All your XPATH expressions start with '//' which means "start from root of the document and search recursively". So when you do this:

link.SelectSingleNode("//div[@class='mlink']//span[@class='argmore']//a[@href]")

You will start not from link, but from the document's root. You probably want to do this instead:

link.SelectSingleNode("div[@class='mlink']...etc...")

which is equivalent to

link.SelectSingleNode("./div[@class='mlink']...etc...")

'.' means the current node. '/' means search only the direct children, not recursively.