I don't know what specific information is contained in the sync file. I know it is a binary file (you need a hex editor or other tool to look through it). I do know, from this experiment, link status, dates, file names, invitees, etc., seem to be contained in it. I figured if I started with a clean sync file (new file) and the problem still occurred then I could rule that out as a problem. This was based on the behavior I was noticing during the sync failures. For instance, prior to the experiment I was getting about 60% or so( estimate based on the progress bar) through the "analyzing FTM for changes" before I would get the error. I cleaned up my FTM info (duplicate people, locations, possible "special characters" in the names, etc., and the progress dropped to under 50% before the error happened. I saw that some folks recommended changing from an automatic sync to a manual sync. When I changed the setting, FTM gave me a message indicating there was an issue with the link file (which I interpreted as the sync file it created) and asked if I wanted to relink to my Ancestry tree. I let it attempt to relink the tree (successfully) and then the next sync got to almost 75% before I got an error. That was the point I started to suspect the sync file may be part of the issue.
Like I said, I've only done a limited experiment (about 1900 people in my file with lots of media) to see if anything changed by doing it this way. I could have tried to relink to the original Ancestry file but opted not to in order to eliminate that variable. Almost all of my changes were done in FTM. The few in Ancestry were minor (more of an experiment after the sync problem manifested itself)..