Shawn

"so you staked your ETH on the Ethereum blockchain to earn yield?" "yes, Dave" "except you didn't want your capital to be locked up so you actually staked it with a liquid staking protocol called Lido?" "that's correct, Dave" "and Lido gave you a liquid staking receipt token called stETH in return?" "yes, Dave" "and then you didn't think that was enough, so you juiced the yield even further by depositing your stETH receipt tokens into a restaking protocol called Eigenlayer?" "you are correct, Dave" "and now you didn't want to lock up your capital, so you actually restaked with a liquid restaking protocol called KelpDAO who provided you with a liquid restaking receipt token called rsETH?" "you got it, Dave" "and then that was surely not enough juice, so you then deposited your rsETH tokens into a lending protocol called AAVE so that you could open a leveraged looping position that borrows ETH against the rsETH collateral and restakes the ETH into rsETH which is then deposited as collateral, except it turns out rsETH used a cross-chain bridge called LayerZero whose security is held together by a 1/1 toothpick, which was obviously hacked by north koreans causing rsETH to become undercollateralized and now these looping positions are stuck and unprofitable, and everyone is pointing fingers at each other, and also DeFi is a very serious industry" "you are 100% correct, dave" jfc.

@aarondotdev Spot on

the ugly truth is that LLMs are clearly plateauing in how "intelligent" they seem. they are NOT to the point where they can fully take over coding. anyone who thinks so is either: >not running a product at scale (it's great for MVPs) >has a vested interest in ppl believing AGI is 6 months out >has no idea how to code yes, there were huge leaps in model capabilities before. but you can count them on one hand. the large companies are quite literally out of data to train these models on, so they're trying different things like synthetic data, better post-training RL loops, task-based fine-tuning. these are all guesses though. and we haven't been seeing much progress since the last big wave imo (grok 4, gemini 2.5 pro, o3) almost all proven AI researchers have agreed (and in some cases predicted) this drop off in model quality based on these scaling laws. in fact, they don't just think that the intelligence returns of scaling will fall off, they think true intelligence is not possible at all with the current LLM based approach. my point is that if you're a cs major, or a teenager that likes to code, or even in your early career you shouldn't consider leaving the field because of AI. right now, this moment in time, this might be the BEST that AI coding tools get. and they are nowhere near replacing competent SWEs. most of what you hear on twitter and in the news are largely propaganda-based glazing of AI because a massive portion of the US economy (all the large tech companies) are basing their future outlook on AI. a ton of people are going to believe this bs and quit learning to code, change jobs, etc. but here's the most realistic timeline of how this is going to play out: >publicly traded companies will layoff a ton of employees and say it's because of AI so stonk go up. my company did this a couple years ago. block did it last week. more will do the same. >cs major will become less competitive because junior hiring has gone down. less people will go into cs. >only the most dedicated and genuinely interested will remain in cs majors. >tons of large companies and small companies and individuals will build a metric fuckload or features using cursor, lovable, etc >these features will look like a mosaic of the most architectural patterns you could ever imagine. how do i know? because i read every single line of code my AI comes up with. probably 70% of it is completely retarded, but if i were to test it, it would "work". >the SECOND any of these vibe coded shills get any sort of traction, their systems will literally melt because LLMs are garbage at real system architecture. why? because any company at real scale isn't letting dario and sama train on their source code. that would be retarded. >now we have a ton of people with products that are making money but need someone to come in and fix their shit. good thing we still have those people who got a cs degree because they actually enjoy this task. AI isn't going to make more software jobs. it's going to make a LOT more shitty software. an unfathomable amount of badly written applications that will need to be rewritten from the ground up.

Team USA just won its first Olympic hockey gold in 46 years. On February 22. The exact anniversary of the Miracle on Ice. Forget the storybook narrative for a second. What happened today is a masterclass in what performance science teaches us about pressure, identity, and legacy. Consider the pressure this team was under. They walked into today carrying 46 years of near misses. The US hadn't won Olympic gold since 1980. They lost the gold medal game in 2002 and 2010...both times to Canada. Last year at the 4 Nations tournament, Canada beat them in overtime. That loss was still raw. The 1980 hero, Mike Eruzione, was in the building. He told the players before the game: "It's just a hockey game." It wasn't. And everyone knew it. Canada outshot the US 41-26. They dominated the second and third periods. Nathan MacKinnon missed an open net. Macklin Celebrini had a breakaway and couldn't convert. Devon Toews had Hellebuyck beaten and somehow the puck stayed out. Then Charlie McAvoy cleared a puck off the goal line with his glove. This was not a dominant performance. It was a team surviving enormous pressure and refusing to break. That distinction matters. How does a team perform under that kind of weight? It starts with the environment the coach creates. Mike Sullivan is now the only American-born coach to win multiple Stanley Cups AND Olympic gold. When he took over the Pittsburgh Penguins in 2015, the team was loaded with talent — Crosby, Malkin, Letang — and completely broken. His description: "There was a dark cloud over the locker room." His first move wasn't a new system or a motivational speech. It was a reframe. He told the team: "There are certain things in life we can control and certain things we can't. We needed to focus on the things that we could control and not dedicate any cognitive resources or worry to things we couldn't control." The team adopted a two-word motto: "Just play." Six months later, they won the Stanley Cup. Tonight, he helped USA do it again on the biggest stage in the world. Sullivan builds what he calls a "safe zone for learning." His video review sessions are explicitly NOT about blame. "We don't want a player walking into our video room on eggshells worried about 'Am I going to be in the film? Is Coach going to yell at me?' It's a game of mistakes. Our responsibility is to learn from them." His guiding principle from his college coach: "Before players want to know what you know, they want to know that you care." It's the difference between compliance and buy-in. Buy-in wins championships. Research backs up Sullivan. Fear-based environments don't produce peak performance. Especially when pressure is already high... They produce anxiety, risk-aversion, and choking. When people feel psychologically safe — when they know mistakes won't be weaponized against them — they take smarter risks, recover faster from errors, and perform better under pressure. We could see it in how Sullivan framed this moment in the weeks before the game. "What an incredible opportunity we have in front of us." Not a burden or expectation...Opportunity. He took the unusual step for a hockey team and kept the team in the Olympic Village instead of a hotel. His reasoning: "The Village is part of the experience." The Hughes brothers roomed together. The Tkachuk brothers roomed together. He didn't try to ignore or isolate them from the pressure. He was embedding them in it, together. And then there's the guy who scored the goal. Jack Hughes came into the Olympics injured, underperforming, slotted on the fourth line. Sullivan moved him up mid-tournament because, as he put it, "We thought by moving him and getting him more ice time, he could impact the game more." Hughes's response: "I believe in myself more than anyone. Wherever I was slotted coming into this thing, I knew I was going to play well." A coach who believed in him when results said otherwise. A player who believed in himself when the lineup said otherwise. Then two teeth got cracked in half by a high stick in the third period. And he scored the golden goal anyway. Everyone's going to remember this as the night the US ended a 46-year drought. On the anniversary. In overtime. Against Canada. But the real lesson is quieter than that. The environment you create determines the performance you get. A safe zone for learning. A focus on controllables. Relationships built on care, not fear. Pressure reframed as opportunity. That's what it looks like when a team is ready, with the right environment and support to tackle the ghosts of history. They built a culture where a team could survive 41 shots and a kid with two broken teeth could score the biggest goal of his life. The 1980 Miracle was about belief overcoming talent. Today was different. Today was talent, preparation, identity, and 46 years of accumulated hunger arriving at the same moment. -Steve

@ThePrimeagen 🙏

that's it i am making my own AI neovim integration i think i have the perfect idea

I couldn’t imagine living without believing I’m as capable as anyone out there That I have as much potential as anyone else That I can reach any goal I truly set myself to

Ethereum needs single slot finality and faster block times for L2s to be truly competitive with Malachite chains

Big milestone for Informal! Malachite is joining Circle to build Arc, a new Layer-1 for stablecoin finance. Proud to see our high-performance BFT consensus engine moving from incubation to real-world impact.

startup idea: an AI that actually disagrees with ya

seeing the 50th mispriced IPO in a row

how could anybody who works in AI or uses AI regularly think that we’re all about to lose our jobs lmao

LLM proompt maxxing is astrology for software devs

1% of Ethereum validators already run mev-commit by @primev_xyz for extra yield, secured by @symbioticfi. At @stakingsummit Dubai, @roth_edgar sat down with @MuratLite, founder of @primev_xyz, to learn more:

Movement drama aside, real talk: I feel bad for builders & founders who were actually serious & spent hours shipping Picking the right chain/ecosystem/vm is one of the most important decision a crypto founder makes, choose wisely.

I'm asking Claude 3.7 to do pretty simple stuff in Cursor, it fails misarably. I feel like I'm on a different planet compared to rest of the Twitter :D

Network highlight: @primev_xyz has launched mev-commit on Symbiotic mainnet, bringing credible commitments to Ethereum block production! They're the first to leverage @symbioticfi's slashing capabilities, enabling trust-minimized preconfirmations for transactions.

@toghrulmaharram I didn't claim my dev airdrop allocation and will forever (butthurtly) support bear posts about TIA

Putting Celestia on same level as Ethereum and Solana is the biggest psyop in CT history

@WalkerAmerica This is the way

6 YEARS AGO: Mark Zuckerberg discovered “smokin’ meats” and began his slow transition into an actual person.

You should prepare for moving your crypto startup to the US. The lawyers will still carve a pound of flesh out of your startup, but once the rules are clear the upside is unbounded. US is the largest unified free market economy in the world.

@AlexCosmonaut No

if libertarian party gets 5% of the vote in CA, they get federal funding. the irony is worth every taxpayer dollar