My perspective on the current Solana VS Hyperliquid drama:
A blockchain claiming credible neutrality does not mean every application built on top of it is credibly neutral.
Real credible neutrality has to survive every layer of the stack.
The base chain can be open, while the app layer recreates private access.
So if we are seriously debating neutrality, and not just marketing, the main question we should be asking is: can an ordinary participant compete under the same public rules as a sophisticated insider?
Few things to consider re: perps:
1. order flow
if some actors get better execution because of private routing, the app is not neutral in the actually strong and useful sense.
2. transaction landing
if your transaction lands better because your RPC has the right validator relationship or stake weighted access or direct peering, private mempool visibility, or better bundle infrastructure, then access to blockspace is not equal.
a DEX can be “decentralized” at the contract level while its actual liquidity depends on a tiny set of professional actors.
you must ask:
who gets rebates?
who gets inventory support?
who gets early access?
who gets routed to by default?
3. frontends
most users do not interact with contracts directly! they interact with websites, apps, wallets, mobile interfaces etc
the frontend decides what is visible, what is hidden, what is warned against, what is promoted, what is blocked, and what becomes default. That affects neutrality.
4. oracles
who defines the index?
who chooses the venues?
who can pause the feed?
who resolves disputes?
who decides what counts as manipulation?
true credible neutrality cannot stop at settlement.
5. liquidations
in derivatives, the liquidation engine is key so latency, queue design, margin parameters, auto deleveraging, insurance fund rules, and emergency controls determine a lot.
if only insiders understand the real liquidation game, the venue is not neutral in practice.
6. listings
who gets to list?
who gets early liquidity?
who gets promoted?
who gets delisted?
who gets protected?
permissionless market creation is a massive step forward, but if listing power simply moves into capital thresholds or deployer discretion, or private liquidity arrangements or even validator mediated enforcement, neutrality claims must still has to be interrogated.
7. governance and emergencies
the most revealing moment for any system is stress.
when something breaks, who coordinates?
who ships the patch?
who can object?
who is heard?
who has enough social weight to matter?
credible neutrality is mostly proven during chaos.
8. jurisdiction
every app has a legal surface.
the more successful an app becomes, the more this matters.
so the standard should be adversarial:
can i access the system without knowing anyone?
can i route the same way without private relationships?
can i build without permission?
can i list under public rules?
can i trade against the same market as everyone else?
can i verify the rules instead of trusting the operator?
can the system handle stress without informal insiders deciding outcomes?
if the answer is no then the system may still be impressive and still be valuable, it may even still be better than legacy finance.
BUT we should not confuse good product design with credible neutrality.
crypto should hold itself to a higher standard than “trust us, bro, the app is onchain.”
Any ceded ground on what is truly “neutral” is lost ground, let’s not lose ourselves to false marketing.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
This is what the Kohaku project is, straight from its core contributors, @VitalikButerin and @ncsgy.
TLDR: it is a serious attempt to rethink online safety and privacy across money, AI, and life.
Let’s dive in.
Kohaku starts from this problem: the internet leaks more than people realize.
Your wallet can leak.
Your browser can leak.
Your app can leak.
Your RPC provider can leak.
Your IP can leak.
Soon, your AI agent can leak too.
Just having private apps/txns is NOT enough.
A system is truly private when it stops leaking data, and secure when it stops leaking money. That means you have to care about every layer the user actually touches.
You cannot just clean one room in the house and leave the kitchen dirty. We need ways to clean the whole house.
adding cryptography to ai..some weekend thoughts 🧵
imagine an agent so intimately woven into your daily life that it’s almost a second brain. it checks your emails, invests your money, and even makes deals on your behalf.
but here’s the key question: how will you trust it?
leading researchers predict personal ai agents will become as common as smartphones in the next decade (see de lange et al)
these agents won’t just answer queries; they’ll evolve with you, learning from every email, transaction, or conversation you have.
BUT the risk today is that if they’re compromised or manipulated, then well...you're f*cked.
the privacy dilemma:
these ai agents will handle everything from financial transactions to sensitive health data. one breach could expose it all. so how do we solve this?
1. secure hardware enclaves (tees): run sensitive computations in isolated hardware zones, even if the OS is compromised, the data remains encrypted within the enclave. see the work of @sxysun1 @socrates1024 @ropirito @NousResearch @flashbots_x to see what this looks like for agents.
2. zero-knowledge proofs: prove an agent followed your instructions (e.g., made a correct calculation) without revealing the raw data.
the ethereum pectra upgrade is coming. it’s one of the most ambitious upgrades yet.
here is what you need to know 🧵
so, what is pectra?
pectra = prague + electra.
it’s ethereum’s glow-up.
let’s dive in.
first, the process:
teams tested over 100 devnets during an event in kenya (nyota interop). they tried everything. from there, they narrowed down 20+ ideas into 11 that are ready for prime time.