How long would it take an average computer to figure out the WWII Enigma machine

[CJay]

The bombe was not a put in the message and wait for the answer type of thing - there were hundreds of talented code breakers at Bletchley Park responsible for obtaining a daily key and the bombe was only a part of that process.

I think someone tried a distributed computer attack in 2008 to break some of the remaining unencrypted German messages from WWII, but it did not do well at it.

https://www.cnet.com/news/distributed-computing-cracks-enigma-code/

[Simon]

The whole thing was relays at best and basically it was a guessing game. The codes were brute forced although this was after obtaining clues. For example there was a German station that every morning would transmit "nothing to report" once they guessed the message contents they could use it to try and find out the code. Some of the peculiarities of the coding system also gave away clues. I think the bombs were simply motor drive enigma replicas with multiple replicas in each bomb so that they could simultaneously run lots of possible options on the days code until one of them spat out an intelligible message (in German). So there were no actual code cracking machines, they were just a very large number of duplicates of one machine built into one machine and then they had several machines. The cracking was done by humans. It was not necessarily the breaking that was the problem but the speed required because every 24 hours the code was changed and different German forces used different codes so ideally on a good day by 1AM they had cracked it and knew the orders before those executing them did. on a bad day they did not crack the code until 23:00 so not much use as in 1 hour it would be changed and they would have to start again. So it was simply parallel execution of the same task on the same data until a result came out. Whittle it down with human calculation and intuition and then try out variants of the very good guesses.

[Beamin]

The bombe was not a put in the message and wait for the answer type of thing - there were hundreds of talented code breakers at Bletchley Park responsible for obtaining a daily key and the bombe was only a part of that process.

I think someone tried a distributed computer attack in 2008 to break some of the remaining unencrypted German messages from WWII, but it did not do well at it.

https://www.cnet.com/news/distributed-computing-cracks-enigma-code/

Radio signal 1851/19/252: "F T 1132/19 contents: Forced to submerge during attack. Depth charges. Last enemy position 0830h AJ 9863, (course]) 220 degrees, (speed) 8 knots. (I am) following (the enemy). (Barometer) falls 14 mb, (wind) nor-nor-east, (force) 4, visibility 10 (nautical miles)."

Quick someone get the navy on the phone and tell them this!!!! I'm surprised they put such sensitive info on the internet.


We still havent figured out if you could use an arduino to crack the code or how long it would take a PC, or how its done for that matter; that youtube video didnt explain how they did it.
 If I wasn't such a beginner it would be cool to build a arduno based enigma machine. I miss living in a house with a workshop/garage/tools. I would build it from an old type writter.

[ledtester]

I found this paper to be a good survey of the various techniques involved in a ciphertext-only attack on Enigma:

https://cryptocellar.org/pubs/Enigma_ModernBreaking.pdf

Also check out the top-level of the website.

This github repo looks promising with regard to the hill climbing algorithm:

https://github.com/arvindpj007/Enigma-M4-Cryptanalysis

There are a lot of references to a 1995 paper by James Gillogly which introduced the "Index of Coincidence" statistic. I was able to find a copy here:

http://web.archive.org/web/20060720040135/http://members.fortunecity.com/jpeschel/gillog1.htm

[Simon]

[cbutlera]

Following a visit to Bletchley Park in March 2015, I wrote a simulator for the bombe and also for the (originally manual) process of extracting the remaining steckerboard configuration.  The input data was a crib, the matching enciphered text and a bombe menu.  The simulator worked through every permutation of and starting positions for the rotors until it identified a valid candidate.  During this process, the bombe stops at multiple candidate positions, but most of these can be eliminated because a contradiction is reached while attempting to extract the remaining steckerboard configutation.

All of the information that I used to write the simulation came from Gordon Welchman's book "The Hut Six Story".  It was a command line, text only simulation, written in PowerBasic.  It was not meant to be in any way fast, efficient or elegant, the only purpose of writing it was to cement my own understanding of the process.  In particular, I tried to simulate the manner in which the bombe operated.  The most processing intensive part was to simulate what the mechanical bombe did easily, which is electrical propagation from the initial active circuit throughout the entire bombe, which typically involves many hundreds of passages through the bombe drums.

My simulator for a 12 Enigma-equivalent bombe (3 Rotor Enigma), running on a single thread of a 3.5GHz i5 ran at the equivalent of 27,000 RPM, compared to the 120 RPM of the later mechanical bombes.

[ledtester]

This video:

Enigma Cryptanalysis Revisited – New State-of-the-Art Attacks on the Enigma Machine in CrypTool 2
https://youtu.be/kptQbcTUVyk

can give you a rough idea of the state of the art.

The program "CrypTool 2" has several algorithms for decrypting Enigma, one known-plaintext attack (similar to the Bombe attack) and three ciphertext-only attacks:

- Gillogly's attack
- Lasry's hill-climbing attack
- Lasry's simulated annealing attack

All of these have their own trade-offs.