Satvik Sharma

(me): so we bundle instructions and data together into “cache lines” to improve efficiency (medieval peasant): wouldst thou not fear false idols when fetching from yon predictive oracle? (me): well, yeah. that’s why we invented “speculative execution” (peasant, trembling): and when this prophecy fails? (me): then we must perform “branch misprediction recovery,” restoring the righteous state (peasant, grimly): truly, thou art plagued by thy hubris, courting divine retribution with each cycle

I'm confused by the very premise of this question, though? My PhD students are not code robots I command to implement what I want. They are human beings I am mentoring to become independent researchers

I call this "California Software Syndrome" Most of the software we use is mostly made in Cali, so if a feature doesn't make sense there, they don't bother implementing it

i would be so embarrassed to be this ignorant of the history of ai while being such an ai booster no wonder the current ai boom is so vacuous everyone who's jerking of to it all has no idea what ai has been or could be, they live completely detached from the field

We’re acquiring Promptfoo. We’re buying Safetysnoot. Proud to announce that the Ragschlorp team is joining us. We own a majority stake in Chunkwad.

an underlying layer that makes AI generations obvious (that better tech can’t fix) is that the people using it aren’t creative enough to think of anything interesting for AI to make in the first place

PSA: for security, put your AI agent in a water-tight sandbox, such as a dedicated VM. Once this is done, you can maximize productivity by giving it your credit card number, email credentials, the ability to write and run arbitrary code, and unconstrained access to the internet

i think the people who were obsessed with crafting high quality software and the people who were obsessed with “disrupting” were two very different cultural groups, but go off

We used to be a country. A proper country. With James Bond dress-up episodes.

Extremely good.

Just tell Claude to keep the reliability high bro. What do you mean software engineering isn’t just ✨producing some code✨ but tradeoffs, edge cases, infra, monitoring, scale, security, and 3am outages?

"The benchmarks we use to evaluate AI coding tools are measuring memorization, not understanding."

New research just exposed the biggest lie in AI coding benchmarks. LLMs score 84-89% on standard coding tests. On real production code? 25-34%. That's not a gap. That's a different reality. Here's what happened: Researchers built a benchmark from actual open-source repositories real classes with real dependencies, real type systems, real integration complexity. Then they tested the same models that dominate HumanEval leaderboards. The results were brutal. The models weren't failing because the code was "harder." They were failing because it was *real*. Synthetic benchmarks test whether a model can write a self-contained function with a clean docstring. Production code requires understanding inheritance hierarchies, framework integrations, and project-specific utilities. Different universe. Same leaderboard score. But it gets worse. A separate study ran 600,000 debugging experiments across 9 LLMs. They found a bug in a program. The LLM found it too. Then they renamed a variable. Added a comment. Shuffled function order. Changed nothing about the bug itself. The LLM couldn't find the same bug anymore. 78% of the time, cosmetic changes that don't affect program behavior completely broke the model's ability to debug. Function shuffling alone reduced debugging accuracy by 83%. The models aren't reading code. They're pattern-matching against what code *looks like* in their training data. A third study confirmed this from another angle: when researchers obfuscated real-world code changing symbols, structure, and semantics while keeping functionality identical LLM pass rates dropped by up to 62.5%. The researchers call this the "Specialist in Familiarity" problem. LLMs perform well on code they've memorized. The moment you show them something unfamiliar with the same logic, they collapse. Three papers. Three different methodologies. Same conclusion: The benchmarks we use to evaluate AI coding tools are measuring memorization, not understanding. If you're shipping code generated by LLMs into production without review, these numbers should concern you. If you're building developer tools, the question isn't "what's your HumanEval score." It's "what happens when the code doesn't look like the training data."

The gray fallacy.

the feeling of breaking Jake Paul's jaw is literally its own reward lmao

The issue with caving to Zionists on “globalize the intifada” is the fact that the very same people who insist that it means “k*ll all Jews” say the same exact thing about the words “free Palestine.” Their problem isn’t the slogan, it’s the concept of Palestinians being free.

Periodic reminder that housing cannot be both affordable and a means of building wealth. Those two concepts are diametrically opposed to each other. If you're going to make housing affordable for young people it's going to reduce the wealth of the Boomers.

the mistake so many people make is seeing university professors as intellectuals when they’re actually employees at a combination hedge fund and healthcare conglomerate that operates a small luxury resort/sports franchise where student-customers occasionally take classes

yet another person posting how they, solo, have 347 agents writing 100,000 lines of closed source code for a project they still haven't shipped yet