Also note that scientific conduct and proper provenance are largely orthogonal to licensing issues. Published science is, regardless of licensing, by definition public and hence open to plagiarism, which is the reason why rules on proper attribution are (or should be) holy to the scientific community. They were introduced ages ago when people realised open knowledge is beneficial to science and impediments to publishing (valid) results should be removed as far as possible.
Plagiarism, if detected, is scientific suicide regardless of the license of the plagiarised work. People shouldn't cite because of CC-BY, they should cite because it's good scientific conduct.
There are some aspects where licensing and scientific conduct are not altogehter orthogonal -- clear licenses make proper reuse easier, and, as Peter already pointed out, publicly accessible texts may help identifying fraud, in particular if bulk analysis of large bodies of text and/or data is covered by the license.