If you had lost 50% of your users, you would have half of the overall site traffic that you had last time.
Let's take HN for example (I don't know if HN considers this their metric or what, just using it as an example.). Let's say they made a change that now requires everyone to login to the site each time they visit. So if you closed your browser window or were inactive for a period of time you would have to log back in. While you could still read posts and comments, you would be unable to submit new stories or comment until you logged in. While you are not logged in you would be considered a visitor. But when you log in, you now become a user.
While at first the traffic, or visitors, may remain the same, without participating users on the site the number of visitors will eventually drop as not as many people are submitting or commenting.
A site that does not require interaction, may on the other hand, simply use the two terms (visitor & user) interchangeably.
They specifically mention that they did not lose any traffic/visitor numbers: "Important: our overall traffic didn’t change. The same amount of people came to this feature’s page."
It appears as though the "lost users" were really lost interactions with the vote button: "our voting numbers decreased by ~50%."
It's fascinating to me how far people are willing to go when skewing the information to pique interest. I'll bet a lot less people would read it without "lost users" in the title.