First, obtain a reasonably-sized group of geeks in mixed company. A university would be a good place to look. The geeks part is important, since you'll have to get a reasonable number of them to understand and go along with this idea.
The group should be large enough so that there are several unrealized possibilities for romance among its members, but small enough that everyone in it knows who most of the other people are.
Have everyone create a public and private key. Arrange a system of pairs of messages (the content doesn't matter)that can only be decoded if two people encrypt them with each other's public key. I'm sure this is cryptographically possible, though it might require multiple stages.
Then, people should create the appropriate messages, encrypted with the keys of everyone he or she would be willing to go out with, and send them to a central repository. Once all the encoded messages are there, each person can then attempt to decode all the messages.
If there is a match, then the two people will successfully decode each other's messages. Voila! If the feeling is only one-way, on the other hand, the other person never has to know.-- PurpleBob, Dec 23 2002 am i hot or not http://www.amihotornot.comThey are doing something like this. Would not be hard to simply use their system and a page of links to the group's profiles. [futurebird, Oct 17 2004] I don't understand the mechanism here. Can you give an example? How do you prevent me from discovering affection without requiting?-- egnor, Dec 23 2002 It sounds a little like that "E-Crush" service that I see pop ups for, only on a smaller scale... Maybe I don't understand the mechanism either(?)-- colaaddict, Dec 23 2002 [The following annotation both misunderstands and misses the point of this idea; it's safe to disregard. Thanks, egnor.]
I don't think it is the e-crush idea, but maybe it should be. If you want to let people discover if interest is mutual without risking rejection, that's a much simpler way to go about it.
Encrypting the messages seems pointless, even if there is some system for ensuring that only reciprocated ones are readable. You may not even want to let people send arbitrary messages: They might express wildly different interest levels or be innocuous or irrelevant or worse (someone might simply broadcast "I'm not really interested in you, but knowing that you lust after me might be useful someday").-- Monkfish, Dec 23 2002 E-Crush etc. rely on a trusted central authority; the idea here is to come up with a cryptographic protocol that does away with the need for such an authority. It's a perfectly fine idea, I just don't understand how the actual cryptography works. I'd also be a little surprised if the problem hadn't already been studied (perhaps in a different guise).
Yes, each round of messages would be only a single bit, conveying whatever level of interest is designated for that round.-- egnor, Dec 23 2002 <sort of related> I was once very methodically propositioned through a series of cryptogrphic messages (I knew the sender but not what the message was). Each message was progressively more difficult and progressively more . . . provocative. The key was always something common between us but increasingly obscure. It was a clever tact.-- bristolz, Dec 23 2002 This idea scares me, but it is a essentially good one. I'll give it a croissant and hope that the boys in the CS department never hear about it.-- madradish, Dec 24 2002 [bristolz] And? Did you get it on, or were you creeped out?-- egnor, Dec 24 2002 [bristolz] I thought we agreed to speak no more of that incident...And, for the record, sending you my phone number repeatedly does not qualify as a mathematical cryptogram. I'm blaming it all on the combination of egg nog and Pimm's Punch.-- jurist, Dec 25 2002 I really like the idea but can't seem to get the encrypting part into my brain...
"Arrange a system of pairs of messages" Why?-- nederhoed, Apr 08 2003 I believe it's impossible to guarantee that both parties will find out about the match unless there's third-party involvement.
What would be needed is a message which has some information readable to Alice that Alice doesn't have *and* some information readable to Bob that Bob doesn't have. Neither Alice nor Bob can ever send such a message.
If a trusted third party is available, there's no need for encryption.-- beland, Nov 02 2003 The simple fact that you want to send an anonymous encrypted message requires a trusted 3rd party. You can send anyonmous messages via remailers that strip headers and forward messages, but here we have no encryption, and a third party. You can directly send encrypted messages via Diffie-Hellman key exchange, but here we have no source anonymity.
Now, if you're ok with the idea of a trusted third party, you can perform an anonymous submission (it's fairly easy to strip headers from an email) of a document encrypted with the keys of everyone you're interested in (of course you'd have to get the keys from the trusted server). These keys would have been generated by the server using a shared secret algorithm for X participants (you could let X be the number of particpants plus a few extras to allow for newcomers without having to regen and redistribute keys). The server would have then stored the public keys for everyone, and distributed the private keys to the appropriate participants via some secure means (D-H, SSL, whatever...). By generating shared secret keys you can do some really neat things, in this case if a message was encrypted with your key, you could decrypt it, otherwise you get jibberish. So when you're sending messages to the server, you encrypt them with the keys of everyone you like, and then whenever someone attempts to decrypt it, if you used their key, then they'd read it, otherwise they'd be in the dark.
This doesn't exactly do anything to prevent unrequited messages though, I'll need to think on this for a little while more. Stay tuned for more information, same Rev time, same Rev channel....-- Reverend_Cobol, Nov 03 2003 random, halfbakery