NavList:

The early historical Delta-T investigations and publications by Stephenson and Morrison (e.g. 1984) were already a quite reliable starting point tool and did include a "best guess reasonable forecast" formula into future centuries from now.

Based on a more comprehensive set of historical astronomical records, the subsequent huge and impressive work led and published by Fred Espenak did confirm the general trends earlier established by Stephenson and Morrison. It also quite significantly narrowed down the Delta-T uncertainties for the past millennia and centuries until the 20th century. It did also incorporate an initial forecast formula into future which unfortunately started to rapidly "diverge" from current measures over the past 2 decades (almost a 3 second unexpected "drift" over the 1995-2022 time frame). Hence (and among a few others) the Espenak's/Meeus' adjusted "variant" mentioned here.

To recap : what seems the best reliable Delta-T "reconstruction tool for the past centuries and millennia" is probably the one published by Fred Espenak while the Stephenson's and Morrison's forecast formula into future centuries should probably be preferred to any other formula for the time being.

And for current years (from 1600 AD until to-day) we now have everything at hand.

For all these reasons, any versatile and performing Astronomical Software and even Celnav Software covering a few centuries (or more) validity time-span should leave the User with his own choice for Delta-T, e.g. software computed and prompted for User's acceptance to be replaced at will by any other value at his own and sole discretion. This recommendation does also hold even for Celnav applications : we all know too well that Lady Moon keeps running quite fast !

To my own view point, such User's option (for own choice of Delta-T value) is quite reasonably unescapable in any such software.

And I fully agree with you Frank ! Things here (e.g. in Stellarium to this very specific regard) can easily become an infuriating mess !

Kermit