In "Time-lock puzzles and timed-release crypto", Rivest et al. describe a method to encode information such that it can only be decrypted after a certain amount of time has elapsed. In practice, this is implemented by requiring the party decrypting the message to repeatedly square a value to obtain a key with which the message can be decrypted. This operation is inherently sequential and, without significant advances in prime factorization, cannot be significantly parallelized.
Run timelock.rb to get usage information. Try out encrypting and decrypting
a message like this:
echo "hello world!" | ./timelock encrypt -t 10 | ./timelock.rb decrypt
The output of timelock encrypt is YAML, with numbers stored in base 16 and
ciphertext stored in base 64
---
n: '0xd424561c0e58fc9627098068ef21c5db4647ed37e9b9f45809bb86ef7899ac11'
t: '0x3b9aca'
a: '0x7ca1d4221b246666310649c6ab3731c7d67779aca244fb5248dba5d2308df00'
ck: '0x795d9a45afe7ae9297f7cf24ec29ebde72c40c627cf43c96c70a8573e7df7843'
cm: |
tjqVnMTCi6Jfr7A+3A==