[kmm]

I'm afraid that's a common myth. For starters, there isn't really an objective measure of how conservative or old-fashioned a language is. But even subjectively, there's not really a reason to perceive American English as more old-fashioned.

All dialects randomly conserve and innovate on features. And when a dialect is split off by, in those case, geography, they will start to conserve and innovate different features. And it's true that American English conserved rhoticity which was lost in most of England, but it did innovate e.g. /æ/-raising where British English conserved the original pronunciation. But that's just two of many more features, and neither geographical region clearly conserved more than the other.

[pessimizer]

> All dialects randomly conserve and innovate on features.

Saying that there are no intelligible reasons for any feature change of a language in any dialect, (i.e. that they are all random) is a strong claim.

[ferzul]

Fortunately it is not a claim in the text you quoted. If a change is necessary, it has already happened everywhere. So neither t-flapping nor th-fronting are necessary, but both are more likely than [t] > [k]. There is no specific reason that one of them advanced through one country and the other spread through another, even if there is a good reason the third change hasn't happened. Or equivalently with the abandonment of “reckon” vs the substitution of “autumn” for “fall”.

[kmm]

Could you expand on that? Do you mean that they are predictable or possibly artificially steered?

In mainstream linguistics sound changes are assumed to be random. Some are more common than others of course, but there's no saying what sound change will happen in English next.