By the very nature of your argument if a Turing-complete blockchain isn't inherently worth something than a stack-based language which isn't Turing complete is certainly worthless.
I feel like your arguing with me for the sake of arguing? What's your point ETH isn't perfect, so what, it's relatively new tech which is exploring an interesting problem. The fact it's valued X _may be_ a fair evaluation or even undervalued given it's impact. The upside being possibly a DAO controlled by an AI efficiently providing resources to humanity in a way human's can't given traditional power structures. Or maybe fiat isn't so bad and we don't want robot overlords so we scrap it all. At the end of the day who cares, as long as your smart about the money you've invested into to the ecosystem you won't be hurt by a Black Swan in the cryptosphere. I think anyone working on ETH up to Vitak would have the same sentiment.
If you want to build a better language for interfacing with the VM your free to do so.
And if you're just going to use ETH to pay for real compute hardware in the cloud so you can run real software, at that point why not just use BTC?
Turing-completeness doesn't automatically make something useful.