There are various opinions on using Unique IDs for records in a family Tree.
A Universally Unique IDentifier (UUID) or Globally Unique IDentifier (GUID) - the latter being the Microsoft version of a UUID, is a tag that is so unique that the chances of duplication are negligible; it would be like winning the lottery jackpot every day of your life. (Don’t hold me to the accuracy of that analogy but you get the picture - it’s possible.)
Attaching a UUID to a person will guarantee that he or she can be distinguished from every other person, in any other database, anywhere in the world. But, UUIDs are very very long, like 550e8400-e29b-41d4-a716-446655440000, and not very friendly, so for simplicity I’ll refer to them a UUID-1 and UUID-2, etc., from now on, if necessary.
The good thing about UUIDs is that they can help you share data with other researchers and avoid duplication. For example, assume two researchers are both compiling information about John Smith who lived in Bristol in 1878 and moved to New York in 1901; they’re both working on the same John Smith, but there is absolutely negligible chance that those two researchers would accidentally generate the same UUID for that John Smith. So when they come to share their information they should discover that they’ve both been working on the same person but have assigned different UUIDs. That’s the time to decide who's got the best information, which UUID should be retained and which UUID should be abandoned. Thereafter, further research into John Smith will be much easier and quicker to compare - instead of having to compare names, dates and other stuff to decide if it’s the same person, they can be sure it’s the same person because of the UUID and then choose how to deal with other information about John Smith.
Another use for UUIDs is for splitting and merging family trees. When a cutting is made from a family tree the UUID will stay with each person. So no matter what changes are made to the details about each person as they exist in separate trees, a researcher can always be sure that it’s the same person even if some information has been changed. Of course then there is a need to decide about the differences - this cannot be automated. And when merging two trees, the check for duplications is much simplified because there is a single field that tells if this is a duplicate. Again, choosing which information to keep cannot be automated.
It seems that many researchers have not been using UUIDs and that an awful lot of assumptions are made as people lift stuff out of web pages. Some of the better programs do use UUIDs in their GEDCOM files but many do not. GEDCOM calls it _UID rather than UUID as it’s a customised tag in those programs that use it.
Please note that XYFT assigns unique identifiers if they have not already been assigned or imported with data from another source. However, XYFT does not use these identifiers when connecting people within its own structure. Later versions of XYFT may do but it will depend on how reliable such identifiers become.
XY Family Tree © 2005 - 2018