As an angry old man, I'm always eager to shit on whatever "the kids" are excited about, and this month that's been the phrase "vibe coding". But today I was feeling a vibe myself, and decided to delay recording of v35 of Breaking Change and just do my first-ever vibe coding session on camera instead.

The only rule: AI only. I didn't touch a line of code in the editor window. That's right, after over 2 years of using GitHub Copilot, I got to be copilot for once. 😎

Stuff you'll learn in this 90-minute video:

• How Japanese real estate listings work
• That a "mansion" in Japan is anything BUT a house
• How well GitHub Copilot's new agent mode performs with GPT 4.1
• How amazed I am by a computer doing what I tell it to do
• What it looks like for Copilot to get sleepy and decide to call it a day
• That the code GitHub wrote is now available on GitHub
• Update: that they already fixed the issue that caused Copilot to take a shit at the end

Oh, and if you're an AI skeptic, you might appreciate that this screencast has ZERO cuts. It's one continuous unedited vibe party, and you're invited.

Don't have 90 minutes? Check the chapters—most of the real action happens in 30 minutes in the middle. But I hope you'll pour a beverage and sit down and vibe with me for this one. ☮️

I realize I'm a year late to dishing takes on Shogun, but since people keep recommending it, I thought I'd offer my 2¢ on a real problem I have with how it deals with spoken languages (and something I haven't heard anyone talk about anywhere else)

Mailbag day! This one from v31 of the podcast.

After 15 years of sugar-coating it, I'm upgrading from kid gloves to… regular gloves, I guess?

The truth is and has always been:

1. That the best way to get better at programming computers is to spend time programming computers
2. That a lot of people don't do this, but expect the same salaries as those that do
3. That one highly-competent developer can often run laps around entire software teams at the typical company

I'm done pretending this isn't true.

…the answer usually depends on everything OTHER than whether you're good at your job.

This clip is from v30 of Breaking Change.

This is the first of what I hope will become a habit of long-form video excerpts from the podcast. This one comes from a section in v30 about DeepSeek and the ramifications it may have for OpenAI and the extent to which it condemns Sam Altman's ideology on how to run a startup.

It was probably easy to miss this one, but v27 of my Breaking Change podcast was a holiday special that featured our friend Aaron "tenderlove" Patterson and which we released as a video on YouTube.

In this video, we sweat the small stuff to review this year in puns, re-ranking all 26 that he'd written this year for the show and ultimately arriving at something of a Grand Unifying Theory of what makes for a good pun.

(One programming note that's kind of interesting: while we recorded locally, we conducted the actual session via a FaceTime Video call and… I'll be damned if it doesn't seem extremely low latency compared to most remote-cohost podcasts and videos you see out there. We were able to talk over each other pretty frequently without it becoming unintelligible in post. Neat!)

(Another programming note that's marginally less interesting: YouTube flagged this video as potential climate change misinformation, which it most definitely is.)

It is finished. As mentioned elsewhere, I gave my final conference presentation at Rails World 2024 in Toronto back in September.

The tremendous organizers did me a solid by humoring my request to provide the audio and video feeds they recorded of my talk, which allowed me to create my own edit in the same basic style I've used since discovering screen recording. You can view it on YouTube if you want.

Why watch this one when the official video is also on YouTube? Well, here's what the very exclusive and deluxe and never-before-seen Searls Cut gets you:

• No obstruction, hiding, or movement of the slides themselves—they're the star of the show, not me
• Myself off to the side (where I belong), manually center-tracked with minimal movements to keep me in frame
• Composed in 4K, with slides upscaled to ~1440p
• Same great audio track. I kept in all the umm's and uhh's to humanize me and also because I'm too lazy to bother fixing them
• Native software capture of my slides using Screenflow (as opposed to the conference's HDMI capture)
• Manual removal of the dreaded macOS green dot
• Gently-blurred wide-angle shot as the background instead of cutting between the two video feeds
• Correction of a slide transition where I missed a click on the remote (you can guess, but I'll never tell you which one)

Anyway, if you haven't seen the talk yet, I hope you'll give it a watch. The presentation summarizes a year of my work but it also embeds countless little things life taught me over the 15 years since I started speaking at user groups and regional conferences.

But this chapter of my life has now concluded. I'm excited to be moving on to other things. In the meantime, you can stay tuned to my podcast and subscribe to my newsletter while I get to work.

Since starting Breaking Change earlier this year, I've wanted to start publishing stock-video-laden clips of my favorite little rants and flourishes. The trouble is, video is a huge time sink. At least for me, relative to the other things I do. So I pulled out a timer and ran an experiment to see if I could turn around a ~3 minute clip in under an hour.

And I succeeded! I think with some template setup work, I could get it down to even less time. Hopefully this means more video shenanigans in the future.

I made a video from the random live photos and video snippets I assembled after visiting the PayPay Dome in Fukuoka for a Softbank SeaHawks game on May 21st.

Enjoy.

(Also, before anyone writes in: users don't get to pick out the thumbnail images for their Shorts, because Youtube knows it can maximize engagement by just scanning your video for any stills of girls. Neat.)

All aboard the Persona Polar Express! Next stop, Uncanny Valley!

I had a lot of fun creating this video about what I've learned from speaking at conferences (YouTube | blog post).

Whether or not you have aspirations of speaking yourself, I hope you'll enjoy the behind-the-scenes look into my process. 💚

I love writing scripts against the ChatGPT API that are intentionally constrained to particular input and output formats. Here's one I wrote this morning to help me write spreadsheets with a REPL-like interface to continuously improve the output until I'm happy with it.

Shortcuts is an underappreciated way to wire up Apple products with third party APIs without writing custom code. I've found can muck with request headers and satisfy whatever authorization scheme most APIs expect, too.

Within 30 minutes of getting API access to GPT-4, I'd wired it up to the Action Button on my Apple Watch for Japanese language practice.

I started a screencast series last week, and I can call it a series now because there's a second one.