While it looks like Grigory Perelman may indeed have solved the Poincare Conjecture, much of the news reporting is glossing over an important part of the story, maybe in part because it is an example of Yet Another nail in the coffin of mainstream journalism, this time for academic journalism.
Mr Perelman did not "publish" his paper, in any of the "accepted" journals. He just uploaded it to an online archive. This was all that was sufficient.
Of course, maths papers are somewhat of a special case, in that classical academic journals provide both too much and too little peer review.