> I don't think the whole Blockchain web3 thing is really doing that
I would argue most of whatever fraction of web3 projects are in good faith are trying to do exactly this. One reason to miss this is that most of the digital goods/services offered in web3 are financial in an economy that really only needs 0.01% the current level of finance in terms of its real value. There are some non financial projects which for the past year have branded themselves as "DePin" (full disclosure this is the area I work in).
> It's almost like you need something that makes use of it in a properly decentralised way. Meaning an app protocol?
The history of open markets for hard digital goods (not finance) is maybe longer than you think. If you read Brahm Cohen talk about his work at Mojo nation in the late 90s you'll see that early attempts at this left people discouraged about user demand, which is maybe what you are getting at with "app protocol". The comments of this old Unenumerated post have more ex-Mojo nation people arguing against similar ideas: https://unenumerated.blogspot.com/2007/06/nanobarter.html
The situation is potentially different today with demand for more control over our digital lives and less trust in spy corps bubbling under the surface of what regular people want. One charismatic example is Blue Sky which is built on a data model ready to go for decentralized storage markets (though not using any afaik). All of the discussion around open models and AI safety also look vaguely like opportunities.
IMO a lot of future success depends on the cost curves of fundamental cryptographic primitives especially snarks (which actually look promising) and also on the potential for disruption with radically new primitives: non excruciating FHE, VDFs that actually work etc