Arjun Narayan

devrel? you mean twitter deployed engineers?

@TristinHopper Not only will it not buy any gratitude there will be a vocal contingent that will be furious for "not letting nature take its course"

One of these days the United States is going to save us all from a planet-destroying asteroid and it won't even buy them five minutes of gratitude from the world community.

@kevinakwok real fumble that they charge $178 rather than 1 share of NVDA

if you are at GTC plz bring me back one of these and i will pay you back

@ByrneHobart Taxing working people to provide infinity dollars for total boomer luxury communism

State capacity tends to increase over time. So: 50 years from now, what current policy will be viewed the way China's one-child policy is viewed today, but worse? (RIP Paul Ehrlich. You... participated. In the discussion. Your work... impacted the world we live in.)

Paul Ehrlich was genuinely one of the worst people of the 20th century. Everything he stood for was wrong; everything he advocated for was evil. x.com/derektmuller/s…

@DeveloperHarris The California coastal commission exists for the sole purpose of ensuring this never happens and that it's all surface parking lot.

all i want is an American coastal city like this

Why don’t they just tokenize the oil in the Middle East and transport it across permissionless financial rails, thereby avoiding the Strait of Hormuz altogether

It's so funny to me that Mary Shelley, her husband, John Polidori, and Lord Byron had a competition to see who could write the best horror story and she wrote fucking Frankenstein. Imagine losing a competition that badly. Imagine just doing a silly little competition with your friend and she basically invents a new genre and creates one of the most famous characters in fiction. Imagine being proud of your little story and then she shares one that people will still read every day in 200 years. Imagine doing a writing competition with your wife and she becomes so recognizable that you'll always be known as Mary Shelley's husband

@srush_nlp I've thought about this tweet every week for the past year. x.com/id_aa_carmack/…

@srush_nlp Seymour Cray was *so* close. The road not taken was just like 3 or 4 design decisions away. I've been meaning to write about this in detail.

Hypothetically, Tensor Cores could have predated most general purpose computing, right? Feels like there should be people creating worlds where GPT-5 exists, but MS Word is wildly impossible.

Ada Palmer on Dwarkesh cancel all my meetings for the afternoon

I’ve been told- very reliably- that if we attack Persia, we will destroy a great empire

@samlambert you're-- you're-- too offline, okay? you've lost transactions.

The people talking themselves hard in blog posts about how they will take over OLTP, OLAP, and everything in between should probably just focus on the OL part first.

Based on the last 10 months of change in coding agents, the SDLC we’ve known will soon be gone. My thoughts on the rise of software factories, agent harnesses, Jujutsu (jj), digital twins, and new programming languages + infra for agentic development: x.com/lennypruss/sta…

Stuck on a cross country flight sitting next to my son, that's a prime five hours for him to run his latest distillation attacks

@nathan84686947 @VictorTaelin 👀

@narayanarjun @VictorTaelin has an epic addition to this coming soon

I'm optimistic that formal verification is the solution to our current situation where LLMs are writing our code and nobody's reading it. Formal methods can give us a world where we write succinct specs and agent-generated code is proven to comply. But we have a long way to go. There are several open challenges that stand between our situation today and that future, but none appear insurmountable. I’ve written a brief overview of what I consider to be the big open problems, and some of the directions that researchers are taking today to address them: from verifying mathematics to building standard libraries of verified code that can be built upon. Here are a few highlights: 1) A Brief History of Formal Verification Verification is fundamentally about understanding what your program can or can’t do, and verifying it with a proof. In order to verify, you must first have a specification that you are verifying your program against. Most of you leverage some formal verification day to day: namely, some of the compiler errors in statically-typed languages like C++ and Java are verification errors. Static type checking is the version of formal verification programmers are most familiar with. Type systems (and related formal verification tools) have gotten quite impressive, and they are becoming a lot more relevant in constraining the behavior of AI coding models. 2) Rust Type checking represents a middle ground for verification. The hard part is choosing the right balance: reject too many good programs and it becomes hard to program in this language as the programmer has to “guess what the type checker will permit”. Recently the language that has brought the most interesting advances from type systems to the real world is Rust. Its ownership type language and associated type checker is known as the “borrow checker”. The borrow checker is conservative, and “fighting with the borrow checker” is part and parcel of everyone’s Rust experience. This gives us the following lesson: we can prove more interesting things, but at a larger burden to the developer. Finding elegant middle points is hard, and Rust represents a real design breakthrough in navigating that tradeoff. 3) Mechanically verified math Recently, groups of mathematical researchers have recently been writing mathematical proofs in a specialized programming language called a proof assistant. This language, LEAN, comes with a powerful type checker capable of certifying complex mathematical proofs. LEAN is exciting, but working in LEAN can be frustrating - because of the nontermination properties of the type checker’s search, such languages rely heavily on programmer annotation. And this is why more complex type systems have stayed relatively academic: the Rust borrow checker sits at a genuinely elegant point in the design space: complex enough to reason about a complex property like memory references, yet simple enough to not need too much extra annotation. But this is a critically important point: Mathematical proofs and type checking aren’t just analogous: they are the literal same task. They are different only in the degree of complexity along two axes: the complexity of the underlying objects, and the complexity of the properties we are proving. 4) There is still a long way to go for proof assistants While the world I describe is exciting, bluntly, we’re not anywhere close to that world yet. Proofs break easily when programs are modified, the standard library of proofs is too small, and specifications seldom capture everything about the program’s behavior. Overall there’s a long way to go before these techniques reach a mainstream programming language with broad adoption. But, AI is a huge accelerant to proof assistants. Much of the energy towards AI-assisted mathematics is coming from AI researchers who see it as a very promising domain for building better reasoning models. Verified math is a domain rich in endless lemmas, statements, and proofs, all of which can be used as “ground truth” - which means we can use them as strong reward signals in our post-training workflows. There are several startups being built by seasoned foundation model researchers - Harmonic, Math Inc - that are based on this premise. I’m no expert here, but it sure seems to me that formally verified code would lead to a clear domain of tasks that have strong verifiable rewards ripe for use in reinforcement learning to build better agents period. I’m excited about the efforts to use verified mathematics in reinforcement learning. But I’d love to see even more experiments in bringing verification to the agentic coding world. This is an exciting time in programming languages and formal methods research. There’s only one way out of the increasingly unwieldy mountain of LLM generated code: We must prove. We will prove.

@harrybrundage Yes. Have you seen the hydro project? An interesting attack on this problem

My experience has been that the formal methods are valuable but that it’s still pretty hard to review the formal model, and it designs into existence a model<->implementation matching problem — it’s really easy to get drift. You get internal consistency for free but you still have to supervise to get correctness

@jacopotagliabue @audreyccheng 👀 wasn't aware of this line of work. Very much appreciated!

@narayanarjun In the OLAP world, "DAG types", pre-post conditionsand general transactionality go a long way (you probably recognize the style by now: arxiv.org/pdf/2602.02335). As for the verification part (latu sensu), check out for example the @audreyccheng's et al work on ADRS.

@RanjitJhala Yes indeed! Check out some links in the post that lead to exactly that modality of LLM copilot assisted proving

@narayanarjun One exciting possibility is that LLMs make it much easier to deal with the borrow checker, and in general, cough up the hints/annotations etc needed for verification. Eg was scared to do any serious refactoring of Rust code but less so now because CC etc has my back!