>Does it matter . . .
To you, no; to me? No.
The same thing happened to me: Rails 3 came along. My web applications weren't so large that the port would be unreasonable, so I gritted my teeth and came out feeling like I had gained something. (Knowledge, experience, productivity.)
---
For every "you", however, there is someone with a non-trivial Rails 2.x codebase that simply won't invest the time or energy to port it. So they'll keep using it; keep developing libraries for it, keep patching it, etc.
Continue this trend, and I'd imagine you end up with something like the Python2/3 schism.
For an example of what this leads to, look at Rust.
Arch Linux uses Python 3 as the default for /usr/bin/python, Rust (0.3 and it looks like 0.4) fails to compile because LLVM can't stand Python 3. [This is described in the Rust 0.4 release notes. Of course the fix is a simple symlink and an EXPORT= away once you install Python2.]
So you end up with _new and exciting projects_ that require you to install and maintain old tools and libraries. Why? Because eventually _a non-negligible_ number of library maintainers stop trying to hit a moving target.
I'm not saying Rails will _necessarily_ go down that road; but Python has already been there, and the Python 2/3 schism isn't pretty. It's a possibility worth entertaining, anyways.