Has it, really, though? What percentage of worldwide crypto traffic represents paying for goods and services (or doing some kind of meaningful product interaction for non-financial applications) and how much of it is just trading?
We are still in very early stages. What most likely will happen is that crypto will start to become the "backend" of the financial systems and the regular people will be using it without even knowing it, similar to the way they use Linux powered phones now. Crypto has applications beyond finance too, some things that may end up being disrupted are art and DNS (for example)
On the blockchain, these trivial questions are resolved with a simple database query.
Bitcoin is 12 years old. Comparing cryptocurrencies to the early days of internet and the web is beyond ridiculous. There's still zero evidence of what will most likely happen, except for more of what we've seen so far: speculation, extortion, money laundering.
Why I say it's still early?
Currently most of the projects are infrastructure and protocol development - kind of like when the internet was still developing. We're still on something like HTTP 1.0 and HTML 2.0 when it comes to crypto. The internet wasn't built in a single decade. It took multiple decades to get to where it is now, this is exactly the same thing playing out now. The amount of building and development is only intensifying.
I'd urge you to take a look at what a16a has been investing in https://a16z.com/crypto/
I agree, and even though avalanche (one of a16z investments in ava-labs works on this) seems to have solved the consensus problem, I have yet to see an l1 tackle the state growth issues similarly.
I think it will require some novel ways of information compression, mapping and/or storage/revival, and I suspect at least one l1 way will be related to the the DAG approach with avalanche however with the relationship between prior blocks (if you consider a block a matrix related to it block size limit at max) a series of transformations such that block x_n = sum of block transformations from 0 to n (this will be heavily reliant upon group theory) so instead of storing blocks themselves, you store the transformation matrices mapped to the hashes (sacrificing some compute for less storage).
[0] https://thecontrol.co/state-growth-a-look-at-the-problem-and...