Isn't it a little like saying everything should be written in Java because it's safe and who can afford to take risks?
Nothing ventured, nothing gained I'd say. Taking a bit of risk on marketing can help differentiate. Isn't it also a risk to go with what's common, blending in with the pack?
• UX is good when it works in tandem with human cognition and psychology.
• HCI science--the study of human cognition and psychology, as it relates to doing UX--can point us toward designs that are objectively better in their human usability.
• A design radical-enough to not look or work like current UX--a novel "UX paradigm"--would, at this point, necessarily be one that ignores what we currently know about HCI. It would therefore, probably, be horrible. (Individual designs from such a novel paradigm could be surprisingly okay, if they manage to find some other local maximum in UX-space--but since the paradigm goes against the grain of human intuition, it wouldn't generate any globally optimal designs.)
Um, because we've figured out what works for landing pages, in an overall sense. Being different probably means forgoing conversions, which can translate into failure.
This isn't to say there isn't still room for improvement, of course. But improvement is now more incremental, with lots of A/B testing and whatnot. And for good reason.
If designers still want to be wild and creative, there are still plenty of places to do it, like personal blogs or personal projects or working on specific projects that value cutting-edge design because it fits their unique market.
But for most websites struggling to turn a profit or just gain traction period, "dreaming" is not a luxury they have. And it's disingenuous of the author to suggest that this is somehow a failure, no matter how faux-inspirational it may sound.