It would be a problem if we couldn't get an initial load of existing articles. But I happen to know that wouldn't be too much trouble to get hold of. As the newest nodes are integrated into the database, the scanner could build a linked list back up to the top level node id.
Of course, I'm not sure if that really matters. If you're tracking a thread, you're working down from the current level, not back up. So as long as a node is at or below the current level that you're tracking, you'd see it.
But, as we like to say, "It's only software"