The strongest line in the whole piece is probably the closing question: "how much of that code should exist at all?" That's the real question.

That is the best strategy and how all great products get born. Then when the bureaucrats, bean counters, and regulators come marching it, all bottom up innovation dies, and is replaced by centrally planned big bets. Which by definition will fail. Proof at Meta: HHVM vs Reality

"AI models are extremely bad at original thinking, so any thinking that is offloaded to a LLM is as a result usually not very original, even if they’re very good at treating your inputs to the discussion as amazing genius level insights." marginalia.nu/log/a_132_ai_b…

My religion is whatever this is

I started a software research company

It is hard to communicate how much programming has changed due to AI in the last 2 months: not gradually and over time in the "progress as usual" way, but specifically this last December. There are a number of asterisks but imo coding agents basically didn’t work before December

Killing 10 startups and 100 open source projects with one tweet.

brutal #artificialintelligence citriniresearch.com/p/2028gic

Check out Vishaka Sathish's YouTube channel for some incredibly soulful Abhangs and Bhajans. The voice which brings so much calm to a busy day! 🎶 Subscribe here: youtube.com/@vishakasathis… #DevotionalMusic #Abhang #Bhajan

The struggle with this is the massive $$$ cost required to "compile" the models. The DeepSeek-V3 training run used 2,048 H800s for 2 months. The electricity alone (at 5c/kWh) costs $200,000 vs $0.01 for a Linux build.

The last 8 years while my friends were getting PhDs and mining Bitcoin, I was on a journey to find the best Bombay Vada Pav in Bangalore. After 1000s of nalla pavs & dry buns, I’m finally excited to announce that I’ve found the one! 🫶🏻

I get this question a lot.

The 80% Problem in Agentic Coding open.substack.com/pub/addyo/p/th…

"The code looks plausible. The tests pass (or seem to). You’re under pressure to ship. You move on. Over time, you may understand less of your own codebase."