The
current status of thermodynamic modeling in aqueous chemistry
is reviewed. A number of recent developments hold considerable promise,
but these need to be weighed against ongoing difficulties with existing
theoretical modeling frameworks. Some key issues are identified and
discussed. These include long-standing difficulties in choosing the
right program code, in comparing alternatives objectively, in implementing
models as published, and in wasting effort on numerous proposed “modifications”
and/or “improvements”. There needs to be greater awareness
of the major limitations that such assorted variations in modeling
functions imply for practical thermodynamic modeling purposes. They
typically lack proper substantiation, fail to distinguish between
cause and effect, and are presented in ways that all-too-often cannot
be falsified. Numerical correlations in particular permit overoptimistic
assertions based only on “satisfactory” fits, neglecting
the dictum that regression analyses can be used to discredit hypotheses
but not to prove them. The risks of “model tuning” should
always be acknowledged and minimized. Recognition of the uncertainties
in data that have not been confirmed by independent measurement needs
to be redoubled. Modeling frameworks incapable of explicit trace activity
coefficient prediction should no longer be regarded as credible. The
current modeling paradigm evidently has to be reassessed, hopefully
to find better ways forward, including new protocols which command
sufficient support to warrant IUPAC endorsement.