Is it possible to create a general-purpose, turing-complete blockchain (i.e. Ethereum) that does not need gas fees? It's to my understanding that Ethereum needs gas fees due to the halting problem (in order to prevent a DOS attack). What needs to be fixed so that gas fees are no longer needed for general-purpose, turing-complete blockchains? Is taking execution off-chain the solution to removing gas fees or something else?
1
-
As you stated, the halting problem is one issue here, and that cannot be solved for a Turing-complete blockchain. But even if you gave up Turing-completeness, you'd still need an incentive for validators to actually include your transactions in a block– ZommuterCommented May 5, 2023 at 14:31
Add a comment
|
1 Answer
Maybe in theory if you could take all computations off chain and post a sufficient proof that will always be constant compute time to the chain, maybe. Generally speaking though, gas fees are needed to compensate miners/validators for the computation power they use to process your transactions. In order to develop a chain where you dont need gas fees, you would essentially need to be able to reduce every transaction to constant compute power with some proof, so that it does not vary at all. Thats a question for theoretical mathematics/computer science if its even possible for such a thing to exist.
And I would say its not just the raw problem of denial of service, its also about incentivizing miners/validators to spend more resources to properly process transaction that vary in computing cost for the members of the network.
Generally speaking though, I don't expect we will see this, and it may be mathematically impossible to design such a system in the first place. I think its less a question of blockchain technology, and more a question of a super advanced breakthrough in mathematics happening that would enable such a system.
And while like maybe you could design some system where gas fees are replaced by some other type of incentive, or maybe a chain that caps computations at a maximum size, in all likelihood the model would be inferior to just properly incentivizing efficient computations by making users pay more for high amounts of computation.
