Gutter.

Upcoming video social app Neptune will allow hiding follower and like counts:

A key distinguishing feature of Neptune is that it lets creators hide their total followers and likes. This “ghost metrics” feature is optional, however, and is designed to help users avoid the pressures associated with follower count, yet still caters to creators who may want to showcase their metrics.

Good. Might as well go all the way and hide counts by default.

TechCrunch: OpenAI ships GPT-4.1 without a safety report. I don’t want to overreact to this, but as an OpenAI customer it is a little concerning. System cards are fascinating and even if they don’t paint a perfect picture, the transparency is good.

OpenAI working on a social app? From The Verge:

Entering the social media market also puts OpenAI on more of a collision course with Meta, which we’re told is planning to add a social feed to its coming standalone app for its AI assistant. When reports of Meta building a rival to the ChatGPT app first surfaced a couple of months ago, Altman shot back on X again by saying, “ok fine maybe we’ll do a social app.”

Stay in your lane, OpenAI! 🤪 But I think OpenAI and Sam Altman work on a lot of different things, so who knows whether a social app would actually ship.

Meta is a problematic business and more people should move to open platforms, but I have to agree with Ben Thompson’s take on the trial. It’s too late and too complicated to try to unwind the Instagram and WhatsApp acquisitions:

…both of those businesses were what they were in 2020 because of the investments that Facebook made in the intervening years; I think it is just fundamentally wrong to be re-litigating regulatory decisions of this nature years after the fact — and that certainly applies to 2025.

My last post to Instagram was 2017. My photos are now on my own blog where they belong.

I like the attempt at transparency in this statement on AI from Revenant, an animation and effects studio in Scotland, via Cartoon Brew. They’ll use AI for rapid prototypes and streamlining workflows, but:

…we’re also clear on what we don’t use it for — we don’t use AI to shortcut the creative process, we don’t lift style or work unethically, and we always put human creativity first

AI 2027 is a good read. Both endings seem wildly unrealistic, but as a warning it did make me think. So it worked. Just seems greatly accelerated beyond a few years from now.

Thinking about the Tesla Superchargers that have been torched recently… Anyone who uses violence or property destruction to further their cause is actually undermining it. We’ve gotta be better than that.

Protests like the marches a couple weeks ago are the model. Peaceful, legal, effective.

Ran a few tests with the new GPT-4.1 family of models. I think I’ll go 4o → 4.1 for photo keywords and accessibility text, but maybe not immediately. The old pricing and capabilities are still really good. Currently using a mix of 4o, 4o-mini, and 4o-transcribe.

Over the weekend I got to ride in a Waymo for the first time, in Phoenix. See @cheesemaker’s perspective in this video. It’s a surprisingly smooth, nice ride. It drives as good or better than an average human driver. I was expecting to be nervous but after about 30 seconds it feels normal.

I don’t have any expectations for the FTC vs. Meta trial starting today. It seems wild and unlikely that Meta would ever have to spin out Instagram a decade after acquisition. Feels too late, too closely tied up in the Meta empire.

New Mac app for Micro.blog is out today with a couple improvements, including a little interface for editing the auto-generated accessibility text for uploaded photos. Here’s a screenshot:

There’s a pretty significant difference between good AI models and great ones. I sometimes think about how I could integrate local models directly into the Micro.blog apps, but I wouldn’t want it to be worse than using (for example) OpenAI. How many of my users really have a Mac with 24+ GB of RAM?

Dave Winer writes a longer blog post about inbound RSS. The idea is let’s have more systems able to both generate RSS feeds and read them in automatically. If you have that complete loop for posts, you don’t need much else to have a social platform.

Apparently there will be big updates to the design in the next iOS and iPadOS, but I assume it will be mostly superficial. I’d like to see a rethink of text selection and editing. It’s still too finicky without a keyboard and mouse.

Rob Fahrni blogs more thoughts on Dave Winer’s call for inbound RSS:

The problem is the platform folks tend to say “use our API.” Which makes sense, but most API’s are painful in some way because of authentication or some hoop you have to go through. If the platform natively supported inbound RSS it would greatly simplify the developer and user experience.

Micro.blog was designed around inbound RSS. We had that before we had blog hosting. It’s still a unique architecture that I haven’t seen any other platforms replicate.

Back in Austin, down at St. Edward’s for Staple! Amazing that the first show was 20 years ago. Got to meet Kazu Kibuishi and pick up the final Amulet book.

Updated my NBA arenas web page for the game last night in Phoenix. A few good moments in the game but Spurs lost badly, missing too many players even beyond Wemby.

Parker Ortolani blogging about the new ChatGPT feature to look at past conversations:

…it makes the tool dramatically more intelligent — and personal. By being able to reference things you have talked about before without hoping that the model would catch it or by manually teaching it, it feels more like talking with a person than ever before.

In all the talk of AI models and the technical bits, many people miss that ChatGPT is a success because it’s an actual product now. A competitor can’t “catch up” to ChatGPT unless they also build everything that OpenAI has added around the model.

Manu Moreale blogs about taking over blogroll.org and the new design. Love the colors.

Stephon Castle layup, Spurs at Suns. 🏀

Changing Hands Bookstore in Phoenix.

Rob Shearer wrote a detailed and fairly scathing critique of Mastodon. I don’t agree with everything in the post, but I do think he’s right about migration: One of the big selling points of Mastodon was that you can pick which instance your account lives on, but it is easy ...

Stephen Hackett commenting on a report in The York Times about Apple not allocating much of a budget to AI servers:

For a company that says it doesn’t like looking back at its own history, very often, Apple makes decisions like it’s the late 1990s and the company is on the verge of failure. That drives it to make incredible products, but it also means Apple can be incredibly stingy. To play in the AI race, you’ve got to be willing to spend piles and piles of cash.

Old train bridge next to new light rail bridge in Tempe.

Dave Winer has written up two improvements Bluesky can make to RSS feeds: including images and including embedded posts instead of the “contains quote post” message. The images are especially important and make the feeds really impractical to use right now because the data is just missing.

Finished reading: Dragonsteel Prime by Brandon Sanderson. I read the last several chapters on my flight. Fascinating to see such an early take on a few pieces that would later turn up in Way of Kings, and other draft ideas for the Cosmere. 📚

The difference between a junior developer and a senior developer isn’t actually about writing better code. It’s that for people with more experience, their gut feeling about how best to solve something is just correct more often. That’s it.

It wasn’t the Amtrak route I had planned for today, but guess this airport train will have to do. Hi Phoenix. 👋

The new park around the old Mueller tower. Walked by here today and hadn’t seen it since the area was under construction.