David E. Weekly

Honestly wonder when we will see the first vibe coded kernel. It would probably start with something very simple, almost a unikernel, no meaningful drivers, just running under a hypervisor.

@kristinatastic @Loh Attorneys General Notaries Public

I've been working on this important list for a couple of years now. What am I missing?

AngelFlight West has just been a fantastic and consistent source of joy. Third time carrying Emma - you get 'em, girl! On the empty leg back, the cockpit still smells faintly of bubble gum and unicorns.

The default transcribing engine for YouTube remains pretty abysmally poor. So I am having to use YT-DLP to pull down the audio and then run it through a real transcription system like assembly AI. I don't think Gemini can just magically generate higher quality transcripts for YouTube videos than the existing captions? But maybe I'm wrong? Can you say more about what you've done here?

@dweekly FWIW, if you want to try Gemini, it can pull straight from YouTube. Gemini-3.1-flash model is pretty good for understanding/transcribing in my experience, you might not need pro. ai.google.dev/gemini-api/doc…

For a 3-4hr multi-speaker YouTube video, what is the best automated process for accurate, diarized transcription and identification of key "chapters" / points in the video? Context: looking to better structure and index our school board's meeting recordings.

@bdurrett "Here, let me completely disregard your explicit intent because someone paid me to be able to yell at you"

@dweekly Looks like "sponsored" beat "open"? 🫤

Not impressed, Google.

@jwatte Per post: these are recordings of our school board meetings. I'm looking to summarize and extract information.

@dweekly Honest question: I don't understand people who watch YouTube videos longer than 10 minutes. What can a 4 hour video possibly bring that's worth the time investment?

First full-on Claude Code *crash*. @trq212 - (/$bunfs/root/src/entrypoints/cli.js:581:98875) - (/$bunfs/root/src/entrypoints/cli.js:581:98702) - a (/$bunfs/root/src/entrypoints/cli.js:582:263) - Lq (/$bunfs/root/src/entrypoints/cli.js:582:3681) - iq (/$bunfs/root/src/entrypoints/cli.js:582:5222) - (/$bunfs/root/src/entrypoints/cli.js:582:6732) - insertChild (/$bunfs/root/src/entrypoints/cli.js:582:31053) - zHq (/$bunfs/root/src/entrypoints/cli.js:582:36718) - iT_ (/$bunfs/root/src/entrypoints/cli.js:568:64966) - iT_ (/$bunfs/root/src/entrypoints/cli.js:568:65074)

@FutureJurvetson Ultrawiiiiiiiiiiiiiide

Loving the new office setup The MacOS background scene IRL is located on the left shore of what I see out of the Windows. And the shallow waters have that glorious green glow.

@aakashgupta 2025 study showed the opposite. "Contrary to expectations, impolite prompts consistently outperformed polite ones, with accuracy ranging from 80.8% for Very Polite prompts to 84.8% for Very Rude prompts." arxiv.org/pdf/2510.04950

Sam Altman said people saying “please” and “thank you” to ChatGPT costs OpenAI tens of millions of dollars a year in compute. 67% of Americans do it anyway. Run the math on why. A 2024 Waseda University study tested LLM responses across politeness levels in English, Chinese, and Japanese. Impolite prompts produced measurably worse outputs: more bias, more errors, more refusals. Moderate politeness consistently beat both extremes. The mechanism makes sense once you see it. Polite prompts pattern-match to higher-quality training data. When you write “Could you help me structure this analysis?”, the model pulls from professional, well-reasoned text. When you write “give me the answer,” it pulls from Reddit. Google DeepMind’s Murray Shanahan explained it simply: the model is role-playing a smart intern. Treat the intern like a colleague, you get colleague-quality work. Bark orders, you get minimum-viable compliance. Now look at the cost side. OpenAI handles over a billion queries daily. Each GPT-4 query uses roughly 2.9 watt-hours, ten times a Google search. But OpenAI just raised $40 billion at a $300 billion valuation. Tens of millions in politeness tokens is a rounding error on a rounding error. 67% of users do it anyway, and 55% of them say it’s because it’s “the right thing to do.” They’re maintaining a behavioral habit that governs every other interaction in their life. The parent who teaches their kid to say please to Alexa isn’t doing it for Alexa. They’re doing it because the alternative is raising someone who learns that being rude gets faster results. Telling 900 million people to stop saying thank you so OpenAI can save 0.01% of operating costs is the most engineer-brained optimization take on the internet. You’re training yourself to treat every interaction as a transaction. And that habit doesn’t stay in the chat window.

@mSanterre I had friends literally bankrupted by the IRS after the first dot com bubble imploded. 😞 Asked to pay tax bills for hundreds of thousands of dollars with literally no cash and equity that was nearly worthless.

Imagine paying taxes on vested RSUs at $120 and it falls by 75% by the time lockup ends.

@trq212 Time for me to take a break, I guess!

We're experiencing issues with Claude Code and Claude.ai where some users may not be able to log in & others may experience slower than usual performance. We're on it and working hard to bring service back to normal, thanks for bearing with us.

Amazon: what is the correct visual depiction of 4.2 stars?

@dflieb Which surface is this on? What app?

I’m bullish on Google for many reasons. One simple example is this feature that we wanted to ship 10 years ago but was constantly blocked by lawyers. Google is maybe finally in wartime mode.

@JeffDean - wanted to make sure you saw this! Gemini saved someone's life. reddit.com/r/GeminiAI/com…

@dfect Makes me think of all the solutions that still have absolutely no idea what to do with an HEIC file

That weird moment where you share 50+ photos in whatsapp (HD quality) with no problems. But then try to do the same in Slack and get an error. We pay a lot for Slack as a proffesional collaboration tool, Whatsapp is totally free.

New success metric for a software project: fill a plaza in Shenzhen with people trying to install your solution.

"Maybe everything will get better if we just blow some more stuff up" seems a perennial optimism of war hawks. Exciting, decisive, and tidy. Weird that things never quite turn out like that.

So where is the M5 Ultra Studio?