There's a lot of valid criticism to be made on web3, but this isn't it. It completely misses the mark and shows a poor understanding of what it is and how it could work.
The most glaring error is the assumption that non-essential user actions/content would be stored on an L1 blockchain. L1 is expensive, a settlement layer for things requiring maximum security. It's absurd to store tweets or likes to tweets on it.
Tweets could go to L2, and likes to tweets could be centralized, for simply not being valuable enough to store on a blockchain. Or, one could conclude that perhaps a social network may not be a good use case for the blockchain entirely, or only parts of it, like media and decentralized identities.
Not a single thought is spent on potential positive/new use cases. For example, it would be trivial for users producing high quality content or even NGOs to get a revenue stream consisting of micro donations. As an alternative to the ad model. And this concept of connected wallets can work across applications.
It would directly cut out the middle man. Similarly, as much as NFTs are frowned upon, for some independent artists they offer new possibilities (with proper anti-fraud mechanisms). Having them deeply integrated on social networks and any other connected apps may provide many new opportunities, especially if the metaverse ever takes of (I have no idea if it will).
I agree with the sentiment that web3 still has to prove itself in a big mainstream use case, and there's lots of ifs and buts, but "omg sending a tweet costs $2"...is incorrect, cynical, anti-intellectual and closed minded. A cheap shot.
Personally, I think a social network cannot be decentralized. Not because of the financials, instead because of the censorship problem. It would soon turn into 4chan or worse, and pressure to remove harmful or downright illegal content would ultimately kill or suppress the network in a way that defeats the point. A more fruitful way may be what Twitter is doing: integrating some web3 aspects into web2.