Nature reports (Errami, M. & Garner, H. Nature 451, 397-399 (2008) ) on research done by two Texas researchers who investigated 62,000 papers available online in the Medline database. They were looking for plagiarism and autoplagiarism (publishing one paper in multiple journals) and found about 1% using their tool, eTBLAST, a "text similarity-based engine for searching literature collections".
As usual, any mention of the words "plagiarism" and "Internet" in the same paragraph causes journalists to suspect that plagiarism is "on the rise" and the call and try and get me to verify this, which I refuse to do. We can't measure the amount of plagiarism, only the amount of what we find. So if we can't measure it, we can't say if it is increasing or decreasing. At least this gave me a chance to spout off on some of my favorite topics, and they broadcast a large portion of my interview this afternoon on Deutschlandradio.
Duplicate papers are indeed a problem. Sometimes, one has a minor bit of new material, and wants to republish. I have even had a journal approach me and insist on paying for a translator to translate my paper on plagiarism into English to be published in their journal. I only permitted them to do this if they let me check the translation (it was not good, would have been easier to do it myself) and if they included a footnote explicitly stating that this was a translation of a previous paper).
But apparently, in the quest for AMPAP (as many publications as possible) people submit multiple copies of papers to different journals in the hopes that no one looks at them side by side and discovery them to be identical.
Is it "okay" to plagiarize oneself on the level of paragraphs or sentences? It also looks bad when a paper consists mostly of quotes of one's own work.
Another fine line between what is acceptable and what is not acceptable behavior.