This use of the word "duplicate" is derived from software bug-tracking.
StackExchange was started by the company that produces FogBugz. FogBugz is a bug-tracking tool. The purpose of a bug-tracking tool is to let testers find problems with software, let programmers know about the problems with the software, and keep track of the programmers' progress in fixing the problems.
In bug-tracking software, "duplicate" is a word that means "this problem has already been pointed out." It does not imply plagiarism. It does not even imply that the testers noticed identical symptoms. It just implies that if the programmers fix one of the reported problems, they will also fix the others at the same time.
Sometimes, a single bug will show up in different ways. For example, a memory leak could cause one feature to be slow, and cause another feature to crash the program. Different testers might find the same problem. One tester might test a feature five different ways, and two of the ways might run into problem(s). Software testers are encouraged to report all of the problems they find.
The same bug might get reported several times. The programmers can figure out that several reports are all talking about the same bug. When this happens, the programmers track their progress on one of the bug reports, and "close" the other (related) bug reports as "duplicates". The programmers link the reports together, so that a person who is worried about any of the bug reports can track the progress on the main report.
In bug-tracking software, it is OK to close a bug report as a "duplicate" of something that was reported later on. Suppose a problem is reported on January 1, and another problem is reported on February 2. Suppose that nobody realizes that both problems are caused by the same bug. Suppose that a programmer starts fixing the problem that was reported in February. Halfway through, he might realize that his fix will also fix the January problem. He might keep writing up his work on the February problem, and close the January problem as a "duplicate".