For those unable to attend, here is the video of my talk about the rsync algorithm: paperswelove.org/2017/video/… #rsync @paperswelovenyc

Claude Mythos is Delusional

A new milestone in automatic formalization: We translated an entire graduate math textbook into Lean using 30K LLM agents. Open-source, large-scale multi-agent inference that actually works > Blueprint Lean: faabian.github.io/algebraic-… > Codebase preprint: github.com/facebookresearch/… 1/7

AI is increasingly changing how we do mathematics. Erdős Problem #650, open for over 60 years, was solved a few weeks ago through a collaboration between human mathematicians, an informal reasoning model (GPT 5.4 Pro @OpenAI) and a formal one (Aristotle @HarmonicMath). 🧵

MCP’s spec glancing over identity binding to sessions leaves me with a bad taste.

We wrote a bit about how we responded to the #gcp outage at @redpandadata redpanda.com/blog/gcp-outage…

Preparing my exit from X and into Bluesky. I can’t take the spam anymore, and of course, the new owner.

3 types of child-locks for sliding deck doors, and they all assume the child can’t reach the lock 🤦🏻‍♂️ #greatdesigns

Simple != Easy

I heard someone call “Merchants of complexity” to some folks who love to add complexity to solve simple problems, instead of solving from first principles. I couldn’t have thought of a better title.

There is probably a very healthy business in helping Azure customers migrate to AWS or GCP because they can’t really grow their business with @Azure’s crappy capacity planning .

AI models are not going to replace people, they are going to enhance them. Execs without full understanding of true model capabilities should stop spreading FUD.

✨ With great power comes great simplicity ✨ If there's one thing that distinguishes Redpanda's #BYOC offering from the rest, it's "data plane atomicity." Simply put, it means that even if your data plane goes down, your application stays up 🤘 That's what true BYOC is all about — better security and reliability for #streamingdata apps, and architectural simplicity for #engineers. Read the full write-up from our Founding Panda @emaxerrno on the #data principles our BYOC is built on and how we're shaping a simpler #cloud-first future👇 redpanda.com/blog/byoc-data-…

Do you want to try @redpandadata serverless for free? Create a cluster in seconds here: cloud.redpanda.com/sign-up #kafka #serverless

Worst mistake a startup can make is to focus on trends or pleasing VCs instead of adding true value to its customers. Usually, what adds true value is not what’s trendy.

I’m not sure you can compare what TCP/IP is to Linux with what k8s is to datacenters. K8s isn’t general purpose. Its design is heavily biased towards running stateless services. Statefulsets are not even close to being useful at running stateful workloads reliably at scale.

I’ve been deeply into Azure the last 4 months. It feels much better than GCP but still below AWS. The only complain is the single Entra ID per tenant, shared across all subscriptions. A single point of failure. In this regard, I prefer the decentralized approach AWS took.

“It’s the only product at a technical level that can be deployed from edge devices, all the way to the largest cloud workloads.” @emaxerrno shares details on @RedpandaData's acquisition of Benthos and delivering an end-to-end streaming data platform — tcrn.ch/4bEpL1z

Applies to companies and people, you are what you do, not what you say you do.