You are giving bad advice. bcrypt's maximum password length is not 32 bytes. In fact the maximum has been debated due to discrepancies from the paper describi

You are giving bad advice. bcrypt's maximum password length is not 32 bytes. In fact the maximum has been debated due to discrepancies from the paper describing it and the first implementation, which actually accepted longer passwords due to how it handled words internally. In either case the max length is at least 20 characters more than you state. Your own words above: "because encrypt/decrypt takes up" is 32 bytes. By most measures a five word passphrase fitting into 32 bytes doesn't even contain 80 bits of entropy.