I have designed one of such card decks and released it under an open source license at DKC - Dining Kid-Cryptographers [0]. This was done as part of an EU Hackathon at Google, Belgium.

[0] Dining Kid-Cryptographers, https://github.com/secYOUre/DKC

Hi. Could you add a section that explains what this is about? The linked repo does not explain what the objective of the game is supposed to be.

I understand that the game is related to "the dining cryptographers", but I haven't memorized that piece. I read it once? I can probably find it on Wikipedia?

I recommend making your game a self-contained work that does not rely on at least one of the players remembering some details about a very specific thought experiment.

Hi! Thanks for your feedback, this is appreciated.

While there is no background information about the game in the DKC repository at the moment, the game comes with a set of cards where the equipment, rules, and finishing criteria are described [1]. If there is interested for that, I will add more context on the game and its contribution towards understanding secure multi-party computations for the youngest. I may still have the slide deck I used to present this work at the end of the hackathon, which may provide some further introductory material.

[1] DKC's rules https://github.com/secYOUre/DKC/tree/master/self-print/300-d...

What does "cut the deck" mean here, precisely? I feel like this word is fuzzy & can mean several things.

You divide the cards stack in two and put the top part at the bottom of the other part (which was the bottom part and will be now the top). The relative order of the cards is maintained: you don’t shuffle them. But if you cut them in this way several times, it obfuscates the original location of the cards (as it’s in practice impossible to follow their position changes) so you can’t tell who put which card where. Therefore the final arrangement, when revealed, is anonymized.

Okay gotcha, what was confusing me was how we could randomize it and preserve the relationships of the cards at the same time, but I see now this shuffle is more like "secure by magic trick" than "secure by cryptography", that makes sense. Or equivalently, we put it into a black box, that flips a secure coin, and either does or doesn't swap the cards. Thank you.

No, it is not magic. It really is cryptography.

Yeah sorry, my original wording was confusing, I've tried to clarify that the _shuffle_ is the magic trick, the overall cryptosystem is real.

The only thing that makes sense is keeping the middle red card in place and swapping the top two for the bottom two repeatedly? Anything else could change the answers?

I dont get the exercise

No, the order of the cards is preserved. The "cutting" of the deck is just a rotation.

I personally prefer the red ball - green ball version, where a colourblind person can find out if both balls are the same color or not. Simpler, with fewer variables and still very clean in its explanation.

Can you send a link to the red ball - green ball version?

Google gives https://www.senzilla.io/blog/2018/01/14/two-balls-and-the-co... as the first result

I don't really understand this. Who is hiding what from whom? Is the idea that the outside observers can discover that either Alice or Bob said no, but not know which one?

Why do that with five cards? Why can Alice and Bob not just hand a single card to you, saying "Yes" or "No". You then mix the cards and reveal them. Now you know if there's a "No" but not who said it.

Just using single cards will let bystanders know when both Alice and Bob say "No". The 5 card solution makes it impossible to differentiate between when both say "No" and when only one says "No".

While the cryptographic part is easily understandable, what are the practical uses of this scheme? It can only verify half of the bits on average.

You could use this to allow two people that both need to say "yes" without anyone else knowing who said "no" if the two yes doesn't occur.

However, it does leak knowledge to Alice and Bob; if either said yes they will know what the other chose. This is probably unavoidable.

If you can verify half of a set of bits, sometimes you can use an error-correcting code to verify a full set of bits. Or sometimes to verify a property of those bits.

Think of RAID storage. If some drives can't be read, but not too many, you can still read all the data.