I suggest you read an interesting article Jeremy Rubin(Bitcoin Core Developer and technical consultant Stellar), it's called "The collapse of ETH is inevitable". And so same the answer Vitalik Buterin on this article. This should add a little understanding to the situation.
link ?
vbuterinEthereum - Vitalik Buterin • 3d
I obviously have every incentive to disagree with this, but I think there are quite a few very critical economic and technical details that the article is missing.
TLDR: we are likely not doing full "economic abstraction".
Here is the core of their argument:
Suppose we’re building a new decentralized application, BuzzwordCoin. By default, following a standard ERC-20 Token template, every transaction on BuzzwordCoin will pay gas in $ETH. Requiring every BuzzwordCoin transaction to also depend on ETH for fees creates substantial risk, third party dependency, and artificial downwards pressure on the price of the underlying token (if one must sell BuzzwordCoin for ETH ahead of time to run a BuzzwordCoin transaction, then the sell-pressure will happen before the transaction requires it, and must be a larger sale than necessary to ensure sufficient funds to cover the transaction).
Instead of paying for Gas in ETH, we could make every BuzzwordCoin transaction deposit a small amount of BuzzwordCoin directly to the block’s miner’s address to pay for the contract’s execution. Paying for Gas in a non-ETH asset is sometimes referred to as economic abstraction in the Ethereum community.
In Ethereum as it presently exists, this is absolutely true, and in fact if Ethereum were not to change, all parts of the author's argument (except the part about proof of stake, which would not even apply to Ethereum as it is today) would be correct. However, the community is strongly considering two proposals, both of which have the property that they enshrine the need to pay ETH at protocol level, and furthermore the ETH gets burned, so there's no way to de-facto take it out of the loop by making the medium-of-exchange loop go faster. The proposals are:
The modified fee market described in the draft paper here:
https://ethresear.ch/t/draft-position-paper-on-resource-pricing/2838 , where average gas usage is targeted to 50% of a (2x higher than today) gas limit, using a self-adjusting minimum transaction fee to do the targeting, where the minimum fee gets burned. This fee would be charged to the block proposer, so the block proposer could charge fees in spankchain tokens or whatever other ERC20, but the block proposer would still be responsible for coming up with the ETH to pay the minfee.
Storage maintenance fees (aka "rent"): pay N wei per byte per block to keep data in storage, or else it gets "hibernated" and you need to submit a Merkle proof to revive it. This fee also gets burned.
By my guesses, well over 2/3 of transaction fees paid could end up being burned through these mechanisms.
Without ETH, a modified version of Proof-of-Stake with a multitude of assets could still decide consensus if each node selects a weight vector for the voting power of all assets (let’s call it HD-PoS, or Heterogeneous Deposit Proof Of Stake). While it is an open research question to show under which conditions HD-PoS would maintain consensus, consensus may be possible if the weight vectors are similar enough.
I actually looked into this back in 2015, and heterogeneous deposit PoS is very hard (maybe impossible) to get right. The problem is, how does the protocol know the ratios between the values of the tokens? One could use an in-protocol decentralized exchange, but (i) this would need to be subsidized to be secure, and (ii) one can construct "pathological tokens" that have rules that are designed to treat any in-protocol penalties as a no-op. So doing this securely would possibly depend on some form of "on-chain governance", which is obviously a huge attack vector (
https://vitalik.ca/general/2018/03/28/plutocracy.html ).
So if the community is not doing HD-POS, then depositing ETH becomes the only way to get access to transaction fee revenues. So altogether, the equilibrium value of ETH in this scenario under even a standard "discounted future returns" model is very much nonzero.
Edit: just saw this:
Detractors of economic abstraction (notably, Vitalik Buterin) argue that the added complexity is not worth the ecosystem gains. This argument is absurd. If the software doesn’t support the needs of rational users, then the software should be amended. Furthermore, the actual wallet software required for any given token is made much more complex, as the wallet must manage balances in both ETH and the application’s token.
Economic abstraction can still happen at the user level; users could pay in spankchain tokens, but the block proposers would still need to cough up ETH. One could also use intermediate solutions, where third parties create "wrapper transactions" that take the fees for operations from users that are paid in spankchain tokens, and the third parties provide the ETH to the block proposer.
https://www.reddit.com/r/ethtrader/comments/9ch5ls/comment/e5av470