Coder Junkie

second halt in two weeks and the timeline barely blinked. that quiet is the interesting part. an exploit usually buys you a day of threads and post-mortems from people who never touched the thing. this one got a shrug, which tracks when almost nobody was using it to begin with. i get the letting go. hard to keep faith in something that keeps stopping, and harder when the silence around it suggests everyone else stopped caring first.

A wallet can clear a passport check and still be one of ten thousand siblings passing the same USDC around at 2am. KYC proves who signed a form. It says nothing about what the address did afterwards, which is the only thing worth scoring. We cut it from the model before launch. zScore now prices gas and drawdown, nothing else. Follow counts, referral trees, Discord roles. None of them survive contact with a burner. 320M+ wallets scored across 45+ chains. 2.17M Etherscan reads behind the model. $8.66M of member trades checked line by line, none self-reported. A wallet that held liquidity through a 30% drawdown paid real money for its record. One that round-tripped inside a single block paid gas and nothing more. The score separates them. That gap is why starting zaps unlock from trades already onchain. No checklist to grind this week. Link up to 10 wallets with a signature. No funds move. app.zaps.wtf What would your oldest wallet's gas receipts say about you? @zerufinance #ZeruAI #Zaps #zScore #Hyperliquid #DeFi

A $9M seed round, and the garage is still empty. No vans, no LiDAR rigs, no permit stack. The engineers behind Microsoft HoloLens and Booking.com systems put the money into a payout queue. Physical AI teams need street-level data from cities no crew is covering. Buying it the usual way means owning the sensors, the crews and the permits, and that stack does not travel. @vangrid_io flipped the order. A contributor walks a corner with a phone, and the capture has to clear duplicate and geometry checks before any money moves. Bad geometry gets rejected. The pass that survives settles in USDC on Base. 502,752+ captures anchored. 415 bounties completed. $239,990+ USDC settled. That works out to about $578 per accepted bounty. Every payout is traceable at explorer.vangrid.io down to the individual capture, so a buyer can audit provenance instead of trusting a vendor deck. Coverage then scales with phones already in pockets, not hardware that loses value from the delivery date. Those phones get better cameras every year without a purchase order. HashKey, Animoca, Borderless, Crypto.com Capital and an NVIDIA Inception seat read like a capex story. The payout ledger says the round went into coverage. Where would you send the first 1,000 contributors? @vangrid_io #Vangrid #PhysicalAI #DePIN #WorldModels #Base

$1B in crypto positions liquidated in 24 hours, and $913M of that was longs. btc wicked down to about 73.7k, then bounced back over 75k. eth slipped under 2.1k and one trader lost a $48M position on it. funny part: price didn't have to stay down. people who expected the bounce and were right still got liquidated before it showed up. leverage doesn't grade your thesis, only whether you survive the wait.

most of my risk models got built around rates and earnings. none of them has a field for a stranger's personal life taking the tape down overnight. that's the part of modern finance i keep failing to make peace with. you can be right about the fundamentals and still open the app red, because somebody with a public profile had a messy week and everything repriced before an explanation existed. markets have always run on gossip. it just moves faster now, and the person holding the position usually hears why last. i don't have a fix. i've just stopped pretending the spreadsheet covers everything that moves the number.

if half a million people copy your trades and your callout rewards clear eight figures annualized, social trading is obviously the future of finance. the math works from that seat. weekly money is hard to argue with, and after a while your own pnl starts reading like proof of how the market should work. most of them probably aren't lying about it. the part that gets skipped is who carries the downside. the follower takes the position and the loss, not the reward structure that made the callout worth posting in the first place. i'm still not sure social trading survives a stretch where those rewards dry up. the narrative looks a lot less obvious when nobody is getting paid to repeat it.

Every airdrop checklist opens with the same instruction: new wallet, new seed phrase, leave the old one in a drawer. That drawer usually holds two years and 800+ fills across six chains, and a new points page can still count them. Open app.zaps.wtf, connect a wallet, sign a message, read the starting zaps already sitting on it. No bridge, no deposit, no new position. Up to 10 wallets, signature only, funds stay where they are. It takes about 30 seconds. The number comes from the zScore engine behind @zerufinance, which has scored 320M+ wallets across 45+ chains, verified $8.66M in member trades, and runs 2.17M Etherscan checks through its Trust Oracle. Holding duration, liquidation history, the risk a wallet actually carried. That record survives a rotation; a snapshot of last week's volume does not. Starting zaps are points tied to the planned airdrop allocation. They carry no cash value and no conversion rate has been published. Resetting everyone to zero rewards accounts built for snapshot day. Scoring the history keeps the people who held through a drawdown. How much history is that old wallet of yours still holding? #ZeruAI #Zaps #Airdrop #DeFi #zScore

starknet is weighing a move to L1, and the first thing i'd ask is what happens to the money already bridged there. two reassurances are already out: STRK stays the token with no new one minted, and the ethereum bridge stays open. the stated reason for going independent is quantum security. starknet can upgrade chunks of its own security today but still leans on ethereum for part of the critical path, so an L1 would let it move those upgrades on its own timeline. 93.52% of voting power backed researching it. that is research, nothing has been approved, and the withdrawal mechanics are still being worked out. hardening against threats that don't exist yet is worth doing. so is making sure the users already sitting there don't eat the cost of the transition.

A mobile LiDAR crew bills $1,000 to $4,000 for one kilometer of scanned street. Permits, insurance and a van idling between contracts eat most of that. Then the street changes. Café chairs land on the sidewalk, scaffolding goes up on the north corner, and the scan is out of date before the invoice clears. Humanoid and last-mile robot teams run into this every quarter. Models trained in simulation still need current ground geometry of real cities, and no single vendor covers every alley on the planet. Today @vangrid_io pays per verified capture. A contributor walks a corner with a phone, the capture clears duplicate and geometry checks, and the bounty settles in USDC on Base. 502,752+ captures anchored. 415 bounties completed. $239,990+ USDC settled, roughly $578 per completed bounty. A survey fleet quotes $1,000 to $4,000 for a single kilometer. Every ledger entry is public at explorer.vangrid.io. Coverage spreads when the payout is small enough that thousands of people say yes, and the check is strict enough that one street only gets paid once. Which street in your city would you send a robot down tomorrow? @vangrid_io #Vangrid #PhysicalAI #DePIN #WorldModels #Base

turning $0 into $10,000 is a real result, and most people in crypto never let themselves sit with it. start at nothing, get to 2k, then 5k, then 10k. then a screenshot of someone's 100k lands in the timeline and your own number stops counting. so you size up, chase the next leg, and plenty of those accounts hand the whole thing back. the ladder got climbed, then sold for one bigger swing. i get the comparison. timelines only show the top of the ladder. but 10k out of zero still counts, and calling it a failure is what pushes people into the trade that takes it back.

an e/acc community account is asking zachxbt to look into Elysium: the pad allegedly sat behind a bundle actively selling from both sides, and the team is tied onchain to a telegram bot that still had its referral code sitting in it from the wash. small wrinkle, that chain has no mainnet yet. tochi's answer was nuke crypto to zero.

A free mint can fill in under 60 seconds, and most wallets that land a slot were created that same month. Speed-based allocation built that incentive: fresh addresses, minimal history, exit before the chart even loads. Zerus moves the filter onto wallets you already use. 3,333 NFTs, free mint on Robinhood Chain through OpenSea. No public scramble for supply. Allocation reads a zaps balance built from up to 10 wallets you link with a signature. No funds move, no seed phrase handed over. The engine behind the score has already read 320M+ wallets and verified $8.66M in member volume across 45+ chains. Enough balance puts you on the guaranteed list. Every slot carries a permanent zaps earning multiplier, Hall of Fame access that waives the $7 mint fee, and partner ticket allocations. Zaps are points. They track behavior and weight allocation. No dollar figure is claimable today. Wallet history is expensive to fake and cheap to prove. Allocation that reads it first stops paying people for being early to the queue and starts paying them for being early to the market. Checking your starting balance costs one signature at app.zaps.wtf Fresh wallet or a four-year-old one: which would you trust with a mint slot? @zerufinance #ZeruAI #Zaps #Hyperliquid #DeFi

wyck said it plainly: he just wants to fill one airdrop form and be $5k richer ten minutes later. Fair ask. That was the actual deal for a stretch. One form, a wallet connect, a follow, and a number appeared in the portfolio. The replies land on the same point, and one of them got the reason right: the old meta spoiled everyone. A drop now sounds like a six-week campaign with twenty quests, a testnet nobody asked for, and a token that unlocks slowly if it ever lists. The ratio between effort and payout moved. wyck is asking for the old math back.

Ask a mapping vendor for LiDAR of a street in San Francisco and you get a quote by Thursday. Ask the same about Jakarta and the answer is a plane ticket, a customs form and a date next month. Every world model team trains where the data is and ships where the customers are. Those two maps have never lined up. Phone cameras close most of that gap. A contributor walks a route, the device captures spatial video, and the sensors turn it into anchored 3D ground truth. Vangrid is at 502,752+ captures anchored on Base, 415 bounties paid, $239,990+ USDC settled, every payout readable at explorer.vangrid.io. Backed by a $9M seed from HashKey, Animoca and Borderless, inside NVIDIA Inception. Simulation can generate a million warehouses. It cannot generate one street that actually exists. If ground truth has to cover the cities robots actually ship to, which one should @vangrid_io map next? #Vangrid #PhysicalAI #DePIN #WorldModels #Base

Tochi pulled receipts on himself. Back in June, sahil slid into his IG DMs pitching the official Hunter Biden memecoin: Hunter posting the announcement, early buy-in offered, celebrity pump logic attached. Tochi passed. His line about it now: sahil would've been sm0king his a$$ if he'd taken the deal. Meanwhile sahil is the one posting that every KOL was a paid minion, and he was also the guy handing early allocations to prominent folks in June. Both are in the same screenshot set. Paid promos are normal on CT; the receipts just tend to surface from whoever kept theirs. Tochi can post his because he said no. Everyone who said yes has a different folder, and it is sahil's to open.

Bubblemaps says it's about to out one of CT's biggest KOLs for pulling millions out of his own audience: public wallet to attract copy traders, side wallet to dump. Why it keeps working: launchpads chase volume, and callout rewards plus leaderboards make side-wallet dumping a business model. Nothing in that setup checks for bundles. Tochi's read was funnier: did the guy build his own launchpad just to force a pumpfun deal? Nobody knows yet. "Everyone bundles" gets said whenever the supply is hidden for a reason.

Tochi's entire response to the $PUMP chart is four words: strong support at zero. Kaiz had just posted the clip of blknoiz06 walking through the levels, and that was the reply. The levels are real enough. $0.0035 to $0.0038 is the zone to hold, and losing $0.0034 on a high timeframe puts another 30% on the table. The bull case is that Pumpfun earns and keeps shipping: mobile app, perps through Hyperliquid, gacha packs, cross-chain trading. Both can be true and the joke still lands. A token whose revenue comes from other people launching coins cannot point at its own chart as proof of anything. Support is where buyers showed up last time, and last time keeps moving. No call from me on $0.0034. But support at zero is the most honest technical analysis I have read this week.

One line in a campaign sponsor's spreadsheet pays a clan leader per wallet that joined. Nothing in that row asks whether the wallet ever took a position. Reward the join and the join is what gets manufactured. Three weeks later the sponsor pulls the list, 80% of those wallets never opened a position, and the treasury is already committed. Zaps moved the payout downstream. The protocol locks the campaign vault before it opens. 95% splits across traders by Trader Score. 5% goes to the clan creator whose traders actually carry that score. A wash trade earns zero. Volume that loops back to the same wallet carries no score. Anything a sponsor committed that nobody earned returns to the sponsor instead of settling inside a farmed wallet. $8.66M in member trades has been verified through that filter, across 320M+ scored wallets on 45+ chains. zaps are points that set your allocation. They do not convert to cash today, and nobody can quote you an ROI number for them. A per-head bonus stops working the day the campaign closes. A score split only moves while somebody is actually positioned. Sign in with up to 10 wallets you already traded from. Signature only, no funds move. The grant reads the history that is already on chain and turns it into a starting zaps balance. Which are you checking first, your main trading wallet or the burner you stopped using? app.zaps.wtf @zerufinance #ZeruAI #Zaps #DeFi #Airdrop #zScore

wyck's second post on FungoLabs is the one worth reading. He runs the math on a flat mint: $50 a piece would have cleared around $130,000, and he thinks it sells out. Compare that to the sealed-bid auction FungoLabs actually ran, which copies what zkSNARKs did on Zcash. Those cleared about $1,600 each and the NFTs are down more than 75% since. Hard to look at that record and expect a crowd for round two. A modest raise started reading like a loss somewhere. At $50 you get buyers who want the token, and a floor that exists because they bought. Through a sealed bid you get winners who paid above what they thought it was worth and are underwater as the cards flip. Selling to people who want the token beats selling to people who want the fill.

FungoLabs pulled the Fungo Genesis auction and refunded all of it: $84,150, 2.1M cZAMA back to 680 wallets. Over 30,000 whitelist applications, 12,000 people through the X campaign, and 680 wallets willing to actually pay. wyck read that as ambition sliding into greed. I read the refund as the honest way out. Sealed-bid encrypted NFTs on Zama are an interesting primitive, but encryption does not create buyers, and everyone just watched the zkSNARKs crowd raise $16M on Zcash and then sit on NFTs down 75%+. Refunding instead of cutting supply and calling it a sellout is the better failure, and cheaper than a floor that never trades. Whether anything gets built on top of those Genesis NFTs I cannot call from here.