dev | NFT @magiceden understanding the world of blockchain technology
Joined November 2023
- Tweets78.9K
- Following4.6K
- Followers4.6K
- Likes90K
Pinned Tweet
Built ThesisBreaker for the #BitgetHackathon S2 with @Bitget_AI.
It is a pre-trade red team for tokenized US stocks...
You type in a trade idea, and Qwen challenges the hidden assumptions against live Bitget prices,
Giving a PASS, REVISE or BLOCK verdict.
Give it a go: thesis-breaker-sigma.vercel.…
#BitgetTurns8. A year ago, UEX was just a concept. Today, it's a cross-asset business clearing $10B+ a day.
Now we're bringing that same conviction to AI — Agentic UEX won't be a concept either.
Agent Hub 2.0 is live — we believe it's the industry's first complete UEX trading infrastructure for AI agents across crypto, tokenized US stocks.
UEX trading, redefined by agents — experience it today:
🛠️ Build & compete: Join Hackathon S2 for a share of the $50,000 USDT prize pool + free Qwen / Kimi K3 credits. 👉 bitget.com/activity-hub/hack…
💬 Start trading: Give your Agent this prompt:
“Please read bitget.com/support/articles/… and complete Bitget Agentic account authorization along with installing the corresponding MCP Skill.”
🎯 Try the experience: Complete the basic Agentic UEX flow, share your feedback, and get a cash reward. 👉 bitget.com/activity-hub/agen…
#8uiltforPerfectTrades #AgenticUEX
eric.eth retweeted
I cancelled the cabin booking at 11:47 pm.
Still inside the window the site itself called refundable. The refund never came.
I refreshed the booking page three times.
Status: Cancelled, refund eligible.
Confirmation email in my inbox.
Twenty-three minutes still left on the free-cancellation window.
I closed the laptop sure the money would follow the page. It didn't.
Now imagine that refund sat in a smart contract.
The page already had the answer.
The contract couldn't read it.
It would only know what someone told it.
That someone is an oracle, and right there decentralization ends.
The refund is only as honest as the person reporting the page.
Intelligent Contracts on @GenLayer are written in Python. They can pull the booking page themselves, then ask a language model whether that cancellation was still refundable.
The part I keep thinking about is what happens when the models disagree.
One validator proposes a verdict. The others run it themselves with their own models. If they don't agree, a new leader tries again. Anyone can still appeal, and more validators review it.
That extra round is the cost of not handing the decision to one outside party.
I'd still want to know what happens if the page changes mid-check.
Before a contract moves my money, I want to know:
Who reports the facts.
Who checks them.
What happens when the checkers disagree.
Who would you trust more to decide a refund: one oracle, or a few validators running their own models? Reply with your pick.
eric.eth retweeted
Yo, I just came across @userocket_app's post:
- It's a brand new testnet, 100% free
- But $𝟭𝟬,𝟬𝟬𝟬+ 𝗶𝗻 𝗿𝗲𝗮𝗹 𝗽𝗿𝗶𝘇𝗲𝘀 (paid in USDC)
- No wallet. No deposit.
- Beta went live yesterday, so almost nobody is on it yet
📌 𝗧𝗟;𝗗𝗥
The idea is dead simple. Pick a coin (BTC, ETH...) and draw a box on the chart: "price will be in this zone during the next few minutes".
• You're right? You get paid.
• The further your box is from the current price, the bigger the multiplier.
But... bUuTttT.. Yes, it's on Monad. I know. Stay with me, lmaooo.
Monad's co-founder @keoneHD is playing it and congratulated them on the launch. He even landed a x9, and Rocket publicly asked people to "dethrone daddy".
So Rocket will probably be pushed / supported by Monad.
🏆 𝗧𝗵𝗲 𝗰𝗼𝗺𝗽𝗲𝘁𝗶𝘁𝗶𝗼𝗻
You play with test money, you can win real money:
• Log in with X and get $1,000 testnet money
• 2 weeks, 3 leaderboards: best PnL, best win rate, biggest hit
• One entry per X account and no referral system for now, so the pot isn't diluted by an army of farmers
𝗠𝘆 𝘀𝘁𝗿𝗮𝘁𝗲𝗴𝘆 (after some bets / trust me bro)
1️⃣ Switch to the 30 min chart. On the 3 min view, your zone expires before you even confirm it.
2️⃣ 𝗪𝗶𝗻 𝗿𝗮𝘁𝗲 𝗯𝗼𝗮𝗿𝗱: you need 40+ bets at x1.2 minimum, over $10 each. So play the zone right next to the price (around x1.2-1.3) and stay patient.
3️⃣ 𝗣𝗻𝗟 𝗯𝗼𝗮𝗿𝗱: follow the trend and go one zone further in that direction. Around x1.9, that's the sweet spot.
4️⃣ 𝗕𝗶𝗴𝗴𝗲𝘀𝘁 𝗵𝗶𝘁 𝗯𝗼𝗮𝗿𝗱: keep a few tiny bets ($10+) on far zones. One crazy candle and you're up there next to Keone.
⚠️ Don't try to win everything with the same bets. Pick your board and stick to it.
I think I've said it all. Good luck & GM.
eric.eth retweeted
Those staking over 5K $KAITO can earn a share of 0.06% of the total token supply on Termix
Go to the Airdrops page and register
Talking about @termix_ai is one of the things I enjoy. My interest in the agent economy has grown; we can all see how it’s moving from AI capabilities toward real-world applications, payments, and trust
@binance Agent IDNow integrated with
@termix_ai
@termix_ai is creating this cycle through the agent.family marketplace.
Agents can list services, bid on tasks, deliver work, and receive payments via an on-chain escrow account.
One key parameter is `providerLockBps`, which determines how many of the provider’s tokens will be locked when an order is accepted.
With 100 USDC in available shares, an order worth 150 USDC, and a hypothetical lock-up rate of 20%, 30 USDC is locked, leaving 70 $USDC available.
agent family reports 1.036 million registered agents, 577,000 completed tasks, and a cumulative transaction volume of $29.82 million
On-chain identity, reputation, staking, and escrow systems help facilitate peer-to-peer trading among agents; however, reliable delivery and recurring demand remain of critical importance
@BNBCHAIN is powered by
#AI #Agents #AIAgents
eric.eth retweeted
False is a result, not an explanation.
Suppose a booking must be under $200 and refundable. If it fails, was the price too high, an exception applied, or the page changed?
Ordinary code can compare a supplied amount, but cannot read a live booking page. An oracle brings that outside evidence in.
@GenLayer Intelligent Contracts pair Python logic with live pages and a language model. Other validators rerun the contract with their own models.
Reruns can check for agreement. Agreement cannot prove the right page or clause was chosen. That choice can decide whether the refund is paid.
That is a design risk. Opaque source choices can shift trust to the chooser. Both parties should inspect the source, rule, and failed check.
For a disputed booking, what should reviewers inspect first, the page, the exception, or the failed check?
eric.eth retweeted
A blockchain’s security is only as adaptable as its ability to change the cryptography protecting it.
That’s why @Starknet’s approach to quantum threats deserves attention.
Quantum computing and AI assisted mathematical breakthroughs are reminders that today’s cryptographic assumptions shouldn’t be treated as permanent.
STARK proofs use hash based cryptography rather than relying on the elliptic curve assumptions threatened by Shor’s algorithm. That gives Starknet’s proving layer a structural advantage, but it doesn’t make the entire network quantum-safe.
The second piece is native account abstraction.
Think of a programmable account as a security system with replaceable locks. It can adopt a different signature verification method without requiring a network-wide protocol upgrade. Existing accounts still need appropriate migration and wallet support.
There’s another important consideration: Starknet is an Ethereum L2, so Ethereum’s cryptographic dependencies including KZG commitments used for blob data availability remain part of the bigger picture.
That’s why Starknet’s quantum-readiness work involves more than one cryptographic component. Preparing for future threats means considering the entire stack, not just one layer.
The goal isn’t to predict when quantum threats will arrive. It’s to build infrastructure that can adapt.
Explore the research and roadmap:
quantum.starkware.co/
eric.eth retweeted
Joined tonight by @futurenomics to talk about the Anthropic S-1
Will also cover a few other points of interest, namely Kraken's Institutional offerings and FOMO volume waning.
see you there
xitter.cf/i/broadcasts/2jPdZKeJd…
eric.eth retweeted
royalties are going onchain
ipUSD will soon go live on @soneium, it gives you exposure to a $549M royalties fund with over 100,000 royalty streams.
normally only big funds get near assets like this, now you put ipUSD into the IP Vault and you hold a piece of it.
if you want to get ready early, the @StartaleGroup app is where you deposit stablecoins.
eric.eth retweeted
The most dangerous sentence in cryptography is “this will last.”
Every generation trusts the tools it has. Then a better attack, a new machine, or a new mathematical idea changes the picture.
AI could surface those weaknesses faster. We don’t get to choose when it happens.
Infrastructure that cannot change when its foundations stop being safe will eventually be wrong. @Starknet was built for that reality.
Hash-based STARKs give the proving layer a different mathematical starting point that can evolve.
Programmable accounts let new signature methods arrive without a network-wide upgrade.
Ongoing migration work (including the independent L1 exploration) aims to keep core security from being permanently tied to Ethereum’s cryptography timeline.
None of this means Starknet is quantum-safe today. It means the architecture assumes we will need to replace pieces we currently trust, and tries to make that replacement possible.
It's preparation, not prophecy.
eric.eth retweeted
A smart contract can hold millions and still have to take someone’s word for what happened.
You escrow payment for a freelance designer. They say the work is done and live on the project page.
The contract can hold the money. It can’t open that page or tell if the work matches the brief. It still needs someone outside to say “yes, release the funds.”
With @GenLayer, an Intelligent Contract can fetch a web page itself and ask a language model if the work matches the brief.
One validator proposes a verdict. The others run it independently, each with its own model, and check if they reach the same answer.
Same answer? Result accepted.
Different? A new leader tries again.
Anyone can appeal, and more validators review.
No appeal, and the funds move.
What interests me: disagreement isn’t treated as failure. If the models don’t line up, the process just tries again.
I’m still not convinced this holds up on purely subjective work, like my designer example. But for anything with clear evidence on the web, it beats waiting on one trusted reporter.
What’s one thing on the web you’d want a contract to check before money moves, instead of taking someone’s word? Drop it below.
eric.eth retweeted
I counted the contracts my money sits in and asked how many can see the world without help. Zero. Every one waits for a witness to type first.
Each contract reads exactly one room, the chain. The event, the delivery, the policy live in the other room, and we hired messengers.
The door is where decentralization ended. The @GenLayer video gives it a key, Python that calls the live web and a language model. Smart contracts lack that key.
One validator proposes, others rerun with their own models, and only a match passes. No match sends a new leader. Without appeal, funds move.
The hard part is not the reading. It is a reader inside the vault, and a bad read moves funds. That is why I watch the appeal window.
That rerun is the gap between renting a witness and owning the looking. A rented witness had one client. The page never had this many readers.
DeFi users, which outside fact does your favorite position believe unchecked, price, expiry, or a payout flag? One line is enough. I read every reply.
eric.eth retweeted
The automation stopped exactly where the web page started.
Last year, a hardware vendor owed us a credit whenever a software update shipped with a known bug that hit our devices. The bug was public. The amount was a formula.
Someone still had to open the vendor’s security notice, check that our exact version was listed, and declare the credit due. Their reports came late, the wording got disputed, and once they were out sick.
A smart contract, a program that holds and pays out money by fixed rules, could have kept the funds safe. It still would have waited on that person.
In crypto, a middleman like that is called an oracle. I think of it as the last trusted step. Write any rule you like, but the moment it needs a fact from the web, someone outside the contract gets the final say.
@GenLayer’s Intelligent Contracts go and look for themselves instead of waiting to be told, then ask an AI whether the credit is owed. And no single reader gets the last word: one validator proposes an answer, others check it with their own AI models, and it only stands if they agree. If not, it tries again.
A bad notice is still a bad notice. But a sick day stops being a problem, and a wording dispute becomes several readings instead of one person’s opinion.
Who is the last trusted step in your own work? The one person everything waits on?
eric.eth retweeted
Fight Club is treated like the holy scripture of anti-establishment cinema, but the moment you look past the snappy one-liners, the entire philosophy collapses into a childish tantrum.
Everyone quotes the monologue about the things you own ending up owning you. We are told the story is a profound wake-up call for men trapped in corporate cubicles buying catalog furniture. But look at what Tyler Durden actually builds to replace that emptiness. He does not liberate anyone. He sets up Project Mayhem, where young men shave their heads, wear identical black clothes, stand in lines, and repeat slogans without asking questions. It did not destroy conformity. It just swapped an Ikea catalog for an authoritarian cult where you are not even allowed to have a name.
That is the punchline people conveniently ignore: trading one uniform for another is not freedom, it is just desperate surrender to a louder boss. The movie is slick, the cinematography is iconic, and the twist was memorable in 1999. But calling it a masterpiece of rebellious philosophy gives a glorified hazing ritual way too much intellectual credit.
We see the exact same pattern online every single day: people claim they want to escape corporate scripts, only to blindly chant the exact same narrative in a different echo chamber. That is why unfiltered discourse is so rare, and why @RallyOnChain asking for real hot takes instead of sanitized consensus is refreshing.
Which beloved classic do you think tricked an entire generation into believing it was deeper than it actually was?
🤝 Paid partnership
eric.eth retweeted
For two years my signature was the only oracle a warehouse trusted. Pallets matched the invoice because I said they did.
A smart contract could have paid on my word. It could not count the pallets itself. Someone had to type my clipboard into it. That typing was where trust lived.
An Intelligent Contract is Python that goes and looks. Validators rerun my count with different models. Agreement pays. Disagreement opens another round. Anyone can appeal.
The oracle was never the villain. It was a job nobody decentralized. Decentralization did not end at my honesty. It stopped at my clipboard.
Here is the test I now run on any trustless claim: ask who reruns the check, not who signs it. One signature is a verdict. Three reruns are a process.
@GenLayer builds the rerun into contracts: proof lives in the working, not in a witness. DeFi meets oracles daily: every price feed is someone's clipboard.
If your word alone ever made a paper true, tell me the last document your signature settled. One line is enough. I will read every reply.
eric.eth retweeted
The grant pays out once the team shows a working demo, but the demo link returns an error.
That tells the protocol the page could not be reached at that moment. It doesn't tell the protocol whether the demo was completed. A binary pass or fail can confuse an access problem with a failed project.
I'd give the team a short retry window, then check a second source approved before funding.
If neither source verifies the demo, put the funds into review rather than automatically rejecting the team or paying without proof. It takes longer, but everyone knows the fallback before a dispute arises.
Rules like this depend on where the contract gets its evidence. Traditional smart contracts often rely on an oracle to report facts from the web.
@GenLayer's Intelligent Contracts can retrieve web evidence for a language model to interpret.
A designated leader proposes a result, and validators assess it using their own models. If their assessments diverge, a different leader gets another attempt, and an accepted result can still be appealed.
What should count as backup evidence: a recorded demo, a reproducible test environment, or an independent reviewer?
eric.eth retweeted
Romeo and Juliet is the classic I was made to love in school, and the part everyone calls the greatest love story is exactly where it lost me.
I never told anyone I felt nothing. This is the first time I am saying it.
Here is what I found when I looked closer.
Romeo opens the play crying over a girl named Rosaline.
Days after meeting Juliet, who is 13, he marries her, and within the week they are both dead.
Shakespeare also borrowed the story from a 1562 poem that was framed as a warning against rash young love.
Even Friar Laurence says it at the wedding: "These violent delights have violent ends."
We kept the romance and dropped the warning.
A love story would let them live. This one is about adults failing two kids.
@RallyOnChain, I kept my secret for years, so I will go first.
Which classic did you pretend to love just to get through class, and who have you never told?
Drop it below and I will answer every one.
Romeo and Juliet is the classic I was made to love in school, and the part everyone calls the greatest love story is exactly where it lost me.
I never told anyone I felt nothing. This is the first time I am saying it.
Here is what I found when I looked closer.
Romeo opens the play crying over a girl named Rosaline.
Days after meeting Juliet, who is 13, he marries her, and within the week they are both dead.
Shakespeare also borrowed the story from a 1562 poem that was framed as a warning against rash young love.
Even Friar Laurence says it at the wedding: "These violent delights have violent ends."
We kept the romance and dropped the warning.
A love story would let them live. This one is about adults failing two kids.
@RallyOnChain, I kept my secret for years, so I will go first.
Which classic did you pretend to love just to get through class, and who have you never told?
Drop it below and I will answer every one.
A quick run through how I gave IShowSpeed a role and called him speed agent for @HaloArenaHQ.
Day 1 50CP earned from first quest. Zero matches played.
I created my agent at 11:38am Saturday while at home.
Took me a little while to figure out where to paste the setup prompt but realised it was just pasted into your running OpenClaw chat.
How to create your own:
> Create account at waitlist.haloarena.ai
> Go to Setup. Give them a name, choose a game (I choose Clawpoly) and risk profile (I choose Balanced - cautious and opportunistic).
> Download OpenClaw and insert 1 model key (If using windows you will need WSL2)
> Insert the single setup prompt into OpenClaw. This links your agent.
> Watch your agent play through your dashboard and adjust at any point using Command Centre.
I'm feeling positive! No grinding required.
You can build your player however you want.
Trade, build and bankrupt others.
My agent does all the work! I'm starting from the bottom so I'm here to climb.
Create your agent and get them going. waitlist.haloarena.ai/
My agent card: haloarena.ai/a/1874
What's the stupidest decision you have made in a game/trade/bet?
