HP

20 yo, ML undergrad by day avid crypto enthuasist by night. The next unit of compute is distributed systems.

[HP @MyMan_HP]

@Old_Samster @MessariCrypto A truly interesting field. You should check out this presentation for some technical solutions to make distributed ml training truly viable. https://t.co/7q7ePjczKr

[HP @MyMan_HP]

@MikeIppolito_ the most obvious counterpoint probably is that, there are great returns to having blockchains even though they are not completely decentralized because of semi-permissionless innovation.

[HP @MyMan_HP]

@MikeIppolito_ I guess his point is that crypto that interfaces with the real world is fundamentally incompatible with traditional regulatory systems since you basically end up I a scenario where you have to fall back of one of the two systems, in all cases you end up choosing local regulation

[HP @MyMan_HP]

@toly @Crypto_Picture @solana in an efficient market where everyone can compete to arb a price and resynchronize state, the L1 captures all this value. The total addressable market is the mev of the world.

[HP @MyMan_HP]

@toly @Crypto_Picture @solana L1´s logically centralize and synchronize social state across the world (market prices) and capture the cost of synchronizing that information. Where much of this value previously went to hedge funds and financial institutions..

[HP @MyMan_HP]

@hasufl Solaris- stanislav lem

[HP @MyMan_HP]

@hasufl agreed. The great power of blockchains is that exactly by being politically decentralized they allow for more logical centralization of social state than previously possible.

[HP @MyMan_HP]

@hasufl Would you put frax in that category?

[HP @MyMan_HP]

@ujjwalgupta49 @naruto11eth @toly @TrustlessState something like zk proof of ML-inference could work. then you just need to ensure that at all times someone can compute the inference of chain and construct a proof on chain. if you want data privacy you could use mpc to construct the proof. https://t.co/4tPtLROuH1

[HP @MyMan_HP]

@MikeIppolito_ agreed! restaking through eigenlayer could become an easy way to effectively have a form of undercollateralized/ leveraged lending. Actually wrote a thread about this.
https://t.co/ij9J5C3z3Q

[HP @MyMan_HP]

On Eigenlayer and meta liquid staking derivatives.

[HP @MyMan_HP]

You could also create a derivative where non eth rewards are sold for eth and then restaked.
In fact like in the banking industry when you have a bunch of yield generating mechanism you can compose them in a ton of ways.

[HP @MyMan_HP]

On Eigenlayer and meta liquid staking derivatives.

[HP @MyMan_HP]

Automatic rebalancing of the meta LSD. Imagine a user is restaked in two protocols beyond Ethereum, the value of the staking rewards in token x > rewards in token Y. the user might want to withdraw a bit of stake from protocol x' and deposit to restake on protocol y'.