Take with the huge mountain of salt labeled "I am not an engineer" - but I believe that's correct. Most of the "sync works via task completion" examples look like they're uploading data through an already-established connection. OmniFocus needs both upload and download, which we
haven't been able to do yet, though we
want to keep trying.
Making the sync server work in that manner is technically possible, but would require a
lot of work, and means that folks that want to host their own data would be out in the cold. I wouldn't rule it out completely, but it's not something we've had a any real discussion about doing.