Clive Mctrain

@MattStopa @simonklee if u know nothing other than whats in the two blogs, youre in the majority we agree

@_xcrunch_ @simonklee I think you are in the minority thinking there were actual any hard truths here.

I like Zig. I'm one of the relatively small number of people who gets paid to write Zig professionally. Yeah, that may sound arrogant, but the reality is that there aren't many "Zig jobs" out there. Incidentally, the project I work on also uses Bun. I want Zig to succeed. I remember talking to Joran from TigerBeetle forever ago, and my only question was something along the lines of how TigerBeetle could succeed, because it's one of the few major supporters of Zig. It's incomprehensible to me that Andrew chooses to publish something like this. The blog post doesn't help the Zig project. It only alienates users and contributors, moves attention away from what Zig is trying to achieve, and makes the project look bad.

@MattStopa @simonklee obviously andrew’s post lacked the level of editorial tact but it was more honest about the motivations of the project. the line about oven’s “cloud project” (+others) shows iykyk. the flaw of the article is mixing hard truths with unecessary language

@_xcrunch_ @simonklee The Bun post laid out of lot of reasons why the language was superior. Not sure where you are getting it being a marketing thing from.

@simonklee @MattStopa but ofc it does right? the bun rewrite was 100% marketing but the blog post made it seem like language and tooling advantages were the reason for the decision? is this not shade on the zig project?

@MattStopa Yeah, the bun article doesn’t matter in regard to what Andrew wrote. It’s just a personal attack on Jarred.

@thdxr if you take money from a16z you shouldn’t comment on this blog post

this part made me lol i was ready for some big reveal on how jarred was trying to corrupt the honorable Zig Foundation in pursuit of self gain and it was that he asked for an LSP

@seatedro the exposition is hilarious to just avoid saying maybe he didnt need to grind for 5 years on a language that was always going to accrue segfaults. maybe deno being written in rust was zigs biggest argument for bun

so tl;dr on bun blogpost is they switched to rust because claude is better at rust and there's only 3 people working on bun so nobody reviews code

@returntoquality unslop.shop seems cool, good take

The hypebeastification of golf obviously must end, but the bigger issue here is the infantilization of the adult consumer. Having this in your golf bag is the equivalent of bringing a $2000 superhero lunchbox to work. Beyond the fact that it looks stupid, it clearly indicates poor judgment and misplaced priorities. For the same price you can get 20 golf lessons. Or hit 10,000 range balls. Or buy 2 months of groceries. Or take your likely neglected children on vacation. But no. You got the putter with the sandwich on it because you think it will make you more interesting, when in fact nobody will care. You’ll still miss that 3 foot putt to extend the match, and just look stupider doing it. The same is true for brand collabs in every industry. They take advantage of attention-seeking manchildren by convincing them to pay MORE to be a walking billboard for them. I think it’s dumb in streetwear, and it feels especially out of place in golf. I genuinely don’t know who the target market is for this. In short: if you buy this putter, I have no choice but to assume you are mentally retarded.

@Valuable bro u make some bullshit bible app…

Anthropic press releases be like: BREAKING: our researchers are AMAZED and SCARED to find out that … well .. you know how your brain experiences color? When we give our models a token related to color Regions we trained it to light up when dealing with that token … they LIGHT UP What the actual… Do models now experience color???

@mhaadiabu @jacobmparis u know where my man

@jacobmparis funny enough, vercel seems to be the largest OSS supporter i wonder where all the hate comes from

everyone: “companies should support the OSS they rely on!” *company puts OSS team on payroll* everyone: not like that I’m outta here

@justinskycak i mean look best of luck to you. the selection of testimonials u have is.. interesting. investor/influencer adjacent at minimum. no one needs a quant in edutech so that sticks out. knowledge graph is not a real thing, its a marketing word. to each their own

@_xcrunch_ We take in students all across the board in their initial math performance. E.g., from just a couple weeks ago, a student who went from 50th to the 87th percentile in a couple months. (By the way, none of these posts are paid or otherwise incentivized.) x.com/Steve_McGinnis…

Nadja's kids, 3rd and 6th grade, both got perfect 5's on the AP Calc BC exam. And her 3rd grader did it in record pace, finishing six years' worth of math in just over a year. Here's some backstory:

@justinskycak How key is student selection in the validity of your process/software?

I know this kind of thing may seem unbelievable, so here's some background to try to help it make sense: It was an the official AP Calculus BC exam, administered at a school in a proctored environment, the same exact thing that many HS juniors/seniors took to earn college credit. 3rd grade is our youngest so far to ace the AP Calc BC exam, but we have a track record including: -- A 5th grader who aced AP Calc BC last year: x.com/justinskycak/s… -- Numerous middle schoolers: the students in our original school program, who were representative of typical honors students, would take AP Calc BC in 8th grade. More info: #math-academy-origin-story" target="_blank" rel="nofollow noopener">justinmath.com/podcast-prep/#… I know this sounds crazy but these are official, proctored, externally-validated exams. The only way we can influence a student's score is by actually teaching them the math. Unlike virtually every other math learning program out there, we optimize for learning efficiency, meaning that every action a student takes on the system is intended to maximize the amount of knowledge they learn and retain per unit time they spend working on the system. We accomplish this by providing students with individualized learning tasks matched to their own knowledge profile and pace of learning. Here are some entrypoints where you can learn in more detail how that all works: ____________ Nerd Level 1 Why Not Just Learn From a Textbook, MIT OpenCourseWare, Khan Academy, etc.? justinmath.com/why-not-just-l… I learned from those kinds of resources myself, and while I came a long way, for the amount of effort I put into learning, I could have gone a lot further if my time were used more efficiently. That's the problem that Math Academy solves. ____________ Nerd Level 2 Overview of our pedagogy: mathacademy.com/pedagogy How our AI system works (it's an "expert system," throwback to old-school AI): mathacademy.com/how-our-ai-wor… Deep dive into our spaced repetition system: justinmath.com/individualized… ____________ Nerd Level 3 The Math Academy Way, our 400+ page working draft book: justinmath.com/files/the-math… It compiles and synthesizes evidence from hundreds of scientific papers to answer the following questions: 1. What techniques exist to maximize student learning and talent development, particularly in the context of math? 2. Why are these techniques so impactful, and if they are indeed so impactful, then why are they so often absent from traditional classrooms? 3. How does Math Academy leverage these techniques? The table of contents of The Math Academy Way is very extensive and functions as a summary of the book; each chapter also has a one-paragraph summary. Hope this helps!

@T3chFalcon just dont use windows? simple as

A 19-year-old hacker used VPNs, tunneling tools, and rotated IPs across 3 countries. The FBI still caught him. Here's the Windows feature that made it possible and why it should concern everyone. Peter Stokes, an alleged member of Scattered Spider (the group behind $100M+ in ransomware extortion), was arrested at Helsinki Airport this April. Court documents revealed a key piece of evidence: Microsoft's GDID. What is a GDID? GDID = Global Device Identifier. It's a unique code baked into every Windows installation. Microsoft uses it for telemetry, crash reports, feature usage, and license verification (it's why swapping your CPU can break your Windows activation). What Microsoft gave the FBI: → Web activity with timestamps → Gaming history → IP addresses used over time → Tool usage (including Ngrok, a tunneling app) → Azure account activity All tied to one persistent device fingerprint. Even though the VPN masked his IP, it didn't affect the GDID. How they connected the dots: Every time Stokes logged in to Snapchat, Apple, or Facebook from a new IP address, the GDID was there too. Investigators matched timestamps across platforms and countries: Tallinn, New York, Thailand, Germany. Different IPs. Same machine. Same person. Microsoft had already identified him in October 2024 and filed a criminal referral. He was still 17. So they waited till He turned 18. Then they moved. Yes, Stokes is accused of serious crimes. But the GDID data exists on every Windows machine, including yours. The infrastructure that handed his entire digital life to the FBI is the same infrastructure running on your laptop right now. The unanswered questions: - There's no public policy on when Microsoft shares GDID data - No known opt-out mechanism - No transparency report specifically covering GDID disclosures - What other criminal referrals has Microsoft quietly filed?

@nsd000 @mymoneytrades @optionscjp you will learn intelligence doesnt help with success once u age

@mymoneytrades @optionscjp yeah he is he’s just smarter than you

Be leopold > wife is chief of staff at Anthropic > get insider info at the dinner table > learn that Anthropic is going to be buying $20b of compute in Australia > become the largest shareholder of the two companies building compute in Australia > $SHAZ $IREN Beast.

@ninja_maths its cool your kids did this, very cool. what makes this a knowledge graph?

For anyone wondering how a third-grader can complete six years' worth of math in a single year AND score a 5 on the AP Calculus exam. This knowledge graph spans 3,000 math topics, from 4th grade to the university level, providing the perfect basis for mastery learning. Students can go as fast or far as they want! There are no restrictions whatsoever. The only requirement is that they must demonstrate mastery of each topic before moving on to the next. Kids are capable of incredible things when given that kind of freedom and support.

@dillon_mulroy are you using worktrees? stop! value your context and the agent’s

yall im ngl its way harder to get joy and satisfaction out of building with ai than it was before constantly straddling burn out, being far less immersed in hard problems, constant context switching i’m tired of the uncertainty of where this is going and how to do it well

@fofrAI this is worse than just directly prompting the agent to create the skill u know u want. if the agent needs to do “deep research”, youre wasting time or tokens and u arent ready to try to define a skill

I am loving this process for skill making: - setup subagents that can do deep research - ask for X research runs covering different angles of a thing - distill the research reports into a single SKILL.md - include research alongside skill for reference

@AnatoliKopadze it should have never been as low as 20-30% so this is marketing

Spotify Chief Architect, Niklas Gustavsson: "Once we implemented loops in our workflow, our agent success rate went from 20-30% to 80%." 26 minutes with Claude Code creator Boris Cherny on how Spotify runs AI agents across 20 million lines of code. Today 73% of their code is written by AI, most merged without a human ever seeing it. The model matters less than the loop you build around it. Watch it, then read the full guide on building loops below.

@max_paperclips problem with these more complex tools is that they are papering over user doing bad prompts and bad or complete lack of skills to target the agent. grep is not amazing but it maps to how we ingest code. theres no such thing as “understanding the whole codebase” for us or agents

because grep alone wasn't a good solution. there were plenty of solutions between "passive semantic rag" and "grep the entire codebase". Why not LSPs that warn the model that it's doing it wrong, or hybrid BM25-semantic indices over the codebase, with active tools to look up what it needs, skills acting as a code map & internal docs. there's a lot of stuff you can do Anthropic sophon-locked everyone by saying "grep is all you need", people just bought into it because hype. grep should be 1 tool it uses among many, not the whole

@neetcode1 stop larping u can innovate with available models fine

How to kill US innovation 101

@unusual_whales wtf is a 22 year old fresh out going to do with an LLM ??

AI employment opportunities are often reserved for those with deep experience, not beginners, per CNN