I ordered a few original articles using an online service. They were exceptionally cheap, and I was thrilled when I received them: they were 100% original (copy / paste several 10-word strings between "" into Google produced no results whatsoever).
I uploaded one of the articles on my website using CSE HTML Validator. All looked perfect.
Then I needed to find a word on that newly created web page and I quickly used Internet Explorer's built-in search. And surprise: the word was not found!
However I was able to see it with my own eyes: there were many occurrences of that particular word on my page. So I had to investigate what was happening. And the "writer" was exposed! He was no writer, just a thief!
Here is how they do it: they copy existing articles that are already duplicates many times. Then they run search & replace operation and globally replace several letters by identical-looking cyrillic or other foreign characters.
So a string that appears as
"beautiful river Danube. The city Park is also an area"
on my site, when copied from my site and pasted into Google gives no results whatsoever (don't try it with the above example; it won't work).
However if I copy the text by hand, I find many occurrences of that particular string.
Using an UTF-8 converter, I found out that the actual code was
beаutiful rіvеr Dаnubе. Thе Cіty Park іѕ аlsо an аrea
In other terms, this is pure theft! Don't be an accomplice. There are more or less clever variants of this scheme. The example above is pretty useless, since Google does not see most words, it has no SEO value. A more elaborated scheme modifies only stop words, so the article does have some SEO value and still appears original... until you are exposed and (most likely) banned.