1) It was easier than porting to CP3.
2) It gave me a tangible benefit by removing all CPU performance worries once and for all. Added "performance" as a feature for Python. Worth the testing involved.
3) It removed the GIL. If you use PyPy4 STM, which is currently a separate JIT. Which will be at some point merged back into PyPy4.
So for me, Python3 can't possibly compete, and likely never will with PyPy4 once you consider the performance and existing code that runs with it. PyPy3 is old, beta, not production-ready, based on 3.2 and Py3 is moving so fast I don't think PyPy3 would be able to keep up if they tried.
Python3 is dead to me. There's not enough value for a new language. I'm not worried about library support because Py2 is still bigger than 3 and 2.7 will be supported by 3rd party libraries for a very long time else choose irrelevance (Python3 was released in 2008, and still struggling to justify its existence...). My views on the language changes themselves are stated much better by Mark Lutz[0]. I'm more likely to leave Python entirely for a new platform than I am to migrate to Python3.
PyPy is the future of Python. If the PyPy team announces within the next 5 years they're taking the mantle of Python2, that would be the nail in the coffin. All they have to do is sit back and backport whatever features the Python2/PyPy4 community wants into PyPy4 from CPython3 as those guys run off with their experiments bloating their language. I believe it's all desperation, throwing any feature against the wall. Yet doing irreparable harm bloating the language, making the famous "beginner friendly" language the exact opposite.
I already consider myself a PyPy4 programmer, so I hope they make it an official language to match the implementation. There's also Pyston to keep an eye on which is also effectively 2.x only at this time.
[0]http://learning-python.com/books/python-changes-2014-plus.ht...