Travis Nesbit

@nicksortor You’re thinking of Blackstone. Modern “journalism”.

🚨 JUST IN: The House has OVERWHELMINGLY passed a Trump-backed bill to RESTRICT companies like Blackrock from buying up single family homes LONG overdue! 🔥 It’ll now head to the Senate, then to POTUS’ desk 🇺🇸

@fivepebbles__ @om_patel5 It’s not “learning”. It’s Bayes-like iterative conditional probability refinement. And it all goes poof when the context is removed.

@geekmdtravis @om_patel5 If it were pure instruction following, it would fail at performing tasks that go beyond simply combining the rigid templates it knows. Instead, LLMs can write novel functions, combine APls in unseen ways, and debug unfamiliar code patterns. Context learning is a real phenomenon.

RESEARCHERS JUST BUILT AN AI MODEL TRAINED ONLY ON TEXT FROM BEFORE 1931 it's called talkie. 13 billion parameters, trained exclusively on text published before december 31, 1930 its worldview is completely frozen in time the reason this matters: every major AI model today (GPT, claude, gemini, llama) was trained on the modern web. that makes it almost impossible to tell if these models actually reason or if they just memorized the answers from their training data talkie breaks that completely because it has never seen any modern information the crazy part: talkie can learn to write python code from just a few examples you show it in the prompt. despite having ZERO modern code in its training data. it's figuring out programming from 19th century mathematics texts. that's ACTUAL reasoning claude sonnet 4.6 was used as the judge in talkie's reinforcement learning pipeline. claude opus 4.6 generated the synthetic conversations used in fine tuning. a modern AI was used to train a model that's supposed to be frozen in 1930 the team already flagged this as a contamination risk they want to eliminate in future versions what they're using it to study: > long range forecasting. how well can a model "predict" the future from a frozen vantage point > invention. can it develop ideas that didn't exist until after its knowledge cutoff > LLM identity. what makes a model itself vs what's just patterns absorbed from the web alec radford built this. the same guy behind GPT, CLIP, and whisper both models are open source on hugging face. they're already planning a GPT-3 scale vintage model later this year an AI that has never seen the modern world can still reason its way to writing code. THAT alone tells you more about intelligence than any benchmark ever will

@om_patel5 Now have it omitted vowels, like modern Hebrew. Save another 20-30%.

I taught Claude to talk like a caveman to use 75% less tokens. normal claude: ~180 tokens for a web search task caveman claude: ~45 tokens for the same task "I executed the web search tool" = 8 tokens caveman version: "Tool work" = 2 tokens every single grunt swap saves 6-10 tokens. across a FULL task that's 50-100 tokens saved why does it work? caveman claude doesn't explain itself. it does its task first. gives the result. then stops. no "I'd be happy to help you with that." no "Let me search the web for you" no more unnecessary filler words "result. done. me stop." 50-75% burn reduction with usage limits getting tighter every week this might be the most practical hack out there right now

@drboorenie @om_patel5 Positon has everything to do with headaches related to sleep apnea. Obstructive sleep apnea is obstructive sleep apnea precisely because of positon.

@om_patel5 positional headaches have nothing to do with sleep apnea that’s idiopathic intracranial hypertension bro who gave this guy a twitter account

THIS MAN'S UNCLE HAD A MYSTERY ILLNESS FOR 25 YEARS. EVERY DOCTOR MISSED IT. CLAUDE FIGURED IT OUT IN ONE CONVERSATION. his 62 year old uncle had kidney failure, diabetes, hypertension, a stroke, and severe migraines that ONLY happened when lying down to sleep. neurologists couldn't explain it. nephrologists couldn't explain it. brain MRI showed nothing obvious. so he brought everything to Claude. Claude caught what every doctor missed: the headaches are positional. lying down triggers them. that's a textbook sleep apnea signal. > pulled research showing 40-57% of dialysis patients have undiagnosed sleep apnea > read his brain MRI report and flagged findings other doctors overlooked > asked about snoring. answer: loud snoring for 25 years > calculated his sleep apnea risk score: 6-7 out of 8 they got the sleep study done. results were terrifying: > breathing stops 119 times per night > oxygen drops to 78% (dangerously low) > 47 oxygen desaturations per hour > 28 minutes per night below safe oxygen levels they put him on a CPAP machine. headaches gone. 25 years of loud snoring and daily exhaustion. every doctor said "dialysis fatigue" or "just getting old." it was sleep apnea the entire time. potentially causing his hypertension, contributing to his stroke, and definitely causing his migraines. Claude even translated the full home care plan and CPAP instructions for his family. AI didn't replace his doctors but it connected dots across nephrology, neurology, pulmonology, and ENT that no single specialist was doing.

The odds of winning the lottery are probably higher than the odds of a string of doctors missing this ultra obvious sleep apnea diagnosis for 25 years. His symptoms and risk factors were so stereotypical it would have taken like 30 microseconds to order polysomnography. We screen for sleep apnea all day, every day. People ignore us a lot, but we also constantly tell people that they’re at risk of sleep apnea and need to be screened. I can imagine some bad docs not caring and not asking. But a series over 25 years? No way. Nonsense marketing thats almost certainly misrepresentikrnthr truth in one way or another; the hallmark of AI marketing in this era.

Cursor is raising at a $50 billion valuation on the claim that its “in-house models generate more code than almost any other LLMs in the world.” Less than 24 hours after launching Composer 2, a developer found the model ID in the API response: kimi-k2p5-rl-0317-s515-fast. That’s Moonshot AI’s Kimi K2.5 with reinforcement learning appended. A developer named Fynn was testing Cursor’s OpenAI-compatible base URL when the identifier leaked through the response headers. Moonshot’s head of pretraining, Yulun Du, confirmed on X that the tokenizer is identical to Kimi’s and questioned Cursor’s license compliance. Two other Moonshot employees posted confirmations. All three posts have since been deleted. This is the second time. When Cursor launched Composer 1 in October 2025, users across multiple countries reported the model spontaneously switching its inner monologue to Chinese mid-session. Kenneth Auchenberg, a partner at Alley Corp, posted a screenshot calling it a smoking gun. KR-Asia and 36Kr confirmed both Cursor and Windsurf were running fine-tuned Chinese open-weight models underneath. Cursor never disclosed what Composer 1 was built on. They shipped Composer 1.5 in February and moved on. The pattern: take a Chinese open-weight model, run RL on coding tasks, ship it as a proprietary breakthrough, publish a cost-performance chart comparing yourself against Opus 4.6 and GPT-5.4 without disclosing that your base model was free, then raise another round. That chart from the Composer 2 announcement deserves its own paragraph. Cursor plotted Composer 2 against frontier models on a price-vs-quality axis to argue they’d hit a superior tradeoff. What the chart doesn’t show is that Anthropic and OpenAI trained their models from scratch. Cursor took an open-weight model that Moonshot spent hundreds of millions developing, ran RL on top, and presented the output as evidence of in-house research. That’s margin arbitrage on someone else’s R&D dressed up as a benchmark slide. The license makes this more than an attribution oversight. Kimi K2.5 ships under a Modified MIT License with one clause designed for exactly this scenario: if your product exceeds $20 million in monthly revenue, you must prominently display “Kimi K2.5” on the user interface. Cursor’s ARR crossed $2 billion in February. That’s roughly $167 million per month, 8x the threshold. The clause covers derivative works explicitly. Cursor is valued at $29.3 billion and raising at $50 billion. Moonshot’s last reported valuation was $4.3 billion. The company worth 12x more took the smaller company’s model and shipped it as proprietary technology to justify a valuation built on the frontier lab narrative. Three Composer releases in five months. Composer 1 caught speaking Chinese. Composer 2 caught with a Kimi model ID in the API. A P0 incident this year. And a benchmark chart that compares an RL fine-tune against models requiring billions in training compute without disclosing the base was free. The question for investors in the $50 billion round: what exactly are you buying? A VS Code fork with strong distribution, or a frontier research lab? The model ID in the API answers that. If Moonshot doesn’t enforce this license against a company generating $2 billion annually from a derivative of their model, the attribution clause becomes decoration for every future open-weight release. Every AI lab watching this is running the same math: why open-source your model if companies with better distribution can strip attribution, call it proprietary, and raise at 12x your valuation? kimi-k2p5-rl-0317-s515-fast is the most expensive model ID leak in the history of AI licensing.

@meganreplies @ThePrimeagen The compiler tells you its location in milliseconds.

@vaxryy I’ve was just thinking about this 💯🔥

honestly atp we might just make the config pure lua. Transition period of like 2 major releases, small utilities stay hyprlang. There's so much we're trying to pack into hyprland that I don't think a config language makes sense anymore. Yay or nay?

@rohit4verse @thdxr

@DanielW_Kiwi Kiwis visiting Mangostein island.

Disturbing

A Johns Hopkins robot known as SRT-H removed a gallbladder by itself with 100% accuracy after watching surgery videos. It identified arteries, clipped ducts, cut tissue, and even adapted when the visuals changed mid-procedure. It followed voice commands like "grab the gallbladder head" and adjusted in real-time, just like a human trainee. It exhibited the expertise of a skilled human surgeon, even during unexpected scenarios typical in real-life, medical emergencies. The surgery was done on a life-like model, not a real person.

@ThePrimeagen @kai_brokering @claudeai 😂😂😂 And I bet this ratios the OP in a similar time.

@kai_brokering @claudeai claude fast mode that'll be about 3 hours and 41 minutes of slop, speed, and product engineering

completely forgot i have $30k worth of @claudeai credits that expires in 3 weeks wtf am i supposed to do with all this😭

How was the show Silicon Valley so ahead of its time?

Nor do I know him. But I do know the first thing he did when he bought X was cut something like 80% of the workforce. That’s fairly well aligned with the idea of running lean. And he’s made his point—it works! That sad, yes, he is undoubtedly more forward thinking than most and taken at his word about his values and the importance of space exploration for the preservation of humanity it is almost certain he would value Antarctic research. That said, I didn’t mean to speculate. The major facts informing my opinion are what I’ve read from researchers in Antarctica. But again, I think we’ve all seen enough recently to question everything that happens. So I’m for sure leaving room to change my opinion. 😂

@geekmdtravis @BGatesIsaPyscho I don't know Mr. Musk personally but I get the impression that, that is NOT how he does things. Just from observations, there seems to be a much much grander plan in play and he doesn't do things for current time he is building hardware for the day after tomorrow.

🚨🌎 Meanwhile in Antarctica Online streamer actually makes his way to the frozen continent, starts filming and is met by a huge team of security” “Really - so no one’s allowed to Starlink or a mobile phone?” “It’s very strict we need to confiscate your stuff” Now ask yourself why? Why is nobody allowed to Antarctica if there’s nothing there? Why are people not even allowed to film or possess a filming device? Who’s investing in all the security? Why is the only global Pact/Treaty that stood the test of time the Antarctica treaty - which prohibits anyone who isn’t authorised in go in there?……..

There is not unlimited bandwidth. That’s the point. There is no such thing as unlimited bandwidth. Blatant violation of the first rule of thermodynamics and also just a massive misunderstanding of computer networks. Bandwidth is limited everywhere in all forms of networking. There are records of what bandwidth is available all over the internet. And those records align well with what a person (with a brain) would expect. How’s it that you have a hard time understanding that Antarctica of all places may have bandwidth constraints? It’s wild.

@geekmdtravis @BGatesIsaPyscho So your argument is now there is not enough bandwidth because they ban unlimited bandwidth by not allowing Starlink. Who’s paying you to cock block? Actually, I’d bet you’re some type of govt employee with that logic

My experience in Silicon Valley was that — outside of marketing — they won’t spend a dime more than they need to. So, it’s possible that they’re only deploying what is needed to keep the folks down there happy. And they are. Morning ago these bases were sharing like 1-10 Mbps bandwidth for everyone. Some still are. That said, there could be more going on. Absolutely. Time will tell and I’ll keep my ears tuned in.

@geekmdtravis @BGatesIsaPyscho You added fuel to the fire. For projects like this communication is essential for survival, so why invest billions of dollars and then limit people's bandwidth and these have been served by satellites other than Starlink, why ban Starlink? too independent? make you wonder?

@daytrader333 @BGatesIsaPyscho And, the point I keep trying to make is even if he is paying for the Starlink subscription he’s still sharing the same bandwidth for satellite constellation. It’s all over Reddit. The researchers not that due to bandwidth limitations personal Starlink devices are banned.

@geekmdtravis @BGatesIsaPyscho I never argued about Antarctica’s contents. My point: No reason to ban the guy from transmitting data on his own paid devices/plans. His satellite bandwidth share would’ve been insignificant. Still observing & forming opinions there; such actions just fuel conspiracies.

As I said. No real idea. The only information I have that’s related to that issue is that there are very clear longstanding limitations on broadband access through the camps. And it’s camp specific. But even recently and at big bases like McMurdo, people report limited bandwidth and consequently personal starlink receivers are banned. Could there be something going on? Sure. But these are valid reasons and in agreement with what I’ve been told. (I do have one acquaintance who did 6 mo of his physics PhD in Antarctica, but that was pre-Starlink).

@geekmdtravis @BGatesIsaPyscho We have plenty of national parks and reserves that people are allowed to visit and record. If there is nothing in Antarctica as we are told, why not apply the same treatment. Just that alone makes the whole situation stink.