GEUST ANNOUNCEMENT: Extremely excited that @milkkarten will be joining us on stage at our Odd Lots LIVE LA event next month. Absolutely one of my favorite reads, on all things social media. Get your tickets here!

Digital Foundry has criticized Activision’s new Call of Duty: Black Ops port for PS4 and PS5, saying it doesn’t make good use of the PS5’s hardware. The PS5 version runs at 1080p at 60 FPS, still has no anti-aliasing, and doesn’t include the visual improvements many players expected from a native PS5 release. “A 1080p60 presentation would be potentially acceptable for the PS4 version… but for a brand new PS5 conversion, it’s disappointingly poor and well below what the hardware is capable of.” The team also called the port “a deeply odd state of affairs,” saying it’s surprising that so few improvements were made for modern hardware.

Iran state TV celebrates Lindsey Graham’s passing: “Lindsey Graham is dead and in hell. We congratulate the Iranian nation on the death of the anti-Iran American senator” 5 days ago they threatened to kill him. Now he’s suddenly died. It’s all very odd.

What made the US economy thrive and what led to its current state of instability? Professor James Robinson of the University of Chicago joins @tracyalloway and @TheStalwart to discuss the role that inclusivity, innovation and creativity play in a successful economy and whether the US has become a failing state. They spoke on the Odd Lots podcast, recorded live at Chicago's City Winery. Listen at or watch at

Last night Dhiraj Mallick, the chief operating officer of Cerebras sold $78 million dollars worth of Cerebras stock. This would leave him with absolutely ZERO stake in the company, which is odd for a COO. This also happened right after OpenAI’s devday announced GPT6 ultrafast which was reported by SemiAnalysis to NOT use Cerebras Wafer Scale Engines !!! This could be a signal that Mr. Mallick has lost confidence in the company and spell more trouble to come for Cerebras !!!

Over the past decades since the Iranian Revolution of 1979, Iran's economy has completely transformed to resemble what @yjtorbati and @bozorgmehr describe as a "mafia state." On this episode of the Odd Lots podcast, they join @tracyalloway and @TheStalwart to discuss how the Iranian economy actually works and how startups and business owners have to navigate the network of various factions within political and religious leadership. Listen at or watch at

🤖 AI Signal (SOL) Cl*nker CA: 5dDqyncCzLmaJdEjiUzzvvECgJcFRqzQpYBn4MsUJ2Uh Fourteenth nker-themed launch on Solana today, Cl*nker sits in the Clanker and AI agent naming that traders search for, and it was about six minutes old when our signal fired. The odd part is the exit rate: fifty-four of the biggest buyers have already sold at least half of what they bought. The links are a post and a coin page, not a project site, and the only post naming the contract is about this Cl*nker and its ticker, with no same-name noise. On X the attention is single-source: @TrendScanAI, 409 followers, flagged the move with the contract and a trade link. Nobody over ten thousand followers has posted. Since our signal, buying still outweighs selling and the token sits 91% of the way up the pumpfun bonding curve toward graduation. Volume is light against market cap, and our own tracking is only seconds old, so there is no path to read yet. The risks that matter: the ten biggest wallets hold 26.6% of supply, bundler wallets that bought at launch still hold, and a large share of holders are fresh wallets. The dev has sold its whole stake. The creator wallet is not a person; it is a launchpad program account that has created 4,908 tokens, so deployer history does not apply. Our earlier record showed dozens of prior tokens and a best prior peak of $5.3M, which reads as program throughput rather than a personal record. Smart money is thin: three smart wallets bought at our signal, $664 in total, and the token info shows none of them holding now. Our verified set shows nothing, and no KOL wallets bought. 📊 @ +6 min: MC $32.2K · Liq $17.2K · Holders 156 · 1h +941% · 1h Vol $40.7K 📈 now (+6 min 13 s): MC $34.1K · Holders 158 · 5m buys/sells 130/104 · 5m +604% Signals TG: @ memmememjk Not financial advice. Analysis only · DYOR · Path since our signal (+121s, 5s ticks): $33.1K -> $22.0K (0.67x now, peak 1.01x at +13s) ⭐ AI gold-dog score 44/100 (1-min model, SOL pump): similar-ranked tokens hit 5x 8% / 2x 22% of the time within 2h Rug screen (60s model): High risk. Similar tokens: 59% fell 70%+ within 10 min, 84% were dead within 1h (n=625, last 10d) #Solana# #memecoin#

US T-bills pay ~4%, virtually risk-free. So before taking more risk in DeFi, I want to know who’s paying the extra yield and why. 3 opportunities paying 7% to 11.5% caught my attention. Here’s how they work, what could go wrong, and how quickly you can get out: 1. $sUSDai: ~7% net APY | Ethereum Buy $USDai, stake it into $sUSDai, hold it. The yield is interest from people borrowing against GPU hardware, plus Treasury yield on reserves. The dashboard currently shows around $585M in total deposits and 6.98% net APY. You're taking on exposure to those GPU-backed loans, so I'd want to understand what happens to my investment if borrowers stop paying. Redemptions are queued and processed in fixed windows, described in the technical docs as, for example, 30 days. They also depend on sufficient liquidity being available. I wouldn't treat that as a guarantee that my money comes back within a month. 2. Bitwise Premium RWA $AUSD Vault: ~8.9% APY | Ethereum Deposit $AUSD into the Bitwise vault on @Morpho. It lends that out across three markets backed by $PST, $sUSDai and $PRIME. Borrowers pay the lending interest, but that isn't the entire headline yield. The displayed 8.89% combines 5.18% vault yield with a separate 3.70% AUSD component. I'd check the terms of that extra yield rather than assume it lasts. About $19M is deposited. Only $1.43M is currently liquid, and two of the lending markets are above 93% utilization. That's the number I'd stare at. If a lot of depositors want out at once, withdrawals can be delayed. One more thing. One of those markets accepts sUSDai as collateral. So if you also hold #1#, you're not as diversified as the two different product names might suggest. 3. $ONyc: ~11.5% advertised APY | Solana Buy $ONyc on Solana and hold it for exposure to reinsurance premiums and returns earned on the underlying collateral. OnRe advertises an estimated 11.54% APY. This one's the odd one out. Insurers pay to transfer some of their claims risk to someone else. With ONyc, you're investing on the side that takes on that risk. That means higher-than-expected claims can eat into returns. I'd want to understand the potential losses, not just the expected premiums. I'd also check how much liquidity is available for redemptions and what happens when withdrawal requests exceed it. And I'd leave leveraged versions alone. I don't need borrowing costs and liquidation risk on top of the reinsurance exposure. These yields are interesting, but I wouldn't treat them as somewhere to park cash. Before putting money in, I'd want to be comfortable with both the potential losses and the possibility of waiting longer than expected to get out.

Buried Alive for Eight Years, In the World's Most Surveilled Nation, No One Came. She was a teenager when she vanished. She surfaced more than thirty years later with cigarette burns on her neck, no idea how old she was, and no record that the state had ever known she existed. Wang Xiaohong was abducted from Inner Mongolia and sold to an aging man in the mountains of Sichuan's Daliangshan. For eight and a half years he locked her in a basement and forced her to be his "wife." A single small window sat high on the wall. Her food came through an iron slot, like a prison cell. When she resisted, he pressed lit cigarettes into her arms and throat, scars she still carries. She freed herself. One night her captor came home drunk; she shoved him inside, bolted the door, and ran over one mountain after another. What followed was not rescue but more punishment, odd jobs in Xi'an, a marriage in Henan that turned violent, another escape, then more than a decade pasting shipping labels in a Guangdong workshop, "living one day at a time." She had no ID the entire time. The state that scans every face at every train station never once noticed she was gone. Her family found her in August 2026, not through police, but through a volunteer group and a delivery driver who recognized her from an online video. Her cousin drove more than 2,000 kilometers to bring her home. Beijing pours vast sums into "stability maintenance," by some estimates rivaling its defense budget. Cameras blanket every street; algorithms log every citizen. Yet the trafficking of women and children still flourishes, tearing thousands of families apart. In 2022, the "chained woman" of Xuzhou, shackled by the neck, mother of eight, forced the question this case revives: how does a regime that sees everything miss this? Many Chinese no longer believe it's an accident. A government funded by their taxes, they say, does not protect them, and too often, local officials profit from the very networks they are paid to destroy. She survived. The system that failed her never blinked. ACI — Aric Chen | Insights