A bulk of my GIS datasets reference Washington State Plane North FIPS 4601 (US Feet). Often I get data in WGS84 that I need to transform to FIPS 4601. Ten years or so I worked though many of the default transformations in ArcGIS to prove to myself that the NAD_83_To_WGS84_5 (CONUS) transformation was in fact the best transformation for minimizing any shifting in the datasets. I do not remember the shift in the different transformation from back then but I have been relying on NAD_83_To_WGS84_5 since that time.
I am now running ArcGIS 10.7.2 and noticed a shift of 1.28 meters in data using the NAD_83_To_WGS84_5 transformation.
I have rerun my experiments using some newly created (AKA clean) point data in WGS84. I have experimented with the NAD_83_To_WGS84_5 (CONUS) and the default ITRF00_To_NAD83 default transformation in the reprojection tool and see that this adds about 1.28 meters of shift in my data.
I also ran a transformation using NAD_83_To_WGS84_1 (CONUS, Canada, Alaska) and see that my data less that 0.000 meters away.
Looking at the ESRI publication ArcGIS 10.7.1 and ArcGIS Pro 2.4 Geographic and Vertical Transformation Tables I see that there is now a newer NAD_1983 to WGS84_OR_WA_41 transformation. This transformation causes about a 0.2 meter shift in my data.
Interestingly, I duplicated this transformation experiment in QGIS 3.4 and see that QGIS returned less than a 0.000 meter shift in the data without ever being prompted for a transformation type.
My question is: Why is the more generalized NAD_83_To_WGS84_1 returning better data than the more focused transformations and is there a better transformation that I should be using?