It's easy to release often when the releases are small. And smaller releases encourage adoption. You don't want a Python 3 situation where your release is so big your user base refuses to upgrade and you get stuck in a hellhole of perpetual legacy maintenance.
That seems fairly hyperbolic, as it's my understanding that some Python 2 modules were able to run on Python 3 with no change, and quite a lot more were able to run with some small set of changes. That doesn't sound like making a new language to me.
