In reading through comments on some other sites, I think there are two things going on that are causing the disconnect between Ancestry and Gedmatch genetic distances. The first is phasing, which has been an issue all along. Some of the segments that appear to be large on Gedmatch may have been reduced by Ancestry's phasing program so that they are smaller in Ancestry's matching algorithm. The second is their new program, Timber, which eliminates matches or reduces their probability based on "pileup," the assumption that a large number of matches on a segment indicates that an apparent match is based on either a widely shared genetic characteristic or a very distant ancestor well beyond genealogical time.

There are two results:

1) increasing genetic distance on Ancestry compared to Gedmatch
2) elimination by ancestry of valid matches in which the shared segments, albeit possibly very old in origin, are in fact common to an identifiable ancestor.

Jim