Louphi @deLouphi

Probably one of the most horrifying statistics I have seen today Married + home by 30 years old in America > 1950: 50% > 1960: 52% > 1970: 48% > 1980: 45% > 1990: 43% > 2000: 35% > 2010: 25% > 2026: 12%

one of the strangest things about deep learning is that high-dimensional spaces behave in ways that completely break our intuition. 1. in high dimensions, almost all the volume of a sphere concentrates near its surface. weird, right? 2. randomly chosen vectors become almost perpendicular to each other as dimensionality increases. 3. in thousands of dimensions, two random vectors will typically have a cosine similarity close to zero. 4. neural networks can use these high-dimensional spaces to represent different features as directions in their activation space, without necessarily making them overlap too much. 5. but features can still interfere with each other and share dimensions. this is one reason understanding what a neural network has actually learned is so difficult. 6. gradient descent doesn’t just adjust weights. through training, it also shapes the geometry of the representations the network learns.

Something to keep in mind

the artidote @TheArtidote

3 days ago

Now THAT is a reframe.

51 3K 24K 392K 4K

One of the tragedies of human nature is that the dishonesty of a few prevents us from fully trusting the goodness of the many

@StockPrinter @PauloMacro @BornInvestor Its ranking algorithms and ad infrastructure weren’t acquired or copied

Cancún generated ~65% of $ASR’s operating income in 2025. Passenger traffic fell 3.5%, but higher revenue per passenger kept revenue excluding construction roughly flat. Operating income fell 1.6%. How long can pricing offset weaker traffic?

@rdd147 Advertisers optimize for return on ad spend, not cost per impression. Flooding X with cheap ad inventory doesn’t magically make it competitive with Meta’s targeting, conversion optimization, and scale

@rdd147 News flash: business owner doesn’t understand corporate finance

@leevalueroach Cigarettes literally kill you

Of course it can. Take all 100,000+ games on steam, scrape their descriptions and metadata, semantically plot them in vector space using embeddings, mathematically choose a location in vector space far away from neighbors, convert that back to a description and metadata, boom, brand new ideas ranked by how unlike any other ideas they are.

Second Chance Game @SecondChanceGme

4 days ago

@SyntheticBeef one thing AI can't do is create a brand new idea

27 1 26 153K 5

@SparkingFIRENC Shaking the feeling that he’s somehow wasting his potential by not pursuing a conventional career

I heard a fascinating story about human behaviour today. A tour guide at a historic castle was getting very high ratings from tourists. They went to see what he was doing differently. On several of the rooms he had used the red velvet ropes to block the doors. These were normally used when the rooms were being repaired or cleaned. He had put them up at the beginning of each day. During g the tour he would look around and say to the group “would you like to have a quick look in here?” And they would say yes in hushed tones. He would ceremoniously unclip the rope and let them through into the room. A room that was always a normal part of every tour. This simple little bit of theatre made people love the tour that much more. Humans love what we can’t have.

The 32 year old that spent the last decade chasing homeruns & PnL screenshots instead of hockey-sticking his equity curve is not a man I envy. The stock market is not the place to take really big risks. Might ruffle some feathers there but if you are in your 20s or 30s, I'm asking you to consider 2 very different paths: One path is making these huge bets. The ones that movies get made about. Maybe one hits, you close your laptop forever, and you’re off into the sunset …or it doesn’t. And instead you are hitting a complete reset every couple years (some of you already know this). Ground zero. The other path is playing the long game, making risk-adjusted bets, living off a healthy diet of papercuts & seemingly small wins that become BIG wins via compounding. Kicking off the compound curve early will make you 1000x richer than trapping yourself in the 5+ year boom & bust cycle of trying to hit a mega trade and walk away…bc it very rarely works like that. I’m simply asking you to consider what your life would look like in a decade if you stopped kneecapping the compounding process every couple of years.

Former $GOOGL employee on why data center capacity and electricity are the real bottlenecks, not GPU production itself, and how unreliable some of the neoclouds are ($AMZN, $GOOGL, $MSFT, $NVDA): 1. According to the expert, open-source models are closing the gap with frontier models faster than expected, with customers increasingly replacing Anthropic and OpenAI models in multi-agent architectures with open-source alternatives starting from Q2 of this year. The shift started with simple tasks like tool calling and MCP, but has moved up the stack to low-level orchestration. 2. The expert identifies capacity as the biggest constraint at the moment, with certain companies' traffic growing 5x while GPU availability, particularly in the U.S. where latency requirements are tightest, simply cannot keep up. Sourcing capacity overseas in Asia or Australia is technically possible but adds latency that most customers will not accept. 3. Hyperscalers offer reliability but are expensive, require multi-year commitments of three or more years, and have limited availability of the latest hardware, with customers already chasing B300s. AI-native companies tend to gravitate toward neoclouds due to hyperscaler pricing and contract rigidity, but neoclouds come with their own reliability issues, with GPUs dropping off unexpectedly and requiring engineering resources to manage fleet stability around the clock. 4. The third tier consists of smaller independent operators who have invested in GPU racks, seeing the opportunity to lease out capacity and recoup costs within a year or two, with the remainder being pure profit. The expert describes these as essentially unusable in practice, with operators who do nothing more than find land, install racks, connect electricity, and set up cooling, with no real cloud management, maintenance, or operational expertise behind them. 5. The expert expects overall GPU availability to increase meaningfully over the next year or two, with $NVDA continuing to ramp production. The real bottleneck is not GPUs themselves but data center capacity and electricity in the U.S. On the supply side, the expert also sees players $SPCX beginning to lease out spare capacity to the market, with more likely to follow. 6. The expert notes that U.S.-based neoclouds typically require 20-30% prepayment upfront, with the remainder paid monthly or quarterly over the contract term, and rebates negotiated at the end for any reliability incidents like traffic loss. APAC-based providers operate differently, with no prepayment required at all. found on @AlphaSenseInc

keyboard_arrow_left Previous keyboard_arrow_right Next

@SeniorScrolls @ShimmyRmcf Immediate termination 😂 I hope you’re never a manager

Guy on one of the aus subreddits asks why his "high growth" retirement account has been stuck at the same number for 5 months despite regularly putting money in, and gets this response:

A fund manager who’s compounded at 60% a year with a Sortino over 2 on why he sizes bigger in Hermès than in Nvidia: Joe Walsh is the Founding Partner of hedge fund Code Partners. Joe Walsh explains: “It’s not enough to do the work on an individual stock basis. Where I generate a lot of edge is on industry structure.” “Understanding where a dollar flows into a value chain, how each part takes a different amount of the profit pool, where the profit pool actually accumulates, and where returns on capital over that profit are exceptional.” “If it’s a dynamically changing industry like semiconductors or tech, that’s a form of fragility. Today’s frontier technology can reach obsolescence when the next generation comes out.” “In luxury, you’re dealing with a handful of brands that have been around for over a century. Hermès, Louis Vuitton, Chanel. The industry structure doesn’t change nearly as much. It’s quite static.” “Where industry structure becomes predictable, profit pools become very predictable. The confidence interval around your forecast compresses, the probability of you being way wrong declines, and the variance of that return comes down.” “Sizing for me is expected value over the volatility of that expected value being achieved.” “On entries and exits I use technical analysis. But it’s all augmenting the fundamental work. You can’t have a view that this is a very good long or short without a great sense of what it’s worth to start.”

Ethan Kho @ethanrkho

a week ago

"Take advantage of volatility. Don't get taken advantage of by it. The last thing you want is to be the guy stopped out at the low." Joe Walsh is the Founding Partner of hedge fund Code Partners. Personal track record: 60% annualized with a Sortino over 2. "The expected value

11 16 170 119K 322