You are right, although we might be able to make it lossless in practical terms with probabilistic approaches like combining multiple algorithms, selecting the

You are right, although we might be able to make it lossless in practical terms with probabilistic approaches like combining multiple algorithms, selecting the hash length, and using heuristics to pick the most likely source.