Add a Cryptography Chapter #278

Liikt · 2018-07-19T08:12:31Z

I want to add a chapter on cryptography.

This is a general overview of what's to come. In the future i want to also write a sub chapter on most of the algorithms and concepts I mentioned here with the occasional code example.

june128 · 2018-07-19T13:51:49Z

I resolved the merge conflict.

june128 · 2018-07-19T20:26:18Z

I think the headings "General principles of Cryptography", "Symmetric Cryptography", "Asymmetric Cryptography" should be ## and not ###.

leios · 2018-07-19T20:54:35Z

Thanks for the chapter submission! It looks like there is a good amount of text here to review, so give me some time there. As a quick question: will you be adding an initial algorithm with this too?

Also: the redirect is probably not necessary in this case.

EDIT: Would you be able to separate all sentences into new lines? This will allow me to review individual sentences instead of entire paragraphs and will make the process easier on my end. It also makes it easier with version control (git). Sorry for the inconvenience.

…hive into cryptography

Liikt · 2018-07-20T05:03:35Z

Yeah I would gladly add an algorithm as well. I wanted to do a sub chapter for every "group" of algorithm. So one for older ones (i.e. caeser or substitution cipher), one for DES/AES one for blowfish (maybe i'll look into twofish as well) and one for RSA

EDIT: Also sorry for the merge commit I pulled from the wrong remote and though I pulled everything

leios · 2018-07-20T05:04:44Z

Ooh, nice. To be clear: those will be additional PR's on top of this one? I can hopefully get to review this on Sunday my time, but if you want to add to the PR, I'll wait.

Liikt · 2018-07-20T05:07:20Z

Depends on what you prefer i think :D If it's to much work to review several PRs I can add everything here. If not I think just for organizational purposes I would do a PR for every sub chapter.

leios · 2018-07-20T05:08:30Z

It's the same amount of work to review one PR or multiple ones. I think you are right that multiple PR's are better for organization

leios

Sorry for taking so long with this. I was working on this review off and on for the past few weeks.

It's a pretty good start, but kinda clunky at times. I added things that I thought would make it an easier read, but it;s sometimes personal opinion.

I think it is missing a lot of information now (mathematical underpinnings, talk of future directions, etc), but I think that information will come with the algorithms as they come in.

leios · 2018-08-01T23:01:54Z

contents/cryptography/cryptography.md

+# Cryptography
+
+For a long time humans wanted to send secret messages that only the reciever understands. 
+The first encryption algorithms go back to the ancient greeks and romans. 


A bit clunky. Maybe something like:

Humans have almost always been interested in sending secret messages that only the sender and receiver understand.
The reason for this is obvious: secret messages should remain secret.
The easiest way for this to happen is to talk behind closed doors, but that simply doesn't work if the the sender and receiver are separated by a significant distance.
In this case, they need to rely on a messenger or mailman to send the message.

For simplicity, let's assume they are sending a written letter for the purpose of negotiating war tactics in ancient Greece or Rome.
Obviously, the message can remain secret if both the sender and receiver also trust the messenger; however, what if the messenger is actually an evil spy?
What if the messenger is killed and the letter is stolen?
What if (in an elaborate ruse), some third party slips into the messenger's tent in the dead-of-night and replaces the letter with another one entirely different?

These are all important questions cryptography addresses.

The idea is simple: we procedurally scramble the message we are sending and only provide the unscrambling procedure to trusted parties.
In this way, the message would seem like utter gobbledygook to anyone other than the sender and receiver.
It doesn't matter if the messenger is evil.
They cannot read the message anyway.
It's also fine if the message is replaced, because then the receiver won't be able to properly decode the message and can just ask for another message to be sent (probably on another path with a different messenger).

leios · 2018-08-01T23:12:33Z

contents/cryptography/cryptography.md

+
+For a long time humans wanted to send secret messages that only the reciever understands. 
+The first encryption algorithms go back to the ancient greeks and romans. 
+One of the most well known encryptions is the so called "Caeser Cipher" which was supposedly used by Julius Caeser. 


Unsurprisingly, one of the first methods of encryption was supposedly developed by Julius Caeser and called the "Caesar Cipher."
Here, every character in the message is replaced by another character based on some pre-defined table or chart that only the sender and receiver have.
The table is created by simply rotating the alphabet by $$n$$ spaces, where $$n$$ is chosen in a discussion between the sender and receiver before-hand.
It is certainly not the most complicated scheme out there, but it is generally the first encryption scheme people come up with when trying to encode secret messages to one another.
Honestly, I remember sending messages back and forth to friends in elementary school, but we would never provide the necessary table to decode the message.
Instead, we would provide enough text that they could find the table themselves from context.
If a bunch of elementary school kids can figure out how to break this encryption scheme, it cannot be too robust.
In fact, it's interesting to see how the field of cryptography has grown since the Caesar cipher was developed.

The Caesar cipher obviously needs it's own chapter, so we'll link to that when it exists.

leios · 2018-08-01T23:12:59Z

contents/cryptography/cryptography.md

+For a long time humans wanted to send secret messages that only the reciever understands. 
+The first encryption algorithms go back to the ancient greeks and romans. 
+One of the most well known encryptions is the so called "Caeser Cipher" which was supposedly used by Julius Caeser. 
+In general there are two different principle of encryption, symmetric and asymmetric encryption. 


This sentence seems out of place. You mention these two methods, but don't talk about what they are until later. It' might be better to move this to the bottom.

leios · 2018-08-01T23:19:25Z

contents/cryptography/cryptography.md

+To explain the two principles I have to introduce you to two people. Enter Alice and Bob. 
+These two names are very common when reading about cryptography.
+A third and (usually) evil person is also needed and that person is called Charlie. 
+The way these three people are set up are so that Alice and Bob want to exchange messages without Charlie knowing what they said.


This paragraph is rather long. Maybe something like:

In the cryptographic literature, there is always a sender, receiver, and eavesdropper.
For some reason beyond my own comprehension, these three people are almost always given the names Alice (sender), Bob (receiver), and Charlie (attacker or eavesdropper).
These names are consistent even with quantum cryptography, so they are here to stay.

leios · 2018-08-01T23:28:19Z

contents/cryptography/cryptography.md

+These two names are very common when reading about cryptography.
+A third and (usually) evil person is also needed and that person is called Charlie. 
+The way these three people are set up are so that Alice and Bob want to exchange messages without Charlie knowing what they said.
+


We need to connect to the next section. I think it flows better to talk about the different types of encryption before talking about the general principles... Simply because the the general principles section also provides advice.

I would maybe add the sentence from above here:

In general, there are two different types of encryption: symmetric and asymmetric.
Both of which are described in the following sections.

leios · 2018-08-01T23:58:29Z

contents/cryptography/cryptography.md

+For that Alice encrypts the message with her private key. 
+This is *NOT* so that Charlie can't read the message, since both Bob and Charlie have Alices private key.
+What both of them don't have is the private key. 
+So Charlie can't decrypt, alter and encrypt the message again without Bob noticing it, because it wouldn't decrypt anymore with Alices public key. 


For example, if Alice wants to send a message to Bob and Bob wants to make sure the message from Alice was not altered, Alice can encrypt the message with her private key.
If the message is altered (possibly by Charlie), then the message can no longer be decrypted with Alice's public key.

leios · 2018-08-01T23:58:38Z

contents/cryptography/cryptography.md

+Asymmetric Cryptography is sometimes called "Public key cryptography" because Bob and Alice both need a public and a private key of which they only share the public key. 
+This makes these algorithms asymmetric because what is encrypted with the public key can only be decrypted with the private key and vice versa. 
+This has numerous fields of application. It is not only used for encrypting messages, but also for digital signing.
+Digital signing is used to make sure that the recieved message actually originates from the person who claims to have written it and you can also make sure nothing got altered. 


This can probably be removed?

leios · 2018-08-02T00:00:00Z

contents/cryptography/cryptography.md

+So Charlie can't decrypt, alter and encrypt the message again without Bob noticing it, because it wouldn't decrypt anymore with Alices public key. 
+Some examples for public key cryptography:
+* RSA. This algorithm calculates a public and a private key from two very large primes. It is (hopefully) near impossible to factor the product of two such primes in a feasable amount of time.
+* ECC or Elliptic-curve cryptography. Here you calulate the private and public key from two points on an elliptic curve. This has the positive side effect that you need smaller numbers than non-ECC algorithms like RSA to achieve the same level of security.


Both of these should be changed to the other format:

RSA: This algorithm calculates a public and a private key from two very large primes. It is (hopefully) near impossible to factor the product of two such primes in a feasable amount of time.

At the end of this section, we need to mention the same thing as above:

This section is currently a work-in-progress. These methods will also have corresponding chapters in the near future.

leios · 2018-08-02T00:06:37Z

contents/cryptography/cryptography.md

+A third and (usually) evil person is also needed and that person is called Charlie. 
+The way these three people are set up are so that Alice and Bob want to exchange messages without Charlie knowing what they said.
+
+## General principles of Cryptography


I don't know if I understand this section. It seems to be talking about implementation details for creating your own cryptographic system, but that is beyond the scope of the AAA. We care about the algorithms, but not about protecting someone from an attack.

Maybe we should say something like:

Cryptographic systems are a cornerstone to modern information technology and lie at the heart of everything from WiFi passwords to bank passwords.
If an attacker manages to crack modern cryptographic algorithms, they could cause serious damage.
For this reason, it is important to keep a few things in mind:

From here, I don't know what to add, though. I kinda feel this section should be removed, as none of the following points seem relevant.

leios · 2018-08-02T00:16:23Z

contents/cryptography/cryptography.md

+* Adding complications don't necessarily make the algorithm better or safer.
+* Always account for wrong design, implementation and usage of cryptosystems.
+A good example for a wrong implementation is WPA2 (the algorithm which encrypts WiFi traffic). This algorithm was mathematically proven to be safe yet an error in the implementaion allowed for the "Krack" attack.   
+


Somewhere in this article, we need to mention quantum computing. One of the major reasons it is such a big deal is because it has a heck of a lot of raw power that will completely blow modern cryptographic systems out of the water. This is why quantum cryptography is such a hot topic.

I don't know if it should be added here or with the asymmetric stuff.

It should probably be with the asymmetric stuff, since quantum computing only really affects that because it's good against the descrete logarithm which asymmetric crypto relies on. Symmetric crypto and hashing stays relatively untouched by it

@leios There are also several basic assumptions that modern cryptography relies on (if you multiply two large primes, it's hard to get the original primes back out) which other quantum computing techniques (carefully sculpted waveforms + observation) could invalidate, but it's unknown if they will yet. That's... probably too much detail for an introductory thing on crypto in general, though; maybe a linked chapter on quantum-resistant crypto?

I would settle for a small post quantum crypto chapter in which we explain the problems and how they are delt with. Though I agree that this would exceed the "Intro to crypto" part of the chapter

nic-hartley · 2018-10-03T21:34:51Z

contents/cryptography/cryptography.md

+It doesn't matter if the messenger is evil.
+They cannot read the message anyway.
+It's also fine if the message is replaced, because then the receiver won't be able to properly decode the message and can just ask for another message to be sent (probably on another path with a different messenger).
+Unsurprisingly, one of the first methods of encryption was supposedly developed by Julius Caeser and called the "Caesar Cipher."


People have been using ciphers since well before 44BC. Substitution ciphers, in particular, have been around almost as long as writing -- there are Mesopotamian tablets which use it. And, naturally, it was to protect someone's trade secrets. See this. That said, the Caesar cipher is a good introductory example of a cipher; I think it can and should stay, just... don't call it the first.

nic-hartley · 2019-01-03T01:41:52Z

contents/cryptography/cryptography.md

+* Adding complications don't necessarily make the algorithm better or safer.
+* Always account for wrong design, implementation and usage of cryptosystems.
+A good example for a wrong implementation is WPA2 (the algorithm which encrypts WiFi traffic). This algorithm was mathematically proven to be safe yet an error in the implementaion allowed for the "Krack" attack.   
+


@leios There are also several basic assumptions that modern cryptography relies on (if you multiply two large primes, it's hard to get the original primes back out) which other quantum computing techniques (carefully sculpted waveforms + observation) could invalidate, but it's unknown if they will yet. That's... probably too much detail for an introductory thing on crypto in general, though; maybe a linked chapter on quantum-resistant crypto?

contents/cryptography/cryptography.md

nic-hartley · 2019-01-03T01:45:34Z

contents/cryptography/cryptography.md

+In general, there are two different types of encryption: symmetric and asymmetric.
+Both of which are described in the following sections.
+
+Cryptographic systems are a cornerstone to modern information technology and lie at the heart of everything from WiFi passwords to bank passwords.


This phrasing is... odd. They're not at the heart of passwords. Maybe more like:

Encryption algorithms are a cornerstone to modern information technology and keep everything from cat pictures to bank information safe.

Though the "information technology" bit is still awkward.

well i guess you are right. I'll be taking out the password

nic-hartley · 2019-01-03T01:46:07Z

contents/cryptography/cryptography.md

+If an attacker manages to crack modern cryptographic algorithms, they could cause serious damage.
+For this reason, it is important to keep a few things in mind:
+* Because crypto has become such an advanced field cryptosystems should be analyzed by trained professionals. 
+Meaning whenever possible use a widely accepted cryptography library instead of writing your own cypher.


Indent by four spaces to continue the previous bullet point instead of dropping a random paragraph between points.

nic-hartley · 2019-01-03T01:46:46Z

contents/cryptography/cryptography.md

+Cryptographic systems are a cornerstone to modern information technology and lie at the heart of everything from WiFi passwords to bank passwords.
+If an attacker manages to crack modern cryptographic algorithms, they could cause serious damage.
+For this reason, it is important to keep a few things in mind:
+* Because crypto has become such an advanced field cryptosystems should be analyzed by trained professionals. 


Maybe make clear that both algorithms and implementations need to be vetted?

nic-hartley · 2019-01-03T01:50:15Z

contents/cryptography/cryptography.md

+Meaning whenever possible use a widely accepted cryptography library instead of writing your own cypher.
+* Kerckhoffs's principle says that when determing the robustness of a cryptosystem it should be assumed that the attacker knows the encryption and decryption algorithm. 
+This does not include any pre-shared or secret keys.
+* With the advances in technology cryptography often hits it's limits.


*its

Also, not all that often -- modern crypto typically stays well ahead of the technological curve, and modern crypto is designed with configurable strength. See, for example, bcrypt, which has a configurable number of rounds, to scale the work done without changing the algorithm. That'll probably stay relevant for decades, unless someone finds a fundamental flaw in the algorithm.

I meant here that we always had to scale the algorithms and are looking for better and better algorithms (i.e. ec instead of rsa, md5 being utter garbo for crypto,...)

@Liikt Ahh gotcha. Maybe rephrase it a little, then? Something like:

Cryptography is always improving as old algorithms get weaker and technology gets more powerful.

Either way works, though, I just misunderstood it the first time.

nic-hartley · 2019-01-03T01:50:41Z

contents/cryptography/cryptography.md

+* **Rot13**: This is a special case of the Caeser Cipher where the alphabet is rotated by 13, hence the name "Rot13."
+* **Permutation Cipher**: Here you choose a permutation $$\pi$$ (i.e. $$\pi=(3,1,2,4)$$) and reorder the the letters according to that $$\pi$$ which is the key.
+* **XOR encryption**: This method works on bitstrings and combines the message with a second message of equal length with a XOR operator.
+To decrypt, simply XOR again with the same second message.


"second message" should probably be "key", for consistency.

nic-hartley · 2019-01-03T01:51:27Z

contents/cryptography/cryptography.md

+Some examples for symmetric cryptography are:
+* **Ceasar Cipher**: Alice and Bob rotate the alphabet by $$n$$ characters and use that as a table to encode and decode their message.
+* **Rot13**: This is a special case of the Caeser Cipher where the alphabet is rotated by 13, hence the name "Rot13."
+* **Permutation Cipher**: Here you choose a permutation $$\pi$$ (i.e. $$\pi=(3,1,2,4)$$) and reorder the the letters according to that $$\pi$$ which is the key.


Pi may well be the variable typically used for the permutation, but for clarity's sake, we should probably call it p or something less likely to be confused for 3.14159...

i was going with what i learned at uni and since the AAA is a pretty scientific collection in general i'd say pi is fine here

nic-hartley · 2019-01-03T01:54:08Z

contents/cryptography/cryptography.md

+This makes these algorithms asymmetric because what is encrypted with the public key can only be decrypted with the private key and vice versa. 
+This can be used for a number of different applications, like digital signing, encrypted communication or secretly sharing keys.
+For example, if Alice wants to send a message to Bob and Bob wants to make sure the message from Alice was not altered, Alice can encrypt the message with her private key.
+If the message is altered (possibly by Charlie), then the message can no longer be decrypted with Alice's public key.


Also add a sentence about how, without Alice's private key, Eve can't make a message which could be decrypted with Alice's public key.

Amaras · 2021-10-23T11:16:45Z

@leios How do you feel about reviewing this chapter submission again?
It has not seen activity since 2019, so it is in dire need of attention (yeah, I'm going through the backlog)

leios · 2021-11-07T17:23:54Z

@Liikt similar to #346, I am not making chapters a priority, would you mind either:

Helping with this PR
Allowing me to work on this PR either by pushing to it or creating a separate PR entirely

If I don't get a response, I will assume 2

Liikt · 2021-11-07T17:29:13Z

@Liikt similar to #346, I am not making chapters a priority, would you mind either:

Helping with this PR

Allowing me to work on this PR either by pushing to it or creating a separate PR entirely

If I don't get a response, I will assume 2

I don't mind helping. But you can also feel free to push into PR as you see fit.

leios

The text here is pretty good. I made a few suggestions (feel free to ignore some of them). Also: it would be nice to have at least one image or something.

contents/cryptography/cryptography.md

leios · 2021-11-07T18:08:20Z

contents/cryptography/cryptography.md

+    Like with XOR or DES you generate a bit string (depending on which AES you use 128/192 or 256 bit long) which is your key.
+* **Blowfish**: This algorithm also was a good contender for the AES but lost to Rijndael.
+
+This section is currently a work-in-progress, and all of these methods will have corresponding chapters in the near future.


Would it be possible to also add some citations here? just add {{ "thing" | cite}} and

### Bibliography {% references %} {% endreferences %}

at the end. Be sure to add the citations to literature.bib

contents/cryptography/cryptography.md

leios · 2021-11-07T18:12:20Z

contents/cryptography/cryptography.md

+    It is (hopefully) near impossible to factor the product of two such primes in a feasible amount of time.
+* **ECC or Elliptic-curve cryptography**: Here you calculate the private and public key from two points on an elliptic curve. 
+    This has the positive side effect that you need smaller numbers than non-ECC algorithms like RSA to achieve the same level of security.
+


Like before, a few references would be nice

leios · 2021-12-28T16:41:38Z

There are a few small changes here, would you mind if I just committed the changes myself to merge this?

leios · 2022-01-31T14:16:04Z

This was finalized in #988

Julian Beier added 2 commits July 19, 2018 10:08

added a cryptography chapter

3cf4501

added the crypto chapter to the book

e0ae690

Liikt changed the title ~~Cryptography~~ Add a Cryptography Chapter Jul 19, 2018

Butt4cak3 assigned leios Jul 19, 2018

Merge branch 'master' into cryptography

8399bf6

june128 added the Chapter This provides a new chapter. (md files are edited) label Jul 19, 2018

Julian Beier added 3 commits July 20, 2018 06:57

every sentece has it's own line now

9769cfc

turned the l3 header into l2 header

66357a2

Merge branch 'cryptography' of https://github.com/Liikt/algorithm-arc…

4db1c45

…hive into cryptography

Julian added 2 commits July 21, 2018 16:21

update my pr branch

481887a

another update

c956d1e

leios requested changes Aug 2, 2018

View reviewed changes

Julian added 2 commits September 18, 2018 01:46

another update

8d132c4

add the proposed changes

c3d0f2e

leios mentioned this pull request Sep 21, 2018

Chapter submissions and planned revision #388

Closed

Julian added 3 commits October 7, 2018 03:27

update the branch once again

1647c7d

update contributors

e154598

add licence to the chapter

347a55f

nic-hartley suggested changes Jan 3, 2019

View reviewed changes

Julian added 4 commits January 3, 2019 11:50

add nic-hartley's advice

43a1fc4

fix spelling mistakes

fa2258d

changed awkward wording and the example

2d7ba69

add a small explenation to stop confusion

063285b

Merge branch 'master' into cryptography

f71d478

leios requested changes Nov 7, 2021

View reviewed changes

table; crypto -> cryptography; license

efd013d

leios added 2 commits January 15, 2022 14:35

Apply suggestions from code review

b0b10e8

Merge branch 'main' into cryptography

6d678a7

leios mentioned this pull request Jan 29, 2022

Small revisions to crypto chapter #988

Merged

leios closed this Jan 31, 2022

Add a Cryptography Chapter #278

Add a Cryptography Chapter #278

Conversation

Liikt commented Jul 19, 2018

june128 commented Jul 19, 2018

june128 commented Jul 19, 2018 • edited Loading

leios commented Jul 19, 2018 • edited Loading

Liikt commented Jul 20, 2018 • edited Loading

leios commented Jul 20, 2018

Liikt commented Jul 20, 2018

leios commented Jul 20, 2018

leios left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Amaras commented Oct 23, 2021

leios commented Nov 7, 2021

Liikt commented Nov 7, 2021

leios left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

leios commented Dec 28, 2021

leios commented Jan 31, 2022

june128 commented Jul 19, 2018 •

edited

Loading

leios commented Jul 19, 2018 •

edited

Loading

Liikt commented Jul 20, 2018 •

edited

Loading