But there are languages like C that don't seem to require that kind of radical change despite having been around for a long time.

C doesn't require much change because it reached its main goal of spreading a slightly less ugly version of PDP assembly code to other processors a long time ago. The language has low ambitions.

Oh man, those Slashdot threads irk me more than they should.

Python has never been morally opposed to breaking backwards compatibility. We went through this in the 1.x - 2.0 shift. Heck, this time around the Python developers are being rather charitable. There will be a 2.6 release around the same time as 3.0 final, and the 2.x line will continue for a while. Tools are being provided to make the minimal changes needed to allow crufty old code to work in the shiny new interpreter. And large organizations with a huge pile of legacy code probably won't be leaping on 3.0 the day it's released.

Some folks have been shouting about loss of backwards compatibility with the looming-ever-closer Perl 6 release despite frequent reassurances that legacy code will be covered one way or another. It doesn't look like 5.x is going to be going away any time soon. And again - despite how cool Perl 6 looks, it's going to be a foolish company that decides to convert all of their existing code to P6 on the day after Christmas.

Major version number changes are the perfect time to change your mind about how a language should be used.

Oh man, those Slashdot threads irk me more than they should.

Especially when they make a big deal out of horribly old news.