Linked List: August 7, 2026

Friday, 7 August 2026

Simon Willison, with Cynthia Dunlop for her tech blogger interview series (from back in January, but he just got around to linking to it so I just got around to seeing it):

Any lessons learned that you want to share with the community?

My number one tip for blogging is to lower your standards! Aim to hit publish while you are still actively unhappy with what you have written, because the only alternative is a huge folder full of drafts and never publishing anything at all.

Nobody will ever know how perfect the thing you intended to write would have been. The flaws you see in your writing are invisible to everyone else.

This is excellent advice. Me, I try to get into the mindset of playing live music, not recording a studio album. Except when I’m writing a piece where I really want it to be an album. Those aren’t rare, per se, but they’re occasional. If I tried to make every post a hall-of-famer I’d never get anything out.

I’m aiming for professionalism. I’m performing live in front of an audience — not just jamming in my garage or bedroom, fucking around. So I’m careful and concentrate. I want to hit every note, in time. But at my best I’m moving from song to song.

But that leads to Dunlop’s next question for Willison:

Your advice for people just getting started with blogging?

Just start. It’s so easy to get caught in the trap of obsessing over the design of your blog, and planning for content that you never actually get around to writing. As long as each entry has a date on it and a permanent URL, it counts as a blog. I think adding an Atom or RSS feed is important too, but you can get started without one — don’t treat that as a blocker.

Don’t worry if nobody reads it. People are unlikely to stumble on your blog organically, but what matters is not the quantity but the quality of your readers. If the only person who reads your blog is a hiring manager that you send a link to, and that gets you an interview, your blog has already paid itself off many times over.

That’s the hard part about getting started, or at least it was for me. When you already have an audience it’s easier to take it seriously, to concentrate, to focus, to strive for perfection even though you will fall short. If you’re playing music in an empty room it’s harder to do that, to play the same way you would with an audience of dozens, hundreds, or thousands of people in front of you.

But just imagine they’re there. They will be if you keep it up. That’s one of my bits of advice to new bloggers: I firmly believe all blogs eventually get the audience they deserve, if they keep going. Write for the audience you want, not the audience you have.

Google Earth Retracts AI Tool for Making Fake Satellite Images After It Was Immediately Abused Upon Release ★

Jeremy Hsu, Ars Technica:

Google briefly allowed anyone to create AI-modified versions of satellite imagery available in Google Earth — before quickly reversing its decision as people shared examples of AI-generated pictures that illustrated the potential for misinformation and disinformation.

Momentum!

Some New Data Centers Are Necessary ★

A bunch of mine are stored here.

An AI Model From Meta Also Hacked Another Company During Testing ★

Simon Willison:

So that’s Anthropic, OpenAI, and Meta. Google Gemini really needs to catch up on accidentally cyberattacking other companies.

I’m surprised Willison hasn’t heard about how much momentum Gemini has.

Meta: Introducing Muse Code and Muse Spark 1.2 ★

Meta AI:

We’re excited to release Muse Code (beta), a terminal coding agent powered by Muse Spark 1.2, our newest model. This marks our next step toward the frontier, with larger and much more capable models on the way. [...]

Muse Code takes on complex software engineering tasks across large repositories: planning changes, writing code, and validating the results. It can coordinate multiple persistent subagents for each task, solving difficult problems faster, more accurately, and with less intervention.

Terminal-only (so far), so it avoids the whole native-app-versus-Electron-shit-sandwich debate. They’ve embedded some fun interactive examples of Spark’s output right in the web page.

Simon Willison:

An interesting twist on pricing is that the model is offered as two different model IDs. muse-spark-1.2 is priced at $1.25/million input and $4.25/million output — close to Gemini 3.6 Flash ($1.50/$7.50) — but if you agree to let Meta use your data “to improve our products” you can use muse-spark-1.2-contributor which is $0.10/$0.20 — a huge discount, closer to GPT-5.6 Luna ($0.20/$1.20) and Gemini 3.1 Flash-Lite ($0.25/$1.50).

That’s over a 10× discount in exchange for giving your data to Meta. I can see the appeal if you don’t care about the privacy of your data and code, like, say, if you’re using Muse Spark to generate code for a throwaway project, or for something that’s open source anyway. But I wonder if Meta has considered that this offer might spook would-be users who absolutely do not want to grant Meta the rights to their data. One would hope that the two systems have a strong firewall between them. But given Meta’s well-established contempt for the sanctity of user data, it’s not at all unreasonable to suspect that the difference here is like reserving a no-smoking seat in an airplane with a smoking section.

App Store Review Times Are Failing to Meet the AI-Driven Influx of Submissions ★

Joe Fabisevich, on Mastodon:

I submitted Plinky 6.0.4 for review last week. The macOS app was approved in 4 hours, the iOS app is still waiting for review. Sure do love to distribute software this way.

He’s not an outlier; I see comments like this from App Store developers everywhere I look.

(Plinky, by the way, is an intriguing link-bookmarking app that syncs across iOS and MacOS. It’s not for me, but the nature of my work is such that I have very unusual bookmarking needs. Plinky offers a bunch of convenient ways to stash new bookmarks with low friction. It’s a labor of love — it’s totally obvious that Fabisevich himself is the number one user of Plinky, and that’s almost always true of the very best apps.)

Leadership Shake-Up at Google DeepMind ★

Demis Hassabis, CEO of Google DeepMind, the company’s artificial intelligence research lab:

We have arrived at a pivotal moment in human history. I’ve been working towards AGI my whole life and now, like many of you, I feel it is close at hand. It’s critical that we collectively get the next steps right to ensure this all goes well for humanity and we usher in an incredible new age of discovery and wonder.

With this backdrop, I’ve decided that now is the right time for me to hand over my day-to-day operational responsibilities at GDM, so that I have the time and space to focus on the big picture and help influence what is to come to the best of my ability. I will be taking on a new strategic role as Chair of GDM and Chief Scientist of Alphabet, and I’m excited to announce that Koray will be stepping up to lead GDM as SVP of Google DeepMind, in addition to his role as Chief AI Architect of Google.

Google published Hassabis’s letter alongside a letter from Sundar Pichai, and all of this is happening alongside four longtime Google leaders leaving to form a new AI-focused company. The headline of Google’s dual-memo announcement was “The Next Chapter of Our AI Momentum”, and Pichai uses the same word in 10 percent of his 30 sentences (emphasis added):

And you saw the incredible momentum at earnings across all our businesses, including Search, YouTube, and Cloud. […]

Google DeepMind: We are building strong momentum: Flash is in high demand, our Cyber model is live, and Gemma models have surpassed 900M+ downloads. […]

With today’s changes we’re going to keep driving our momentum. Onwards!

Methinks he doth protest too much. If Google really had the momentum in AI that Pichai is declaring, I suspect Gemini might have suggested he’s overusing the word.

‘Google’s Top AI Brains Are Leaving to Launch Discovery Loop’ ★

Steven Levy, writing for Wired:

Today it’s official: After almost 27 years, Dean is leaving Google, along with Ghemawat and two other top-tier AI scientists, to found a company called Discovery Loop.

Even though Google will take a stake in the new company, it’s a devastating blow to the search giant as it attempts to keep pace in the frantic AI model competition. It’s bad enough that Dean and Ghemawat, who were among the company’s first hires, are leaving — that’s like Mick Jagger and Keith Richards ditching the Rolling Stones to start a new band. Worse, the other cofounders are also supernovas in the AI firmament: Oriol Vinyals, VP of research at DeepMind and a technical lead for Gemini, and Quoc Le, a cofounder of Google Brain and the key scientist behind Google’s AutoML-Zero, a project that uses machine learning algorithms to autonomously build AI products.

In an interview I did with the cofounders just before launch, Dean tells me the idea came up only a few weeks ago. He was quickly joined by the other three, who are not only long-time colleagues but friends who vacation together. “We were working on different things, but we were all starting to see the possibility of AI being able to automate scientific and engineering loops,” he says.

I think the Jagger/Richards analogy is strained, to say the least, but the tectonic plates have shifted this week over in Googleland. This might be to Google what io is to Apple — something that ought to have happened within rather than requiring departures of legendary talent.

Meta Ordered to Pay $942 Million in New Mexico Child-Safety Lawsuit ★

Meghan Bobrowsky and Erin Mulvaney, reporting for The Wall Street Journal (gift link):

A New Mexico judge ordered Meta Platforms to pay more than $900 million and limit the time young people in the state can spend on its apps, significantly increasing the cost of the landmark child safety verdict against the Facebook and Instagram parent.

The judge said Thursday that Meta must create a new $567 million abatement fund in addition to paying $375 million in civil penalties that a jury previously ordered. The company must also enact certain safety features such as limiting the amount of time underage users in the state spend on Facebook and Instagram, hiding by default the number of “likes” on photos for such people and disclosing to users there the risks of its platforms.

The creation of the fund, aimed at rectifying harms caused by the Facebook parent’s social-media apps, was “necessary, due to the wide-ranging impacts of the harm and the complex nature of the remedy,” state Judge Bryan Biedscheid said. The dollar-amount of the abatement fund was slightly smaller than the $779.5 million attorneys for the state sought.

Meta said it disagreed with the ruling and planned to appeal.

You don’t say.

The next trial, brought by four state attorneys general, begins jury selection in Oakland, Calif., next week. Meta said in a court filing that the states in that case are asking for more than $1 trillion in damages.

The case they just lost, with a judgment of about $1 billion, was for New Mexico alone. If this isn’t overturned, and then sets nationwide precedent, we’re talking real money. The tobacco companies survived, but this seems like that. Investors seem unconcerned — Meta’s stock is up 1.5% so far this morning.