New benchmark out of Meta FAIR, Stanford, and Harvard called ProgramBench. The setup: you get a compiled executable plus its docs. Source code stripped. Rebuild the program from scratch in any language you want. Tests check input/output behavior against the original binary. 200 tasks, from small CLI tools up to FFmpeg, SQLite, and the PHP interpreter. 📊 Results across 9 models: Zero tasks fully solved. Opus 4.7 was the best, passing 95% of tests on only 3% of tasks. GPT 5.4, Gemini 3.1 Pro, and Haiku 4.5 hit 0% in that bucket. The interesting part is section 5. Even the model solutions that "worked" looked nothing like the human reference. Median 1,173 lines vs 3,068 in the original. Flat directories. Fewer functions, each one longer. GPT 5.4 wrote 96% of its final code in a single turn on most tasks and never modified existing files on roughly 40% of runs. 🎯 Why it matters for us: The benchmark separates writing code from designing software. Models can produce syntax all day. They cannot yet decompose a real system into coherent modules, pick the right abstractions, or organize a codebase the way a working engineer would. That gap is what computational orchestration points at. It is also where the durable value lives. 🛠 Try it: Pick an easier task from the repo (the paper flags nnn, fzf, gron, and jq as more tractable). Run it against Claude or your model of choice. Watch where you and the model split. Note the design decisions you make that the model never even raises. Post your runs and attempts to create a harness that would allow the model to do it. Wins, failures, weird outputs, all of it. 📍 Paper and Repo: ProgramBench I'm building something on top of this right now. More soon.

Tell us three things: 1. What you do (job, industry, student, career-changer, whatever) 2. What brought you to Clief Notes 3. One thing you're trying to figure out right now related to computing or AI I'll respond to every single one. And read each other's intros too because the person who's stuck on the same problem as you might already be in this thread. I'll go first I am Jake, I have been working in tech for 15 Years, building with Generative AI for 3 Years straight now! Excited to teach and learn! That's it. Simple, scannable, gives you data on who's joining and what they need, and keeps the feed clear for content that retains people past week one.

@Aaron Quiroz been crushing it as our Community Triage Lead. Inbound across Instagram, Skool, and Discord has gotten genuinely out of hand, and Aaron stepped in and built real order out of the chaos. He's fast, he's thoughtful, and he actually reads what you send before responding. That matters Going forward, if you're trying to reach me about work, partnerships, projects, or anything that needs a real conversation, please send it to Aaron first. He has visibility into my calendar and priorities, and he'll get things routed correctly. If it needs me, I'll be there. If someone else on the team can handle it faster, even better. I'm still around and still reading. This just makes sure your message actually gets seen and answered instead of buried under hundreds of others. Thanks for rolling with the change. And seriously, thank you Aaron.

I first found Clief Notes and @Jake Van Clief a couple weeks ago! It wasn't even from Instagram, was from a sneaky recommendation I saw somewhere. Since then, the idea of avoiding the AI hype and looking at the real structure underneath, and using folders and .MD files changed my life. _____________ Add to this post in the comments to express how Jake's no-BS "avoid the hype" wisdom has impacted your life so far! _____________ I’ll go first… I've been pulling out my HAIR for MONTHS - hitting usage limits on Claude after ONE prompt... I asked everyone - NO-ONE had the answer. And everyone I asked was obviously not a power user. The new system, means I stopped hitting usage limits! It's like free compute power! I use Claude to help my clients achieve INSANE results fast - and hitting usage limits meant days and days to do work that should be done in a handful of prompts... but hitting the walls was killing me. Thanks Jake!

A Claude code Eng just dropped this. Anyone been doing this switch yet or will try? Curious to see what @Jake Van Clief has to say. It’s a file still just a different format. Same thing right… thoughts? Pros and cons? Here to learn and discover https://x.com/trq212/status/2052809885763747935?s=46&t=Ayzo8Ebbgb8PZhLNm057bg