#crunchlabs

SVG artwork for laser engraving/cutting plexiplas replacement tiles for Mark Rober's/CrunchLabs Build Box #6

I've been following along with CrunchLabs (site at www.crunchlabs.com) and I'm woefully behind, but did pick up this build out of order Because Cryptography! It's a cool CrunchLabs build kit that lets you build & use a pretty snazzy gadget for cipher encryption.

If you're familiar with Criptoquip puzzles (a popular staple of newspaper puzzles back in the day), this is basically the same thing. It's not a shift cipher because Mark and the team at Crunchlabs have essentially "scrambled" the connections between the top and bottom plates:

(Hat tip to Mark for giving a quick breakdown of how the German "TRITON" 3-wheel Enigma machine worked in WWII in the build video)

The way it works is that the letter wheels actually come off of the central contraption; you can line up a particular pair of letters with the white arrows on the red and blue side, and snap the letter wheel back into place. This lets you index the whole thing to a keypair (two letters).

And thus, every different keypair combo results in a different cipher for each plaintext letter. If you know the keypair and ciphertext for an enciphered message, you can flip the whole thing over in its base so that the blue side is up and decipher the plaintext. Which is pretty darn cool. There are 676 possible settings (2626, if my math is right*), so that'd be a lot of work for someone manually trying to figure it out if they had obtained one of these machines, and if no one had monkeyed with the initial cable wiring.

*I've never been really great at math, so take anything I say going forward with a grain of salt.

But there are still some problems here. First is that if you use the same keypair to encipher an entire message, then your message is vulnerable to "criptoquip" solving techniques.

First, if your message contains duplicate letters, they'll all encipher the same way. Let me give you an example of the worst possible message to encipher: "A BITTER QUITTER LOST THE LITTLE BATTLE".

Why is this terrible? It's terrible because of all of the repetition. In English, there are certain letter patterns (doubled Ts, "-er" or "-le" at the end of the word) and certain letters are going to appear more frequently (e.g., "e" is the most common letter).

So by choosing this awful message, I've already given the codebreaker some juicy "cribs" (clues). If you're encrypting messages with your friends, don't do this. Because if you set up your cipher machine with the keypair of red=N and blue =S (for Reasons), the message will end up like this:

OYPII KUREP IIKUZ QMIIT KQPII ZKYOI IZK

(by convention, cryptographers split up ciphertext into blocks of five characters when doing it manually like this*). You'd be tempted into thinking those Is are Es, but when it doesn't work, the next most common letter is T...and it shouldn't take you much longer to crack this one.

*Tradition!

So how can we make this better? One way is to shift one or both of the wheels. I could specify, say, that I'm going to rotate the red wheel to the right by 3 many characters after I encipher a letter. Three isn't a divisor of 26 so I'm sure to get through all 26 letters with no repetitions. I'll indicate this on my message sheet by using a lower-case c (the third letter) to the right of my Red Wheel letter (Nc). Resulting ciphertext:

OVJZW VCWGO EBAHJ XQJGO COBRO CKVIZ WKS

Okay, this is better. But there's a problem: the message is more than 26 letters long, so the keypairs are reused, and worse, some of the reused keypairs encipher the same letters (e.g., B and T and T and T is a repeat). If the message > the number of possible keypairs, keypairs will be repeated. I need a way of rotating both wheels (red and blue) so that I work through all 676 possible combinations.

Obvious ways are to include a marker for the blue wheel, too. Say I send the keypairs "Nc gS". This means that the algoritm is "start with N. Encipher the first letter and then rotate the red wheel to the right by three letters. Encipher the next letter. Rotate the red wheel to the right by three letters, repeat until you get back to N. Then rotate the blue wheel LEFT by 7 letters (because the g is on the left side of the S). Encipher the next letter. Rotate the red wheel to the right by three letters, etc., until you get back to N, then rotate the blue wheel left by 7 letters again, etc.

It's complicated, but in this way, you work your way through all 676 keypairs, enciphering each letter with a different keypair. So let's do that. Ciphertext:

OVJZW VCWGO EBAHJ XQJGO COBRO CNBCO LVV

As long as my message is not > 676 characters or I don't keep reusing the initial keypair setup, I'm not going to be repeating keypairs in a periodic way.

I came up with another, degenerative method. Starting with an initial keypair (again, Nc S), encipher the first letter. Rotate the red wheel to the right by 3 letters, then set the blue wheel to the first letter's ciphertext. For example, with "N S" as the initial keypair, A becomes O, so the next keypair is K O. Encipher the next letter, the ciphertext of which ("D") becomes the next setting for the blue wheel, rotate the red wheel to the right by 3 (new keypair: H D), and repeat. Result:

ODHBJ LNMQU QAZIR SQIUB ZZKHQ DIWDG UDQ

All those Qs and Ds! The double Z! And it almost reads "SQUIB". It's just delightfully awful. I LOVE IT

"Okay," you might be asking yourself. "What if my ciphertext with the key info, and the device, fall into the hands of The Enemy*?" Is there a way to encrypt the initial key setting?

*Mom & Dad

Yes! I'd suggest a book cipher, because they're fun. It involves having two identical copies of the same book (the text needs to be exactly the same). Then you can agree on how to reference letters on the page. Example: page-line-word (and look at the initial letter of the word).

Let's use identical copies of Neal Stephenson's Cryptonomicon, because why wouldn't we?

If I write "208-14-1", I'm saying "page 208, line 14, word #1". Looking at your copy of Neal Stephenson's "Cryptonomicon" (softcover), you'll see that the word is "No" and starts with the capital letter "N" so this is your red wheel key letter.

This page is a scene in which Lawrence Waterhouse imagines a scenario involving Alan Turing so as to explain how polyalphabetic cipher (the core of how the Enigma machine works) to the reader. You may not believe it, but I did pick actually this page at random.

All of this does require that you and your partner a) have identical copies of the same book and b) have agreed on a format & scheme for conveying information, such as what you're referencing & how in your book cipher, and what the resulting info means & how to use it.

Here is the setup to my degenerative, book-cipher-enhanced encryption: 1) x-y-z for book cipher meaning "page, line, word (first letter of the word)"; first line is the red key letter, next line is the letter that indicates how many places & direction you'll rotate the red wheel.

If the next line page number is AFTER first line's page number, rotate the wheel to the RIGHT. Otherwise, to the LEFT. Since the first page number is 208 and the second is 424, we'll be rotating the red wheel to the right.

The third line is the blue key letter, and the lack of a fourth line means that the blue key letter shifts degeneratively: set the blue wheel to the cipher text of the previous operation.

Is it a lot to agree upon and remember? Sure. But even with the explanation, above, anyone reading this thread would have trouble deciphering the ciphertext unless they happen to have an identical copy of "Cryptonomicon".

Since there's a pattern involved in this (and also unconscious bias when it comes to choosing pages, lines, and words for the book cipher), it's of course vulnerable to a number of codebreaking techniques, and I'm sure that a modern cryptographer with computer tools would break it quickly.

But I think this is a pretty secure method for amateurs messing around with a couple of Cipher Encryption kits (Cryptonomicon has some mature themes, tho, so maybe better for older players). Thanks to Mark Rober and the team at CrunchLabs for putting together this really awesome build!

I have always been skeptical of deliveries by drones. I’m an engineer and know the many problems with such systems. But two recent ones have shown how they can be successful. One is domestic, one foreign – in Africa! This moderately lengthy YouTube video shows how, and in a very entertaining way at that.