I totally agree that first attempts may be "doing it wrong", and peer review is needed to strengthen the security through the cleansing fire of criticism. Theoretically, a collaboration of many developers where some, many, or all use pseudonyms, could be workable. (For instance, it looks like the I2P git repo is itself located (hidden?) on the I2P network.)

My thought was indeed to establish verifiable pseudonymity through the usual public key cryptography tools.