Here is a fuller image.

Modern ciphers generally fall into 3 broad categories, Secret Key, Public Key and Personal Key. Their main distinguishing features are:

Secret Key:  Sandra has a secret key which she uses to encipher messages.  Riva has a corresponding secret key, which she uses to decipher those messages.  This may be the same key or an inverse key.  Usually Sandra controls the key.  When Sandra changes the key, she must send the new key, or its inverse, to Riva.  This is the standard paradigm of classical cryptography.

Public Key:  Riva has a public encryption key which she makes known to everyone.  Whenever Sandra wants to send Riva a message, she enciphers it using Riva’s public key.  Riva also has a secret decryption key, known only to herself, which she can use to decipher the messages that she receives.  To make this scheme work, it is essential that nobody else can compute this secret key from the public information.  The dominant Public Key method is the RSA algorithm invented by Ronald Rivest, Adi Shamir and Len Adelman about 1975.

Personal Key:  Sandra and Riva each have a personal key which they share with nobody, not even one another.  Since no keys are ever transmitted or shared, Personal Key cryptography is sometimes called keyless cryptography.  Here is how it works:  (Pass 1) Sandra enciphers the message with her personal key and sends the enciphered message to Riva.  (Pass 2) Riva enciphers that message with her personal key and sends this doubly-enciphered message back to Sandra.  (Pass 3) Sandra deciphers that message using her personal key and sends this back to Riva.  The message is now enciphered only with Riva’s key, which she uses to read the message.

The tricky part here is that Sandra’s encryption and Riva’s encryption need to commute.  That is, you must get the same result regardless of whether Sandra encrypts first or Riva encrypts first.  Symbolically, we express this as SRM=RSM, where M is the message and S and R are Sandra and Riva’s encryptions.  The advantage of Personal Key cryptography is that anybody can communicate securely with anybody else without having to prearrange any keys or transmit any keys, so there is no possibility of a key being intercepted.

Personal Key cryptography is also called the Three Pass Protocol. A protocol is just a sequence of steps used for some purpose such as transmitting a message.  In other words, a protocol is an algorithm.  The basic idea for the Three Pass Protocol was invented by Adi Shamir about 1975.

Symmetric vs Asymmetric Cryptography

Many books state that cryptography can be divided into two types, symmetric and asymmetric ciphers.  The idea is that in Secret Key cryptography Sandra and Riva use the same key to both encrypt and decrypt the message, while in Public Key cryptography Sandra uses one key while Riva uses its inverse.  This dichotomy overlooks Personal Key cryptography, which is neither symmetric nor asymmetric, as well as the various classical methods.  Moreover, the symmetric/asymmetric classification is not always accurate.  The Hill cipher, for example, is a Secret Key method where encryption consists of multiplying the message by the key, and decryption consists of multiplying by an inverse key – just like Public Key cryptography.

Categorizing a cipher as either symmetric or asymmetric is not particularly useful.  It fails to capture the essential difference between Secret Key and Public Key cryptography, namely that in Secret Key cryptography all of the keys are kept secret, while in Public Key cryptography each party keeps one key secret key and makes one public and available to everyone.

Public Key cryptography and Personal Key cryptography both came out around 1975.  Public Key cryptography fired the imagination, so Secret Key and Personal Key methods have received scant attention since that time.  Public Key cryptography is fully covered in many books.  This book focuses primarily on Secret Key cryptography, the mainstay and bedrock of cryptography.

 Block Ciphers vs Stream Ciphers

Another classification is to divide ciphers into block ciphers and stream ciphers.  Block ciphers operate on blocks of characters in the message, say blocks of 5 characters.  Usually all of the blocks are the same size, and the same key is used for every block.

Stream ciphers operate on one character of the message at a time.  Each character has its own key, called the character key, typically taken from a larger key called the message key.  In older stream ciphers the message key was repeated.  For example, if the message key size is 10 characters, then the first key character would be used to encipher message characters 1, 11, 21, 31, … of the message, the second key character would encipher message characters 2, 12, 22, 32, … and so forth.  Ciphers using regularly repeating keys are called periodic.  In newer stream ciphers the message key is usually as long as the message itself, and is called the key stream.  This aperiodic, or non-periodic, style of enciphering is called the one-time pad.

The block/stream classification is not exclusive.  There are hybrid ciphers where the message is broken up into blocks, but different blocks are enciphered with different keys, so the cipher operates on a stream of blocks rather than a stream of characters.

Mechanical vs Digital

Ciphers can also be classified according to the means used to produce them.  The earliest ciphers were done entirely by hand.  Not pencil and paper, but rather stylus and parchment, or stylus and clay tablet.

The first mechanical means of encipherment was the skytale or scytale (pronounced SKIT-a-lee) used by the ancient Greeks and Spartans, probably as early as 700BCE.  This consisted of a rod with a narrow strip of leather or parchment wound around it so the edges of each turn carefully match the edges of the adjacent turns.  In other words, there would be no gaps and no overlaps.  The letters of the message were written across two or more turns of the strip.  When the strip was unwound only disconnected pieces of the letters were visible so that an enemy would not recognize that it contained a message.  Additional squiggles or patches of color could be added to make it look like a decoration.

The sender keeps the staff for reading and writing future messages.  Messengers could wear the strip as a belt, use it to tie up their hair, or to cinch their saddles.  The recipient would need a rod of identical diameter to reconstitute the message.  Messengers, of course, would not be told the purpose of the ribbon or thong.  It might even be sewn into their garments without their knowledge.

Here is an image of a skytale from the 1593 edition of De Occultis Literarum Notis by Giovanni Battista Porta.  Notice how each Greek letter spans several turns of the leather strip.

The Greeks kept the secret of the skytale for about 700 years.  The Romans, however, were not so successful.  Eventually their enemies in northern Europe learned the meaning and use of these rods.  So, the Romans invented a special measuring tool consisting of a hollow brass or bronze dodecahedron, a solid shape with 12 identical pentagonal faces, with a circular hole on each face.  These holes allowed them to make wooden rods of precisely the correct diameter.  When a governor, ambassador, or spy was sent to a post that required travel through hostile territory, it was safer to carry this tool than to carry an actual skytale that could be captured.  The 12 holes had different diameters to allow for secure communications with other governors, ambassadors and spies, for example small for Londinium (now London), medium for Lugdunum (now Lyon), and large for Tarraco (now Tarragona in Catalonia).

So far as is known, the purpose of these dodecahedrons was never discovered by the northern Europeans, nor, for that matter, by modern archeologists.  Archeologists have proposed a plethora of preposterous purposes for these artifacts, such as children’s toys, saddle ornaments, practice pieces for blacksmiths, candleholders, range-finders for artillery, or the answer of last resort, religious objects.

This is a bronze Roman dodecahedron found near Tongeren, the oldest town in Belgium, and displayed in the Gallo-Romeins Museum.

Here is an interesting side note:  If you look up skytale in Wikipedia and other websites, it says that the skytale was used to produce a transposition cipher by writing each letter within one turn of the strip.  This is incorrect.  Such a strip would easily be recognized as a cipher message.  Whether or not the enemy could read the message, they would certainly not let the messenger deliver it.  A thorough examination of the whole letter vs broken letter issue can be found at  In 1841, Edgar Allan Poe, who was a talented cryptographer, wrote an essay, A Few Words on Secret Writing, which gives a good description of the skytale and his method for decrypting these messages.

To compound this error, if you look up “transposition cipher” in Wikipedia, it says that the skytale was used to produce a “rail fence cipher,” also called a zigzag cipher.  A rail fence cipher has columns which alternate up and down.  Writing a message either along or around a rod does not involve any changes of direction.  So, if a skytale were used to produce a transposition cipher, the result could be a columnar transposition, never a rail fence.

I tried to correct all these errors in Wikipedia, but gave up.

A 1960s version of the skytale was to sort a deck of computer punch cards, write the message in pencil on an outer surface of the deck, then thoroughly shuffle the deck.  When the deck was run through a card sorting machine, the cards would be restored to the same order, and the message could be read.  This idea was widely discussed by programmers, but I do not know if it was ever put into practice.  Another modern equivalent is to write the message on the blank back of a jigsaw puzzle, then scramble the pieces.  The receiver needs to solve the puzzle, then flip it over to read the message.

Another mechanical cipher was the Jefferson Wheel Cypher invented by Thomas Jefferson sometime between 1790 and 1793.  It consisted of 36 same-sized wooden disks threaded on an iron rod to form a wooden cylinder.  Around the exposed edge of each disk the 26 letters of the alphabet were written in some scrambled order.  The disks could be rotated independently to spell any message.  Versions of the Jefferson cipher using disks or paper strips were used as recently as the 1960s.

Many types of disk ciphers were developed from the 15th through the 19th centuries.  The most common type used several thin flat concentric disks that could be rotated around a center pivot.  Each disk would have the alphabet, or some set of numbers or symbols written in some order around the rim of its upper face.  The disks would get progressively smaller so that all of the alphabets could be seen at the same time.  The disks would be aligned in some position, and encipherment would consist of finding the plaintext letter on one of the disks, then using the corresponding letter or symbol on one of the other disks as the ciphertext letter.  Later types of disk ciphers would advance the inner disk after each letter was enciphered, either manually or through a clockwork mechanism.

This is a picture drawn by Augusto Buonafalce of the Leon Battista Alberti cipher disk from his 1467 book, De compendis cifri.  Image distributed by WikiMedia Commons.

Starting in 1915 a long series of electromechanical rotor ciphers were invented.  The most famous is the Enigma machine developed in the 1920s by German engineer Arthur Scherbius.  Dozens of types were marketed up through the start of the computer era.  They all produced stream ciphers.  The basic idea was that the substitute for a letter was determined by the path that the electric current took through the series of turning rotors.  After each letter was enciphered some of the rotors would turn, controlled by various cams, gears, lugs and pawls to change the substitutions in myriad ways.  So, if the word INFANTRY came out as PMRNQGFW, that might not happen again for billions of turns.

Since the 1960s, cryptography has become increasingly computerized and digital.  The Data Encryption Standard (DES) was developed by IBM in 1975 and certified by the National Bureau of Standards in 1977.  This touched off a series of block ciphers with names like Serpent and TwoFish, culminating in the Advanced Encryption Standard (AES) adopted by the National Institute of Standards and Technology (NIST) in 2001.  This class of ciphers is covered in chapter 11.

The progression has been from manual mechanical electromechanical digital.

 Why Choose Secret Key?

In this era of Public Key cryptography the question naturally arises, why would anyone choose Secret Key cryptography?  There are several reasons.

Secret Key cryptography is much faster.  Even the strongest, most complex Secret Key methods tend to be hundreds or even thousands of times faster than the leading Public Key methods.  In fact, the main use of Public Key cryptography is to encrypt keys for Secret Key cryptography.  The keys are sent using Public Key methods, but the messages themselves are sent using Secret Key methods.

Public Key Cryptography, PKC, requires a Public Key Infrastructure.  There must be Public Key servers to distribute public keys to potential correspondents.  Public Key Cryptography is subject to a variety of man-in-the-middle and spoofing attacks, where an opponent poses as the sender, and/or the receiver, and/or the key server, so PKC requires a great deal of authentication and verification.  The person requesting a public key must prove membership in the same network as the recipient.  The message containing the public key must be verified to assure it came from the server.  The receiver must be authenticated when the public key is first posted on the server, and every time it is changed.  When a new party is added to the network, the person authorizing that new party must be authenticated.  When a new network is added to a server, every party involved must be authenticated.  The receiver must verify that received message has not be altered or replaced by some third party.  This all makes for a profusion of messages.

Secret Key cryptography can operate without any of that administrative overload.  Two individual people could exchange secret key messages without involving anyone else, or any intermediating system.  When several people are exchanging secret key messages, the only authorization required is that each party has the current key.  An unauthorized person won’t have the keys and can’t read the messages.

Exchanging messages is not the only use for cryptography.  An equally important role is securing the secrecy of data files stored on a computer, on an external device such as a flash drive, or in cloud storage, often for a long time.  Public Key cryptography cannot be used for this purpose.  Only Secret Key methods are suited for keeping data files secret.

When a message needs to be broadcast to many receivers at the same time, this can easily be done with Secret Key methods.  All that is needed is for each party to have the key.  They could use a special broadcast key separate from their personal keys.  Or, each party could be sent the message key by using a separate key-transmission key.  With Public Key methods you would need to get the individual public keys of all of the recipients, with all of the attendant authorization and verification.  This cannot be prearranged because the participants are free to change their public keys at any time.

The most common public key method is the RSA method.  The strength of this method depends on the fact that it is currently very difficult to factor large numbers.  Given a 200-digit decimal number with no small prime factors, there is currently no feasible way of factoring it.  However, when quantum computers become available, this all changes.  MIT professor Peter Shor developed a quantum algorithm which can easily factor a number that size.  When that happens, all of the RSA messages stored on computers will be able to be read.

There is, so far, no known way to use quantum computers for cracking secret key ciphers.  If quantum computers are a concern, Secret Key cryptography is the only choice.

Why Build Your Own?

If you are a cipher hobbyist, it is obvious why you want to build your own ciphers.  You build your own ciphers because that’s your hobby.  Model train hobbyists design, build and run model trains.  Model plane fans design, build and fly model planes.  Cipher hobbyists design, build and solve ciphers.

If you are a cryptography student, building your own cipher is good training.  It is the best way to learn how to build and how to evaluate ciphers.  The current standard cipher, AES will not last forever, and somebody will need to design its replacement.  If you want to be part of that effort, this book may be your best starting point.

If you are a serious cryptographer with responsibilities for protecting data and communications, you might build your own ciphers out of a healthy skepticism that the government-approved ciphers are as safe and secure as your government claims.

Let me give you one story that will support your doubts.

Circa 1975 IBM proposed the cipher now called DES, the Data Encryption Standard.  It would become a worldwide standard for secret key encryption.  As IBM originally designed it, DES had a 64bit key.  The National Security Agency (NSA) required that the key be reduced from 64 bits down to 56 bits, with the other 8 bits used as a checksum.

This made no sense.  If a checksum were really needed, then the key could be increased from 64 to 72 bits.  It was widely believed that the real reason the NSA made this demand was that they knew how to crack messages using a 56-bit key, but not messages using a 64-bit key.  This proved to be true.

You could reasonably conclude that the NSA, National Security Agency, would never approve any encryption standard which they could not crack.  In that case, you could infer that the NSA can crack all of the different forms of AES, the Advanced Encryption Standard.  And, if the NSA can crack AES, then it is likely that its Russian and Chinese counterparts can crack AES as well.

There are just a handful of experts who construct the candidate ciphers from which the worldwide standard ciphers are selected.  It is well-known that these experts receive briefings at NSA headquarters in Fort Meade, Maryland.  During these meetings NSA personnel advise them of techniques that which might either strengthen or weaken the ciphers.  It is possible that hidden among the recommended methods is some backdoor that lets the NSA, and only the NSA, solve these ciphers easily.  It is also plausible that the NSA could offer jobs, contracts and research grants that might induce the experts to adopt those methods.

There is a good bit of speculation here, but cryptographers tend to be very conservative.  If you can imagine a plausible weakness or vulnerability, whether or not your opponent can realistically exploit it, it is best to guard against it whenever you can.

Finally, you might just be after speed, simpler implementation, or cheaper hardware.  You might want to construct your own ciphers to achieve these goals without giving up security.  You will find methods in this book that can help you do that.

That said, remember that there are plenty of pitfalls.  Don’t just create a cipher and assume it is “strong enough.”  Lots of ciphers turn out to have unexpected weaknesses.  Even the strongest cipher can be defeated by operator errors, such as starting every message with a standard header, frequently reusing keys, or sending the identical message using different keys.  For example, many German messages were solved during WW II because they all began the same: “Heil Hitler, …”

This book contains all the information you need to construct an unbreakable cipher, but remember that reading only one book about cryptography will not make you an expert overnight.

That’s all for this article. Thanks for reading.