Every "version" is pure w.r.t its current rules. Hard-forking is something that impacts a new chain. The old chain still exists even after a hard-fork.

_Anyone_ can run old Bitcoin versions and participate on the old chains using the pre-hard-forked-changed rules.

There is no way to "turn off" the previous chain, at least not in the protocol.

The old version could become unmaintained, have no remaining miners, no remaining validating nodes, etc. That's true. But for this to be the case, there would have to be universal disinterest in maintaining the previous tech, or mining with the previous rules, operating and running fully-validating nodes, etc. Universal disinterest cannot be measured.

I cannot offer you any justification I agree with for the previous hard-forks. At the time that those hard-forks were deployed, various justifications were given, and as of 2016 I am not sure that I can agree with those reasons now. Perhaps at the time I would have found merit in the idea that the Bitcoin ecosystem was small enough to tolerate a hard-fork or something. Another possible rationalization is that, at the time, the Bitcoin software was horribly inefficient in hundreds of ways, this is why verification took >10 hours of a blockchain that was less than 30% the size of the current 2016 blockchain.... and 32MB was probably a security hole that everyone at the time was able to universally agree must be plugged. This sort of reasoning resonates with me even as of 2016; if there was a hole that bad in modern Bitcoin, heck yeah we would all have to do a hard-fork, it's unfortunate but security is important and I like avoiding holes.

FWIW, I don't care about purity.

Here is some text on validation costs:

http://diyhpl.us/wiki/transcripts/scalingbitcoin/validation-...

http://diyhpl.us/wiki/transcripts/scalingbitcoin/hong-kong/v...

> other releases that required a hard fork as well the world didn't end

Bitcoin isn't going to destroy the world, nsh is. Anyone saying otherwise is just trolling you.