AI haters, you can skip this post. I installed Codex CLI to explain the Ghost ActivityPub implementation to me. I can ask it questions about how it parses JSON-LD and handles various fields. Learning a lot!

Build the wall. First time in northwest Austin in a while, more work on the 183 toll road construction.

I’ve posted the first week of prompts for the June photo challenge. 📷

I’m now at the point in ActivityPub debugging where I’m just reading the Ghost source code, trying to understand what it needs. This is always the problem: you get an HTTP 202 for Create activities, then have no idea why nothing happens when the request is later processed.

Just found out from my son that the latest Playdate SDK has networking! This opens up so many things. Looks like a nice API for HTTP from either Lua or C.

As I fix compatibility with Ghost, there are some surprises. For example, Ghost notes don’t appear to exist on the web. No url in ActivityPub, no permalink in the Ghost dashboard. Maybe a temporary limitation.

It’s always been a strength of Micro.blog that short and long posts are the same thing.

The redactions in this internal OpenAI document are hilarious to me. You could turn it into a Mad Libs:

Our long-term growth depends on [something surprising]. For H1, our top focus is [a dramatic goal], but we’re also pursuing [a fun side quest].

Enjoying the reporting from The Verge’s Adi Robertson on the Google antitrust trial, covering potential remedies:

Mehta points out a contradiction: the government wanted to exclude a bunch of other search-engine-like services to establish Google had no meaningful search competitors during the liability trial, and now it wants to add new ones during remedies.

In my conversation with Vladimir Prelovac on yesterday’s episode of Timetable, we also talked about whether a remedy could be sharing the search index with other search engines like Kagi.

From yesterday, at the Texas State Capitol.

Tickets are available to The Talk Show Live from WWDC. No Apple execs as guests this year. John Gruber:

This year I again extended my usual invitation to Apple, but, for the first time since 2015, they declined.

I’m excited about this. It’s good to mix it up and get some different perspectives.

I wrote a draft post about billionaires and the open web a couple months ago, and the couple folks I mentioned it too have told me not to post it. Good advice. So instead I’m working on a long blog post about Sam Altman, a topic which can’t be at all controversial. 🤪

Looked at Threads for the first time in months. I haven’t missed it. I’m glad we have automatic cross-posting to Threads from Micro.blog, but I don’t personally need it and I rarely hear from people who use it. Maybe it works perfectly, or maybe other folks have lost interest in Threads too.

Steve Klabnik blogs about how AI views online have become extreme and frustrating to read:

What is breaking my brain a little bit is that all of the discussion online around AI is so incredibly polarized. This isn’t a “the middle is always right” sort of thing either, to be clear. It’s more that both the pro-AI and anti-AI sides are loudly proclaiming things that are pretty trivially verifiable as not true.

My interview with Kagi founder Vladimir Prelovac is now up! You can listen on the web or subscribe to Timetable wherever you get your podcasts.

In many ways I’m a terrible sys admin. For the last couple years I’ve been dealing with ballooning memory usage in Redis, confused, and only now have finally taken some time to understand it. I’ve been running a cleanup script for the last week that has trimmed memory from 45 GB to 16 GB. Whew!

Federico Viticci provides an early look at the new LLM-powered automation tool Sky:

The real strength of Sky lies in its ability to mix and match the non-deterministic nature of LLMs with the deterministic approach behind scripts, combining the two in a new kind of hybrid automation that is smarter, more flexible, and more accessible.

Looks impressive. The preview right before WWDC seems like significant timing too.

Matt Mullenweg on how a thought goes to idea to writing to blogs:

Once you publish publicly, you open yourself up to the beauty and chaos of the wider world. The best reason to blog is comments, the people who find you and add to your thoughts, who you never would have imagined.

This is probably a dumb, self-inflicted privacy leak, but as an experiment I asked ChatGPT to look at the last 5 months of bank transactions. No major surprises: we spend too much on eating out, cell phone plans, and streaming services.

Bono was great on Jimmy Kimmel last night. I was lucky enough to see his show a few years ago, but still excited to watch the filmed version when it drops on Friday. And hints at a new U2 album! 🎶

Letting go of Core Intuition has created a podcast void in my work. Today, I just posted a new episode of Timetable. Just 2 minutes, remembering how to podcast. Tomorrow I’ll have another new episode — an interview with Vlad Prelovac.

I’m rooting for The Browser Company folks, because it’s good to have competition in browsers, but their messaging has been all out of order. They could’ve quietly maintained Arc without making a big deal about it. Also, wait until Dia is available so people focus on the new stuff. (Posted with Arc.)

The Browser Company has a long post about why they’re working on Dia instead of Arc. On how chatbots and browsers will merge:

Our tabs aren’t expendable, they are our core context. That is why we think the most powerful interface to AI on desktop won’t be a web browser or an AI chat interface — it’ll be both. Like peanut butter and jelly.