No, I mean your original comment is inaccurate. The paper presents a time-space lower bound for parity learning, but the encryption scheme based on this result

No, I mean your original comment is inaccurate. The paper presents a time-space lower bound for parity learning, but the encryption scheme based on this result is only 'unconditionally secure' in a model where the adversary is restricted to having at most (n^2)/25 bits of storage. This isn't a general-purpose unconditionally secure encryption scheme, which is what your original comment implied.