Non-uniqueness of atmospheric modeling

We focus on the deceptively simple question: how can we use the emitted photons to extract meaningful information on the transition region and corona? Using examples, we conclude that the only safe way to proceed is through forward models. In this way, inherent non-uniqueness is handled by adding information through explicit physical assumptions and restrictions made in the modeling procedure. The alternative, 'inverse' approaches, including (as a restricted subset) many standard' 'spectral diagnostic techniques', rely on more subjective choices that have, as yet, no clear theoretical support. Emphasis is on the solar transition region, but necessarily discussing the corona, and with implications for more general problems concerning the use of photons to diagnose plasma conditions.