These sorts of extreme costs are inherent to any trustless decentralized system like Ethereum. Storing and verifying contracts across tens if not hundreds of thousands of nodes is not free.
Do you have a demonstration of that proposition? I see no reason why that needs to be true at all. There's a big gap between "free" and "extreme costs".
That is only one factor in the equation. The cost of performing a calculation is tiny, on the other hand. How you would, even conceptually, be able to arrive at some minimum dollar figure for such a thing, that is large in comparison to, say a person's average income is unclear.
You can't just put something on thousands of devices, indefinitely, for low to no cost.
In fact, it _has_ to be at least more expensive that whatever the cheapest highly reliable storage provider offers. Otherwise I'm going to be using everyone's machines to host all sorts. Same with webhosting, and any other kind of hosting.
- removed execution from consensus. I believe consensus just gets the execution result from elsewhere and thus doesn’t get slowed down by execution (the bottleneck when I was working on libra/Diem)
- have a random committee that verifies the execution, as opposed to everyone on the network like in Ethereum
What enforcement benefit does the network provide? I'd contend it's entirely superfluous to a smart contract system, which could just offer building blocks from a trusted third party instead for constructing contracts, and that trusted third party could also notary the transaction and provide arbitration.
All of these complex problems are already solved (trust, arbitration etc) and involve recourse to real legal systems whether you use some distributed contract or a traditional one.
The area in business contracts which is a mess and could do with attention is standard clauses for different industries, which are well understood by everyone, diffs between contracts etc etc.
They are autonomous agents that act as your counterparty.
But are they autonomous and do they really have any agency outside the very limited confines of a scripted response to known situations (like if x happens send x ETH)? Are they able to send contract disputes to arbitration or courts in case of disagreement? If not that doesn't strike me as particularly useful.
I do really like the concept of standard contracts though which everyone agrees to just use - could save a lot of lawyer fees on contracts for M&A for example which really should have been standard components plugged together with very little customisation but instead take 6 months of bickering and use copy-pasted contracts inherited from 100 former deals at those lawyers with all sorts of archaic language. If I were in that space I'd be trying to attack that problem and partner with a law firm to take on either very small contracts like wills or very large contracts like M&A in a given vertical.
The reason I called them "autonomous agents" because they don't really fulfill any part of a standard contract. No choice of jurisdiction, no severability clause etc. More of a judge&jury&executioner all in one.
They are rather rigid and somewhat limited today, but I'd guess they are autonomous in the sense you can make them nearly unalterable once deployed. I suppose wih time we will be interacting with AI-agents at some point, there are some rather wild projects trying to do just that.
It may not strike you as particularly useful, but some decentralized exchanges (uniswap, dydx) that now clear more trading volume than Coinbase, a 100 billion dollar company. This has to mean something.
I have no idea if they could ever supercede legal contracts. Maybe in a 100 years.
where does the code actually execute? what's the code writen in?
whats stopping malicious code running?
On every Ethereum node that verifies it. It's usually written in Solidity and compiled to EVM (Ethereum Virtual Machine) bytecode (so it can also be written in other higher-level languages).
> whats stopping malicious code running?
It's sandboxed and since it's limited to the EVM, there isn't really any "malicious" code that could break out of the EVM AFAIK. Inside the EVM it's probably more of a philosophical question whether any code can actually be considered "malicious".
if the code is sandboxed, then what useful things can it do?
There are many networks that claim to do this, but we have yet to see how this is going to play out. A big question here is if these networks are in practice as resistant to attacks and decentralized as (post DAO) ETH.