I was writing code for a proof of concept for a new content tool and checking the content I generated against Copyscape. What I found is that it seems as though people are doing something similar to what I was coding does. Most of the articles I generated listed Copyscrape results of articles matching around 50% of the content. This tells me that some of the most popular news sites in the world are most likely using an automated method of generating new articles before altering them a bit. There's no other way I could be getting these results. For instance, this article I just generated shows 7 Copyscape matches, each around 50% likeness. How much similarity is too much for the content to be truly useful? Is it more like a gradient or is there a cutoff point?