Very true, we would need to create another branch of web2py, do the initial conversion to Python 3, then try to maintain it, coding updates and new features twice: once for Python 2.x and once again for Python 3.x, since the two have different coding requirements. Then both would need to be tested extensively to ensure that they both work identically.
Don't get me wrong, I'm all for making web2py as cutting edge and portable as possible, but this may be one of those cases where the potential problems outweigh the benefits. Besides, Python 3 has been benchmarked to run 10% slower than Python 2.x, so until they can get that fixed, I doubt any high-traffic sites would risk it. Since Python 3 is so different by nature, that might be the time, where Massimo mentioned that a web3py would be created. New features, ideas, and ways of doing things could be created without breaking backwards-compatibility because your Python2.x web2py apps wouldn't work on Python 3 anyways, so it would be the perfect time to try new things. But again, it's a lot of work, and if a web3py were to be created now, it would probably be nothing more than a pet project until Python 3 gets it act together and becomes more popular.
