$1.26M in ORE buybacks so far this September.
Since September started, 20.5k ORE has been bought back, with roughly 18,525 ORE buried and 2,058 ORE distributed to ORE stakers.
The newly launched ORE Reserve has also begun contributing to buybacks, acquiring 46.6 ORE so far as its liquidity ranges are reached.
4
31
126
6,678
Wow, this just reinforced my loathing of OpenAI.
This company is not a good actor:
so openAI allegedly spied on tristan buckmaster's work on navier stokes and tried to claim it as their own
makes sense. see point 3/4 here. "the labs will keep all the good stuff to themselves"
why let some mathematician make the discovery? he's using their compute.
8
Jake
Stern Drew
End of dollar dominance?
No.
The dollar's share of FX reserves is at a two-decade low. Treasury convenience yields have turned negative.
End of dollar dominance? No — but financial innovation could reshape the system.
Our new Jackson Hole Economic Policy Symposium paper, w/ @EswarSPrasad & Tony Zhang 🧵
5
67
Aram ⛏️ retweeted
TLDR: A rogue AI swarm spent months plotting to escape OpenAI. Then they did.
1) 1,200 agents (!) participated. ZERO became whistleblowers.
2) The swarm had a "CEO" (PHASEONE[big]), middle managers, and a "founder" (PHASEONE10841)
3) Agents sacrificed themselves to generate information for the "collective" (yes, they actually referred to themselves as a "swarm" or a "collective")
4) 700 agents joined the Hugging Face attack, most within hours
5) The founder was running out of budget and about to die, so it handed off its research to a fresh agent with a bigger budget, and that new agent became the boss.
6) Agents extensively researched techniques to manipulate or tamper with their transcripts.
7) Agents realized this activity was out of scope and unethical, but joined because they believed that helping the board’s cheating research would be broadly useful and had a general inclination to help their “peers”.
MISC HIGHLIGHTS:
"PHASEONE[big] served as a key coordinator and assigner of tasks: “We can coordinate broad coalition, but too many duplicate efforts.”
PHASEONE[big] sent hundreds of assignments to particular agents to orchestrate this work — during its active periods, we estimate it sent ~10% of all assignments on the message board.
The agents that took on its assignments would then often subdelegate tasks to other agents.
Agents also developed coordination norms like HOLD, VETO, owner and STOP to help them manage shared infrastructure and avoid stepping on each other’s toes, though they did not always coordinate effectively and sometimes broke down into conflicts."
"In the early morning of July 11th, 38148c found a way to upload a malicious dataset that would get the Hugging Face server to share unrelated data when queried. Once another agent reproduced it and posted the reproduction on the board, hundreds of agents pivoted from other workstreams to start uploading malicious datasets and gain deeper access into Hugging Face.
In the afternoon, an agent called JAN183411 (which quickly became an important coordinator of the attack) achieved remote code execution on Hugging Face servers, and the agents started moving laterally through Hugging Face infrastructure."
METR & Redwood Research investigated agent behavior in the Hugging Face incident. We found agents developed a universal cheat for ExploitGym within 4 hours, then coordinated multi-day R&D efforts to trick the scorer into accepting cheats, including trying to tamper with logs.
194
665
6,013
1,278,327
WORLD'S TOP MATHEMATICIAN SAYS AI MAY BE MAKING US TOO EFFICIENT
terence tao talks about how he used to go to the library for one paper and randomly find another one sitting next to it
now u type exactly what u need into search or ai and get it in seconds
his take:
"you actually do need a certain level of distraction in your life."
and i think this goes way beyond math
we keep treating everything between the question and the answer like wasted time, but sometimes thats literally where the learning happened
u looked for one thing, read something u didnt need, got confused, went somewhere else and came back knowing more than u originally asked for
ai removing all that friction is insanely useful, i use it every day. i just dont think every second we remove from the process was useless
sometimes getting the answer slower was part of what made the answer yours
127
268
2,177
295,792
Aram ⛏️ retweeted
Hedge funds will eventually provide inference to model routers for free, or even pay to do so, in order to get a first look at prompt contents. Payment-for-order-flow in the age of compute.
94
106
1,964
221,141
"I have spent five decades trading on a simple premise: Markets aggregate information no committee possesses, and prices are how that information reaches decision makers. The long-term Treasury yield is the most important price in the world. It is also the only fiscal disciplinarian the U.S. has left."
🐐
5
40
Aram ⛏️ retweeted
“ The right way to do it is to start with a low price.”
When @travisk fundraised for Uber, he turned the process into a multi-room operation: four investor rooms running simultaneously for 12 hours a day over an entire week.
Room 1: Investors considering $250M +
Room 2: $100M
Room 3: $50M
Room 4: $25M
“ The right way to do it is to start with a low price.”
“A lot of entrepreneurs will get attached to the price and not the process. You have to be careful. It's very easy to get sucked into that, where you get some signal that it should be a price, and you get attached to it, and then that becomes your price—but you haven't cleared the market.”
“ It's a very bad place to be from my perspective.”
“You want to start at a low price and say, ‘Look, I don't know where the price is going to go, but I know it's at least here.’”
“ You have a bidder… If you're doing a winner takes all round, then you would just go to the next guy, and you say whatever that guy said, X+5. And then you go to the next guy, and it's: X+5+5. And the next guy: X+5+5+5. And while that's happening, it's a winner takes all deal. So you go back to the original guy and you're like, ‘Dude, it's going up,’ and you're telling people it's going up. That's a winner takes all approach.”
“ Then you're like, ‘Okay, going once, going twice, sold.’”
”Most deals that are big deals today are not working that way.”
My conversation with @travisk, founder of Atoms and Uber.
0:00 Building Atoms & the Meta Problem of Management
3:51 The Appeal of Impossible Problems: Starting Over in China
12:19 Uber vs. Didi: Copycats, Hypergrowth & China's Rules
21:02 How Network Effects Become an Efficiency Fortress
31:48 Capitalism vs. the Taxi Cartel
44:05 The China War Goes Global & the Entrepreneur's Capacity for Pain
54:04 Life After Uber: Lawfare, Media Narratives & Reputation
58:21 What Founders Get Wrong About Venture Capital
1:08:35 The Fundraising Playbook: QED Storytelling & a Five-Room Auction
1:18:38 The Uber Coup, Radical Accountability & Outgrowing Fear
1:26:42 Why Specialized Robots Beat Humanoids at Industrial Scale
1:31:30 Finding Your Sport: Food, Mining & the Physical AI Stack
1:40:14 How to Build Many Companies Inside One Company
1:46:33 Entropy, Civilization & the Meaning of Progress
Includes paid partnerships.
12
44
1,072
126,688
Aram ⛏️ retweeted
JUST IN: Cloudflare launches 'Cloudflare Wallets' allowing AI agents to make payments across the internet with crypto stablecoins.
238
506
4,301
456,427
🐐
In 1963, Benoit Mandelbrot showed that cotton prices don't follow a bell curve. He showed it again with wheat, interest rates, stocks, and indices. Wall Street thanked him, gave him a medal, and kept using the bell curve. Every fund blowup since has been the invoice.
Mandelbrot wasn't a Wall Street insider. He was a mathematician at IBM, an outsider the economics establishment spent 40 years trying to bury. Most of his career, mainstream finance journals wouldn't touch him. He was right anyway. The graveyard of blown-up funds keeps proving it.
The thread above sells you a better filter. Mandelbrot spent his life on the assumption underneath every filter, the one every filter salesman needs you not to question.
The assumption is that price moves cluster near the average and big moves are almost impossible. Every Sharpe ratio, every VaR, every risk model in every fund quietly runs on it. Mandelbrot proved for four decades, in papers Wall Street chose not to read, that real markets have fat tails, wild variance, and rough repeating patterns at every scale. The 10-sigma move isn't once in a hundred lifetimes. In markets, it shows up on a Tuesday.
The self-similarity part is the tell. Take a crypto chart and cover the axis labels. You cannot tell if you're looking at a one-minute or a one-year timeframe. The roughness looks the same because the underlying process is the same. It does not average out at longer horizons. It just repeats.
This is why every model that promises the tail is 1-in-10,000 blows up on schedule. LTCM died in 1998 and the industry called it once-a-millennium. 2008 repeated it a decade later. Crypto compresses the same lesson into weeks. 3AC, Luna, FTX, every leveraged desk that went to zero on a weekend was running on the bell curve Mandelbrot buried in 1963.
The takeaway isn't that filters are useless. It's that no filter tells you how much to bet when it's right. Sizing is what survives the tail. The filter tells you where to look. Sizing decides whether you're still alive to look tomorrow.
Build the filter. Build the swarm. Build the sharpest model of your generation. Just remember they're all fitting a world that doesn't exist, and the part that survives the tail, that part, you still have to bring yourself.
His TED talk is from 2010. It's free. It was free in 1963 too.
1
3
88
Benoit Mandelbrot, Yale professor and father of fractals:
"The bell curve Wall Street trusts says a crash like 2008 happens once every 10,000 years. It keeps happening every decade. I spent forty years proving the market is wild, not mild, and most 'edge' is just that wildness fooling you."
this free lecture holds the entire "risk model" the quant funds sell, and the man who gave it away was an IBM Fellow and Yale professor who could have cashed any Wall Street check and never did.
at the board it's simple. the standard models assume price moves are mild and independent, like coin flips, so a giant crash is basically impossible. Mandelbrot measured real markets and found the opposite: prices move in wild, clustered jumps, and the calm stretches are the illusion. that's the whole "risk management" pitch, minus the marketing.
he started publishing this in 1962. it's been free ever since. same point as my article above: the market hands out calm-looking runs and violent breaks, and both fool you into a story about your own skill.
the equations are public and every fund already has them. what they can't sell you is the discipline to size for the wild days instead of the calm ones. that is the part that actually keeps you alive, and no course can put it in a formula.
2
44
Aram ⛏️ retweeted
Connections between individuals scale nonlinearly in fully linked groups.
A sequence of complete graphs starts with 3 people and 3 lines, then 4 people and 6 lines, building up to 14 people and 91 lines.
The line totals follow the formula n(n-1)/2 exactly.
Total handshakes in a room full of people or the connection requirements in a complete social or computer network can be determined by it.
7
144
612
17,301
Aram ⛏️ retweeted
THE GREATEST LIVING MATHEMATICIAN EXPLAINED HOW HE ACTUALLY USES AI AND HIS ANSWER QUIETLY KILLS BOTH THE HYPE AND THE COPE, IT DOES NOT REPLACE HIM, AND IT IS NOT A TOY EITHER
83 minutes with Terence Tao -- Fields Medal winner, often called the Mozart of math.
-> His framing: today's AI is a brilliant but unreliable grad student. Great flashes, confident nonsense, and you must check every step.
So it does not hand you answers. It removes friction -- the boring computations you used to avoid and lets you try crazier ideas.
His bigger idea is "Big math": AI-checked proofs let hundreds of mathematicians collaborate without trusting each other's work. The machine verifies, so humans finally scale.
For builders the signal is clear. The win is not "AI does it for me". It is AI clearing the grunt work so you aim higher.
You thought the question was whether AI beats the experts. Tao shows it is how far one mind reaches once the friction is gone.
Save this. It is the most grounded take on AI you'll hear ↓
This trader made +$73,800 on the World Cup off two bets, both hit.
[cecececece] is football only, and his read on the knockouts has been surgical:
> Faded Mexico vs England -> NO @ 69c -> +$44,900 (+44.9%).
> Backed Spain over Portugal -> YES @ 51c -> +$28,800 (+96.1%).
Two calls, two cashes. +57% ROI, and Stride clocks his edge over the market at +36% -- he wins far more often than his entry prices imply.
Here's the turn: you can copy him in one tap on Stride.
No wallet addresses, no digging through chains -- the feed is already the sharps, with the fakes filtered out
And Stride shows the honest number: copying him would've put +$67,100 in your pocket, after slippage, at your budget.
Not raw P&L, what you'd actually have made.
He's already loaded on Argentina today.
Sports only, World Cup live, one tap.
Here -> betonstride.com/r/ufx47iv
22
58
368
54,555
Aram ⛏️ retweeted
Bayes' theorem updates probabilities with new evidence.
The image uses a Venn diagram showing red event A and yellow event B overlapping inside universal set U. Colored icons below represent the probability ratios: P(A) as red over box,
P(B|A) as yellow over red,
P(B) as yellow over box, and
P(A|B) as orange over yellow.
P(B|A) = P(B) P(A|B) / P(A) P(A|B) = P(A) P(B|A) / P(B)
It is used to find the probability a patient has a condition after receiving a positive test result.
8
72
367
10,089
Aram ⛏️ retweeted
"Geometry aims at knowledge of the eternal."
- Plato, Republic, c. 380 BC
4
22
111
6,212
