Ben Thompson’s article about Anthropic and the Pentagon is worth a read, although not everyone is going to be satisfied with it. Anthropic is an interesting company because it feels like they don’t fully believe in their own product. Like Dario Amodei kind of wishes he was working on something else.

Good morning, Atlanta.

Steve Troughton-Smith blogged about the projects he worked on over the last month with Codex 5.3:

It didn’t just blow away my expectations, it showed me the world has changed: we’ve just undergone a permanent, irreversible abstraction level shift.

Blazers in Atlanta. 🏀

I cancelled my NYT subscription two years ago in part because it felt like they threw Joe Biden under the bus, but they really do have good reporting most of the time. Some background details on the Anthropic / OpenAI and Pentagon negotiations:

Mr. Michael, who was on a call with Anthropic executives, demanded that the company’s chief executive, Dario Amodei, get on the phone to hash out the language, the people said. But Mr. Michael was told that Dr. Amodei was in a meeting with his executive team and needed more time.

Sad to hear about this shooting in Austin. It’s also weird to be out of town when something like this happens:

Three people died, and at least 14 others were injured in a shooting on West Sixth Street overnight Saturday into Sunday, according to Austin-Travis County EMS.

East 6th is more crowded and generally a little crazier, although I wouldn’t consider it unsafe, so perhaps this could’ve been worse. Too many guns.

Centennial Olympic Park.

Strange juxtaposition with the war in Iran happening at the same time as the controversy around Anthropic, OpenAI, and the Pentagon. OpenAI has a post with the details of their agreement, which also includes red lines on mass surveillance and autonomous weapons:

We think our agreement has more guardrails than any previous agreement for classified AI deployments, including Anthropic’s.

I’m confused about how Anthropic’s proposed contract differs from the contract that OpenAI has shared. If the Pentagon offered the same agreement to Anthropic, would they accept it? Lots of questions.

World of Coca‑Cola. 🥤

Aviator Coffee. ☕️

Congress needs to reclaim its rightful role in declaring war. We all know this and yet somehow we’ve let one man control far too much. 🇺🇸

Airport train in Atlanta. Here for the weekend!

Traveling this weekend, going to send details about the RSS beta in the morning. If you signed up on this form, note that I didn’t consider asking for a Micro.blog username… Make sure you use the email address registered on Micro.blog or you won’t be able to access it.

Last day of early voting in Texas. Looking at my sample ballot and got super confused for a minute reading the propositions… A few of them are strange, then realized I was accidentally looking at the Republican side. 🤪

I’ve been critical of Instagram forever, and stopped posting on principle 9 years ago, but still it’s good to see Meta’s new work on alerting parents to a teen’s search about self-harm. The way Meta is handling this seems reasonable. Might’ve saved lives if it was in place years ago.

Open letter from employees of Google and OpenAI in support of Anthropic:

They’re trying to divide each company with fear that the other will give in. That strategy only works if none of us know where the others stand. This letter serves to create shared understanding and solidarity in the face of this pressure from the Department of War.

The leadership of all the AI companies is fascinating to me. Dario Amodei perhaps the most so. I thought his essay Machines of Loving Grace was excellent, but I’ve watched many interviews with him and I sometimes come away kind of depressed about the future.

Catching up on Paul Frazee’s post comparing AT Proto’s decentralization to ActivityPub’s federation and Nostr’s “magical mesh” approach:

Our near-miss similarity to the two common models of decentralization is at least partially why we catch heat from them. We’re really similar, but we introduced changes that remove the legible markers of each technology: multiple app instances in the case of federation, and an absence of servers in the case of magical meshes.

It’s a good read. Most of the confusion in the fediverse about AT Proto is because people judge it based on Mastodon’s architecture.

Love that feeling when a new feature sort of actually works. All downhill from here to the release.

Terry Godier posted on the aftermath of shipping Current:

To not let all of the feedback (both good and bad) alter your ability to think clearly and put one foot in front the other and make a thing that’s true to you again. There’s such a strong pull mentally/emotionally to do more of what people liked, or less of what people didn’t, on the next “thing”

The best products take feedback from everywhere but filter it through the original vision. Otherwise you’ll eventually get a watered down or bloated thing with no uniquely defining purpose.

This upcoming book about Steve Jobs during the NeXT years sounds really good, via John Gruber:

With unprecedented access to unbroadcast footage of Jobs in NeXT meetings, private company documents, and interviews with his closest colleagues, Cain offers the definitive account of how failure transformed a brash wunderkind into a true business genius.

I got my first Mac during those years. Steve was legendary, NeXT machines felt almost mythical, and I’m not sure I ever considered that he would return to Apple. What an extraordinary life.