The promise of Breaking Change is that with every major version (this being the 60th episode of the program, but only the 53rd such release), I will break something. Well, I finally did it. I think I broke the show. Find out how by listening for...
Update on using MacBook Neo for dev tasks: this is definitely a device you should to reboot daily, even without Chrome or any Electron apps.
This morning, Zed (a native Rust app) was 500% CPU without a single window open, 3 macOS internal daemons had...
Using MacBook Neo for "real" development work and it's simultaneously juggling:
• Running Claude Code in three tabs at once
• Compiling multi-package Xcode builds
• Automating two iOS Simulators and a Mac build of my app
Hasn't missed...
Claude sure goes down a lot for being a product targeting businesses.
Maybe Anthropic is just doing humanity a solid by helping us understand how much it will suck when the era of subsidized pricing for LLM-based products ends.
Sent the DLSS 5 video to some not-very-online friends and they thought it looked incredible. Shame on them for not knowing that real gamers are supposed to be outraged! (I happen to think it looks slick as hell and am looking forward to using it.)...
Quincy Larson over at freeCodeCamp had me on their podcast to discuss how the rapidly changing software industry is impacting junior developers and what they can do about it. I don't normally spend time talking about this stuff, because I started...
We blew past this milestone without much fanfare, but it bears repeating: building awareness & goodwill by releasing open source no longer makes strategic sense for many companies. Agents increasingly consume & adapt OSS—often without...
Ben Thompson's latest rests on a single load-bearing assumption: that the harness and the model are tightly coupled, the way Apple's hardware and software are.
It follows, then, that if agents require integration between model and harness,...
Marathon is the first PVP-heavy game to get its hooks in me since… Unreal Tournament in 2001?
I fucking suck, but even playing solo it can feel incredible when you do manage to come out on top. If you want to roll with me, I'm Searls#2430 (or just...
This one goes out to all the testing neophytes who only recently realized that it's useful to have an automated means of verifying their code does what it claims to do.
For the last month, I've been working on prove_it, a framework for...
Pro-tip to any devs who only discovered TDD thanks to coding agents: refactoring is inherently directional. It's more like prefactoring—you rearrange code to make the next change easy. That means you (and your agent) should know the next planned...
I'm still iterating on my experimental Claude Code verification harness, prove_it. This week my focus has been on nudging agents to practice test-driven development. Traditionally, we called this "TDD", but which has recently been renamed...
Claude's electron app for macOS is such a buggy mess that I've uninstalled it and sequestered it to a Safari tab, just like I did to Slack, Discord, etc., six years ago. ChatGPT is a native app where things like the stop button actually work...
I'll admit, it's hard not to get frustrated by all the posts about coding agents going viral lately that are saying the same shit I've been blogging and podcasting for multiple years at this point with little to no fanfare....
GPT 4.1? In 2026? Is the State Department on a budget? Is every diplomatic cable just going to have 300 space-delimited emdashes now? https://www.nextgov.com/acquisition/2026/03/state-offloads-claude-underpinning-model-flagship-statechat/412022/
Want to buy Becky's iPhone Air? I can vouch that it's in excellent condition (unlocked, $300 less than MSRP). Buy here on Swappa or send me your best offer [email protected] https://swappa.com/listing/view/LACO93287
Barely a month goes by where I can check the news without being reminded of the Cracked skit about two comedy writers who were hired to run a fake Donald Trump campaign and then accidentally got him elected https://youtu.be/8qIQbydyHwc?si=COq7K0HLng3LHI1R
Today, we're joined by a very special guest, Len Testa! You might know him from The Disney Dish podcast or from his excellent theme park travel planning app Touring Plans. Or you might not know him at all! No wrong answers.
This episode is all...
For the extremely narrow Venn diagram of people who love both I Think You Should Leave and Disco Elysium, this is amazing https://youtu.be/6OKgdTXLWIg?si=RyS95BqxFBSEVwyQ
If Nothing releases a folding phone, they should absolutely name it the Nothing Burger https://www.pcmag.com/news/hands-on-nothing-4a-series-midrange-phones
Dear prospective sponsors,
You can't pay me to say nice things about you, but you CAN pay me to not say mean things about you. Please send inquiries to [email protected]...
A lot of people were confused by Trump's immediate embrace of Saudi Arabia and the other Gulf states, but it makes a lot more sense when you consider that he thinks they spell it "Golf"
Can confirm, Claude Cowork is very cool, but if you actually use it for more than a few minutes it'll bring any Mac to its knees. Entire system becomes unresponsive eventually. https://github.com/anthropics/claude-code/issues/22543
The M4 iPad Air is notable, because it will set Apple apart as the only company to release a product in 2026 sporting more RAM for the same price as the outgoing model. (8GB to 12GB) https://www.macrumors.com/2026/03/02/apple-announces-ipad-air-with-m4-chip/
Sure feels like some combination of AI, the US military, and the AI military could bring an end to the world any day now, so I figured I'd better record one last show for posterity. Welcome me on this version's speedrun to the apocalypse!
So...
Was always so inspired by that speech where JFK said "we will go to the moon... or do some other stuff instead" https://www.theguardian.com/science/2026/feb/27/nasa-changes-delays-moon-missions
Finally got a chance to visit @izukiorlando with @instantsal this week.
Generally, I set my expectations for sushi in Florida to quite low, especially if the restaurant isn't Japanese-owned. Izuki blew away those admittedly reserved expectations...
The new Star Trek Voyager game is such a thorough simulation that I think it's convinced me to abandon my dreams of joining Starfleet when I grow up. Being a captain is no fun at all!
The first time I visited GitHub's HQ2 in 2012, they had a TV showing off their first animations of Mona and were using it to push their new tagline: Social Coding. The phrase certainly captured the moment we were living in, so in 2014, I borrowed...
If you've posted an issue to any of my repos and I never responded: GitHub apparently disabled automatic watching for new repositories last year and nobody noticed. Issues are now graveyards from Day 1.
No wonder my inbox has been so peaceful!...
My coding agent harnesses are designed to enable parallel serial work—multiple agents running in multiple tabs, all committing to main instead of worktrees.
turbocommit does this by linking each session's commits:...
Ever hear of the Cart of Thesius? It's when every single item you ordered from Instacart has been replaced with something else. It's a thought experiment: are the bags on your porch still your order?
Their customer support contends that it is....
Speaking of me nailing predictions, I feel like this one from 2023 is aging like wine. Not least of which because the decades-long debate over "10x developers" has been rendered moot by agents....
Are they sure it was the Citrini Research sci-fi substack that caused the crash and not my much tighter, more grounded "Brace for the Fuckening" post written three days prior?
Who's to say. https://apple.news/AaV5OdPkJR12OiHpdMqOdYQ
Common feedback when I write about AI: "if job loss is a real risk, then tell people WTF to do beyond, 'generate more revenue than you cost your employer!'"
Answer: if there were a process or playbook to make yourself irreplaceable, AI...
Over and over again, agents like Claude Code exhibit relentless productivity, exhaustive research, and mind-bending speed. But there's one thing they constitutionally lack: innovative solutions in the face of apparent obstacles.
Humans who lack...
A VP of Engineering wrote in to say my latest essay, "is fantastic and terrifying and absolutely necessary for us to be having conversations about." And that most people still, "have no idea what we're actually on the precipice...
Just had a really clever idea, but because I now work with 5 instances of Claude instead of actual humans—and knowing they'll all forget how smart I just was in a median 125k tokens from now—it feels kind of hollow.
So FYI, I just had a great...
It's been a weird experience enabling turbocommit on my repos and watching it do a better job titling commits than I ever do. That it preserves my agent transcripts in the commit message alongside code changes is really nice!...
It was only once I read Andrew Yang's "The End of the Office" post the other day that I realized how few political leaders are seriously grappling with this question: what will happen to civilization if all this AI investment actually pays...
Honestly after 15 years of hiring reporters to write but not report, maybe it's time we tried hiring them to report but not write https://www.cleveland.com/news/2026/02/journalism-schools-are-teaching-fear-of-the-future-letter-from-the-editor.html
If you're with me on team #CommitsDontMatter, then you might be interested in this Claude hook that just auto-commits after every turn with the title coming from Claude's result and the body containing the originating prompt...
One thing I haven't heard many people talking about is that coding agents can more or less cure RSI after decades of mashing out programs by hand.
Not saying programmers will be remembered as coal miners or 9/11 heroes or anything, but I'm not...
Pointed Claude Cowork at two Japanese lease applications in Excel. It asked me a few questions, then filled both perfectly. (Better than I could—when I did the same via Excel for macOS, I broke the lookup formulas.)
Welcome to the identity crisis,...
Bought a bag of pears 3 weeks ago and they're still hard as rocks. Really looking forward to that 8 hour window next week when they all simultaneously ripen before rotting the following day.
At first glance, it looks like Apple has provided podcasts hosts no documented way to host video for their podcasts except through a third-party hosting provider. Boo. https://podcasters.apple.com/support/5593-how-to-publish-video
I feel like I'm getting saltier, and it's concerning. If you want me to tone it down and/or up, let me know. It'll go a lot better if you just write to [email protected] instead of yelling at your phone.
As usual, I brought the goods. Now...
Enjoyed Dave's post coining Deterministic Core, Agentic Shell. Functions aren't going anywhere, but imperative code is shifting from if/else to an LLM's decisions.
Lotta folks losing the thread here....
One month left on Anthropic CEO's prediction that, "in 12 months, we may be in a world where AI is writing essentially all of the code." Myself and the folks I talk to have already been there for awhile....
One year into living with Tesla Full-Self Driving and: it's good.
The improvement from v13 to v14 is remarkable. Tap Start and it pulls out of the garage, drives you, and parks itself. Another tap and it drives home, opens garage, parks itself....
I'm developing apps for Apple platforms for the first time in 16 years so stop me if this is nuts, but the best feedback loops I've managed are when I've made the Mac build the primary one. No simulator jank. No waiting on devices.
Claude...
PSA: iPhone Air's microphone is located on the left side of the bottom edge of the device, AKA where a right-handed person's pinky would naturally rest when gripping the phone with one hand.
Anyway, that's why all your videos sound like...
Added a /prove skill to my prove_it library. It pushes Claude Code to go beyond merely analyzing source and running tests to demonstrate the code is working.
Example: it spun up a complete example project and ACTUALLY proved my cached test runner...
One happy accident of the fact that Claude Code Opus 4.6 turns seem to reliably take 3-5 minutes in my experience is that it's proving to be the perfect companion to a strength training workout.
I just write prompts between sets.
Free idea: hyperbolic_links.
Let users create symbolic links mapped to HTTP resources. Since you can't literally link a file to a URL without changing the actual file system, hyperbolic would centrally handle journaling, cache/etag, updates.
We ordered a wood chest that arrived with cosmetic damage. After logging the damage in their customer support interface, it prompted me to start a chat with their AI virtual assistant.
What happened next:
It immediately offered me a 15% refund to...
My prove_it CLI is already proving valuable. Just watched Opus 4.6 complete an hourlong rewrite without me… ONLY because 6 times prove_it's own script & agent hooks blocked Claude from prematurely declaring victory. It's letting me focus...
Mitchell Hashimoto, founder of Hashicorp and, more recently, Ghostty in a post on his relationship with AI coding:
Instead of giving up, I forced myself to reproduce all my manual commits with agentic ones. I literally did the work twice. I'd do...
My friend Eric Doggett became a Disney Fine Artist a couple years back and he's currently being featured at EPCOT's 2026 Festival of the Arts. Each day this week, he's holding court to talk to people about his work at a pop-up gallery just...
Elon has combined 3 of his 4 businesses and everything makes sense. Also, I went to Japan and all I came back with was another weird story about animal sperm. Other stuff happened too, but let's be honest, it's the typical AI schlock...
"3 things: reusable rockets, AI via satellite; and a real-time free speech platform. Rockets, space-based AI, and free speech. Rockets, AI…are you getting it? These are not three separate companies, this is one company, and we're calling it...
A recurring theme in coverage on the effect AI will have on wealth/income inequality suggests there's a strong case to be made that AI is going to be bad for poor people.
I have nothing useful to add to that discussion, so here's a word I just...
Whether or not the Xcode agent is any good (I'm dubious), the fact Xcode itself is exposing first-party MCP tools is great news—trying to get Claude/Codex to do fucking anything right in iOS is agonizing....
Glad to see Jerod properly follow up on this one:
In September of last year, I covered a post by Mike Judge arguing that AI coding claims don't add up, in which he asked this question:
If so many developers are so extraordinarily productive...
Two new searlsbrew projects today:
prove_it: verification-oriented baseline rules and hooks for Claude Code https://github.com/searlsco/prove_it
scrapple: scrapes and indexes Apple's SDK docs/videos/sample code https://github.com/searlsco/scrapple
For however expensive LLM inference supposedly is, OpenAI continues to be stupidly stingy with respect to web searches—even though any GPT 5.2 Auto request (the default) is extremely likely to be wrong unless the user intervenes by enabling web...
Early in my career, I met a few COBOL developers who came out of retirement in the run-up to January 2000, getting paid $300+ per hour to remediate Y2K bugs when nobody else was left who knew COBOL.
Suspect a similar trajectory for highly-skilled,...
Coding agents are a TON of fun if you are (1) extremely ambitious and (2) have middling standards.
They are still fun if you are (1) extremely ambitious and (2) ruthlessly exacting standards, but markedly less so.
Now that I've spent ten hours with Claude Code after a few months with Codex CLI, I can say with confidence:
Claude is much much faster
Claude makes much stupider mistakes much more often, even with Opus 4.5
With either agent, I end each session...
LOL, also apparently Adobe Premiere on iPad will frequently silently fail when generating captions. And there's no way to export them as text, subtitle files, etc.
This is starting to seem like bad software that nevertheless gets recommended to...
Trying Adobe Premiere for the first time since version 6.0 in 2002. Paid for Creative Cloud Pro. First thing I tried: start a project on iPad, sync via cloud, finish on my Mac.
LOL, nope. Their "cloud" can't sync projects. It's just a...
I visited 蕎麦 鷹乃 新静岡セノバ on January 28, 2026. I gave it a 3.2 on Tabelog.
Shizuoka's Kaki-age is like Japan's answer to Outback Steakhouse's Bloomin' Onion.
Over the last 20 years, my time in Japan felt like Season 1 of Pluribus. A nation acting in harmony, going from happy to see me everywhere I went to constantly signaling they "need some space"
In cities, it's thanks to over-tourism. In the...
Tooling for coding agents is overly focused on scaling numerous parallel workers instead of ensuring correctness. That continues to be where all my time goes and is the real barrier to scaling up. (e.g., Why am I exploratory testing this UI when...
This Opus 4.5 failure mode rarely happens with Codex: Claude copy-pasted a source listing by mistake, edited that instead of the correct file, declared success. When I pointed out its error, it confidently asserted: "The code is correct - this is...
"Brakeman, hosted on GitHub under the stewardship of developer Justin Searls—known online as presidentbeef"
Pump the brakes, man. https://www.webpronews.com/brakemans-static-vigilance-securing-ruby-on-rails-from-code-to-cloud/
Is anyone really surprised that Entropy fell apart over time? https://www.theblock.co/post/386942/entropy-a16z-backed-decentralized-custody-startup-is-winding-down-and-returning-capital-to-investors
After hearing lots of Opus 4.5 hype, I decided to switch to Claude Code this month from Codex CLI (with which I have a separate set of frustrations), and I've learned two things:
Claude Code is significantly more productive (parallel by default)...
Commuter trains in Tokyo are so packed that the tip of my iPhone triggered proximity contact sharing via AirDrop with the passenger next to me.
I didn't want to admit it was a mistake, so we're getting lunch together now. Will probably be in...
Given that it's just as often programmers who have no clue what to build with the coding capacity offered by AI agents, I WOULD say the issue is that many people are horribly uncreative. https://simonwillison.net/2026/Jan/24/jasmine-sun/
Twenty years working on Windows command line tools? Glad the dude is retiring but I'm sure he's a shell of his former self. https://tech.slashdot.org/story/26/01/23/1915259/powershell-architect-retires-after-decades-at-the-prompt