I'd just like to point out that, regardless of the potential for further reducing the uncertainties in instrumental records of SST (pre-satellite era), much excellent work has been done recently to more fully understand and quantify those uncertainties.

For example, HadSST4 provides "a set of interchangeable realisations that capture the temporal and spatial characteristics of the estimated uncertainties in the biases [and] the measurement and sampling uncertainties," which can be combined into a comprehensive uncertainty estimate.

That said, surely there are more surprises to be uncovered in the raw data (like national-level offsets, studied by Duo Chan and Peter Huybers!), and if we can understand their impacts on the reconstructed fields, there will still be some measure of correction to enjoy.