I don't see the benefits of these. Just because I can manually encrypt and decrypt I don't have more confidence in the cipher. If I already trust the cipher I might as well trust a program that does the encryption and decryption.
https://en.wikipedia.org/wiki/CipherSaber
The core goal of which was the idea that people should have basic access to some sort of reasonably strong cryptography as a matter of personal liberty.
The theory goes that if it's an easy enough algorithm to do manually if you had, that makes it an easy enough algorithm to reimplement as needed in a situation where you have to forge your own tools if need be (all you have is a deck of cards, pencil, and paper, or all you have a TI-83 graphing calculator and its BASIC programming manual; can you securely send a message?).
1. Key generation using tiles as proposed is truly random, not susceptible to attacks on a PRNG.
2. The key is never connected to any network in any way--it's effectively airgapped. This removes whole classes of key-stealing and side channel attacks.
That said, while real randomness is still uncommon, there are electronic solutions, and the benefits of networking generally outweigh the security risk for most use cases.