Pinned Tweet
Ever wondered what a hand-drawn Google Maps would be like? Thanks to the USGS historical topo collection you can now find out.
Ian Jeffries retweeted
Coding agents are getting alarmingly good at finding exploits.
But they also make one of the strongest tools for preventing entire classes of bugs accessible to almost anyone: formal verification.
@poteto's new 'create-formal-verification-skill' is really impressive. It:
- Finds high-value properties in your codebase worth proving
- Chooses the appropriate verification tool
- Sets up verification and check that it's proving the intended property
- Generates a reusable skill so future coding agents can maintain the verification
All without you having to become an expert in formal verification.
HOWEVER, there's an important catch.
With tools like TLA+ and Quint, you're often not proving your actual executable code. You're proving a simplified *model* of it.
And when coding agents keep modifying your implementation, that model can drift out of sync. Your proof might still pass, but you're no longer proving the behavior of the code you're shipping.
There are some really interesting approaches to closing this gap:
1. @VictorTaelin's Bend2 lets your agent write executable code and proofs in the same language. It's built with coding agents in mind. The code you're proving is the code you're running, eliminating the separate-model problem.
2. For state-machine-heavy applications, a more experimental approach is to make states and transitions explicit in the actual application code, then verify temporal logic properties directly against that same executable representation. No separate model to maintain. The challenge is making sure the checker itself is trustworthy.
And we're already seeing formal verification being applied in increasingly ambitious ways:
- @Anthropic used Claude to prove no-crash properties for 239 entry points across 123 open-source libraries, working at the compiled RISC-V code level.
- @HarmonicMath just launched Aristotle for formal verification of C software and Verilog hardware designs, with proofs that can be checked on every commit.
- And with @poteto's skill, individual developers can now start applying formal verification to their own projects without years of specialist expertise.
This could fundamentally change how we build reliable software with coding agents.
If you're thinking of adding formal verification to your coding-agent workflow, it's worth understanding the tradeoffs before you start trusting the proofs that pass the check.
We explored all of this in more detail in our latest Superlinear episode.
🧵👇 Links to Discord, YouTube, Apple Podcasts, and Spotify below.
00:00 Coding agents changed formal verification
02:04 Tests vs. formal verification
06:36 Poteto's formal verification skill
08:16 The catch: verified models drift from your code
12:17 General purpose languages like Bend2
14:28 TLA+ and Quint for state-machines
16:27 Build verification into the application
20:38 Statecharts and experimental TypeScript checker
27:27 Closing thoughts and Superlinear Discord
Ian Jeffries retweeted
Retweeting this one, as my current engagement is coming to a close soon! If you want me around to say funny things in your company chat, or make your infra better/more reliable, or your backend more robust, I can help!
Reminder that I'm looking for a new role.
I'm an experienced infrastructure engineer. I've worked with Kubernetes, Terraform, Postgres, AWS, with some Azure and GCP as well.
I've also done a lot of valuable work as a backend engineer, having worked primarily in Haskell and Rust, though I've also gotten my hands dirty with Python, some Go, C++, and a bit of TypeScript.
I've also had the privilege of being a public speaker, and have also lead training seminars for peers on how to use different tools, sharp edges to avoid, and getting them to a functional mental model of how things work quickly, so they can start to ask meaningful questions and debug problems on their own.
I take ownership quickly, care deeply about reliability, correctness, and delivering quality service to customers. I say this up-front because this comes with a number of downsides, including:
* I dislike shoddy work, buggy code, leaky abstractions, or flaky services. "Don't let the perfect be the enemy of the good" is fine and well, but tech debt brings an aggressive drag coefficient to your business, and that fact gets swept under the rug far too often.
* I don't optimize for shipping a ton of code quickly. I tend to be on the lower end of lines of code/sprint, and I'm not terribly keen on changing that. I write code to be maintainable, have clear invariants, be easy to refactor as business needs change, be easy to debug, and to quietly work while you're focus is elsewhere.
I'm also a quick study, and have a long history of getting spun up quickly on the new parts of a stack I encounter at the various roles I've filled over the years.
If this sounds like a good fit, please do drop me a line. I'd love to talk about how I can make your life better.
thewizardtower.github.io/McC…
Ian Jeffries retweeted
I was 8 years old on 9/11. Old enough to remember, not old enough to get it.
And I think many people in my generation struggled to get it because we were children, and almost everyone who died on 9/11 was an adult.
But every year, as I got older and the stories remained the same, the deeper they hit.
I read Todd Beamer’s story again. When I first read it, I must’ve been a child — and of course it’s hard to imagine yourself on flight 93, but as a 10-year-old, how do you relate to a grown man — father, athlete, employee, business traveler, and in the end, hero?
When I was reading the story today, I came across the photo that previously had looked so old to me, and I thought “he looks pretty young”. He was 32 that day, a few months shy of his 33rd birthday — younger than me today. His whole life was still ahead of him.
As so many years have now passed, it dawned on me that so many of the victims then were younger than I now; just starting out in their 20s. I’d see them as kids today. Robbed of life, youth, and the great journey ahead.
Many others are writing that the 25th anniversary is hitting them unusually hard. Part of it is that with a longer arc of hindsight, it is becoming clearer and clearer what we lost. As for me, age adds perspective.
Replying to @michaelspori @usgraphics
Okay, that's legit sexy. We should do something @usgraphics.
Ian Jeffries retweeted
Thank you Guillem
I believe development that focuses on user experience and rich features is what enables this fast pace.
Technically the dev speed is enabled by #Electric from the Hyperfiddle team (led by @dustingetz) and the associated Electric-specific LLM skills.
Yesterday we had the first Incremental Learners community meetup. @adham_clj shared the most important updates of freememo.net.
The development speed is incredibly fast-paced for a solo side project. Just the July updates feel like more than 4 years of a SuperMemo big release!
Ian Jeffries retweeted
Is there some standard for organizing multiple Git repos, so you only have one checkout for each origin URL and different tools know where on the filesystem to find a repo given a URL?
Ian Jeffries retweeted
I'm reaching a point where I want a declarative, GitOps style workflow for DNS for my personal projects.
Anyone have opinions? The obvious options seem to be Terraform or DNSControl.
Ian Jeffries retweeted
File under "the liquefaction of everything"
Imagine every pixel on your screen, streamed live directly from a model. No HTML, no layout engine, no code. Just exactly what you want to see.
@eddiejiao_obj, @drewocarr and I built a prototype to see how this could actually work, and set out to make it real. We're calling it Flipbook. (1/5)
I'm closing in on the beta release of U.S. Flashcards. It's the easiest way of all time[1] to learn classic art, architecture, constellations, and a few other things. If anyone wants to try it out early and check for bugs that would be very appreciated.
[1] I think.
usflashcards.com
/cc the @tan_stack community. TanStack has been incredible for this project and I've tried to make you proud, but more testers definitely could not hurt.
4
123
Ian Jeffries retweeted
I recently made a flashcard for the first building: the HSBC Building in HK, designed by Norman Foster. It's surprising how often my "trivia" flashcards turn up in normal life. I think it's an example of how knowledge changes the structure of attention.
Hot off the wrangler deploy, a number of the best flashcards of all time.[*]
[*] For their respective prompts, not of all possible prompts
The front for each is just text, eg "Chicago World's Fair - year?" Here's the back:
1
4
195
Ian Jeffries retweeted
I have a lot to say about this. The short answer is that if you design the language just so, it becomes trivial to separate the program cleanly from the live object soup. 1/n
Ian Jeffries retweeted
“historically, our batting average was pretty good. I suspect it will be again.”
Lori - always appreciate the interest and commentary!
We never officially moved the timing of Artemis III to 'late' 2027. A reporter wrote that after misinterpreting my quick response to a question during a budget hearing. In the same hearing, I also said we were gaining increased confidence in interoperability tests with both landers in 2027. I am quite sure at least one will incorporate an ECLSS demonstration.
As for 2028, we have always said we would protect for up to two landing attempts, if required. That story has not changed, and the South Pole was always the landing target.
The only goalposts that have moved have been in the direction of achievability--standardizing the architecture, adding missions, focusing resources, and rebuilding in-house competencies. We have tried to communicate to the greatest extent possible in this regard, while respecting the proprietary information of our commercial partners.
I will say, I am very proud of the NASA team and the new pace we are moving at. The public will learn in the weeks ahead which astronauts will undertake Artemis III, and I would not be surprised if you see some early wet dress testing at 39B before the end of this year. Now, if your point is that the schedule is ambitious, I agree. And NASA’s recent track record on schedule has not been great, but when it comes to landing astronauts on the Moon, historically, our batting average was pretty good. I suspect it will be again.
You say you love the bureaucracy of the United States, yet you haven't integrated national parks with a tiled map of USGS 1:250k topos inside your spaced repetition program? Curious.
3
149
Ian Jeffries retweeted
One easy trick to help you learn Roman history is to remember all of the names of the important people and the dates when they did stuff.
Ian Jeffries retweeted
Sellersville, Pennsylvania Post Office
Carrying the Mail - Harry Sternberg (1937)
Photo by Jimmy Emerson
Ian Jeffries retweeted
Replying to @DeltaClimbs
Without significant insulation that crosscuts their power to weight ratio, microturbines are difficult to sonically dampen, and fuel inefficient, low they are widely multifuel capable (important for logistically austere environments, where BTUs may need to be scavenged).
I am partial to free piston linear generators, because of their form factor, noise profile, and potential for efficient, dynamically adaptive multifuel operation. However, the control problem (consistently returning the piston to top dead center with low variation each combustion cycle) is not yet solved for small, mobile FPLGs; all existing commercial offerings are large, stationary installations.
I am making progress, though:







































