Today Microsoft announce "Majorana 2", claiming they will build a “practical” quantum computer by 2029. This is of course massive PR bullshit. I could also speculate about the timing of this announcement, but unfortunately that will have to wait. For now, let's look at their preprint. 🧵 1/x

I stumbled upon this this morning and I've been staring at it for literally multiple hours today https://sandbox-science.com/particle-life

Agreed - but that's on us for not doing a better job of explaining. This is a super article on this topic by @e-astronomer.bsky.social www.science.org/doi/full/10....

With friends at the University of Warwick (in particular, Rocco Caprio and @adriencorenflos.bsky.social), we've recently arXived some work (arxiv.org/abs/2605.30253) on a method for approximate inference known as "Coordinate Ascent Variational Inference", or "CAVI" for short. Let me explain:

okay ❤️ yay ❤️

The trick is, when among mathematicians, to say you are a physicist so that they forgive you for not knowing very much math. Then, when among physicists, say you are a mathematician so that they forgive you for not knowing very much physics. In especially dire cases I might say I am a philosopher.

Always quite jarring for me to see discussions about arXiv which reveal how many people see the moderation as gatekeeping, or otherwise negative. Muddies the waters for me as to what they think the point of the website is.

Teaching LP and Randomized Rounding tomorrow! Excited to explain to my students how to express s-t mincut, a problem we know how to solve efficiently, as an ILP—a problem we don't.

If you are in Oxford and would like to learn about applied category theory, I'm giving a course at the Maths department. It's on Fridays, at 11 am, in L5. We start this week with monoidal categories and string diagrams.

Unpopular opinion: Computer science syllabuses were designed back in a time when working AI was nothing but scifi. A course that does not teach a deep and practical understanding of how LLMs work is no longer representative of computer science, and you need so much material to do so that massive […]

An uncomfortable fact is that it is impossible to write down a translationally invariant Hamiltonian for a lattice many-body system in a classical and constant magnetic field. Infinite PEPS with gauge tensors can counteract this magnetic translation symmetry: arxiv.org/abs/2604.24574

Was re-reading my old MSc thesis for inspiration: "upsettingly heuristic" seems to be the mantra of my scientific career 😅

davidbessis.substack.com/p/the-fall-o... An interesting read, with lots of food for thought. It reminds me of my enthusiasm for "mathematical clean-up work", i.e. going back to the foundations of a topic, making the constants sharper, proofs nicer (and more modern), and the like.

I just published a new blog post! Types and Neural Networks ( https://www.brunogavranovic.com/posts/2026-04-20-types-and-neural-networks.html ) It's about what comes into view once dependent type systems become a part of neural network training.

Fun twist in Google's zero knowledge proof of quantum circuit.

I never wrote about it here, but as of some time ago I figured out a basic implementation for named axes in TensorType: https://github.com/bgavran/TensorType This means that now you're: a) forced to assign some meaning to all your axes)

scirate.com/arxiv/2604.0... I’ll admit I was always skeptical of any QML/classical-data-processing advantages in quantum, but this paper will at least give me pause!

The Moon: oh wow you guys decided to come back Artemis II crew: earth’s haunted

"We [...] introduce a new measure, felinity." arxiv.org/abs/2604.02793

The quantum adiabatic theorem says that a slowly perturbed system stays in its instantaneous ground state. Now researchers have shown that this principle holds for the opposite limit: The ground state remains the most likely state even for a system subjected to an instantaneous perturbation.

Artemis II lifts off

The forward-looking Oratomic time-efficient architecture claims to "potentially enable runtimes of 10 days for ECC–256 with ≈26,000 qubits, and 97 days for RSA–2048 with ≈102,000." This is based on stitching together several gadgets and plausibly assuming they can work together on an atom platform.

We're making such good progress at reducing the number of qubits needed for factoring that we might accidentally overshoot and bring it down to zero, at which point we would have a classical factoring algorithm. arxiv.org/abs/2603.28627

Looks like Oratomic is coming in with a splash!

"our most time-efficient architectures can potentially enable runtimes of 10 days for ECC–256 with ≈ 26,000 qubits, and 97 days for RSA–2048 with ≈ 102,000 qubits" arxiv.org/pdf/2603.28627

More than 1,500 mathematicians are demanding that their field’s most prestigious meeting be moved from the U.S.

hello 👋 i have developed a new tool called blüuümpíi. blüuümpíi will convert every file on your hard drive to a .wav file without asking you, maximize your system volume, and play them on every bluetooth speaker you have ever connected to. if you offer feedback on blüuümpíi i will take legal action

I know I've said this a trillion times but I have to say it periodically to keep from going insane: The right ran a huge, well-funded, & ultimately successful campaign to completely take over political media & the infosphere more generally. This is by far the most important political development...

Charles Bennett and Gilles Brassard have been named the winners of the A.M. Turing Award, one of the highest honors in computing, for their work establishing the foundations of quantum information theory. The award comes with a $1 million prize. www.quantamagazine.org/quantum-cryp...

The APS Meeting in Denver starts in earnest tomorrow and as sort of an experiment to see if this still works, I've resolved to actually cover the conference on social media (like we did in days of yore), so bear with me for what will be a couple days of posting about physics.

Garnet Chan reflects on recent progress using classical heuristics in computational quantum chemistry—and what it means for quantum computing. The lessons he draws can help to steer both classical and quantum approaches in scientifically productive directions. quantumfrontiers.com/2026/03/12/t...

The arXiv is becoming a nonprofit separate from Cornell, and they're looking for a CEO. Please repost to let good candidates know! @dabacon.tachyon.institute, @harrowing.bsky.social, @michaelnielsen.bsky.social jobs.chronicle.com/job/37961678...

New fantastic work from my lovely colleagues at Phasecraft! "Onset of Ergodicity Across Scales on a Digital Quantum Processor" by F Alam, M Crichigno, E Crosson, @sflammia.bsky.social F M Gambetta, M Hunter Gordon, M Kreshchuk, @quantumashley.bsky.social , A Nocera & R A Santos

The distinction I would make is that writing is "thinking" in two simultaneous dimensions. One is the attempt to capture an idea in one's head. The other is that during the act of writing the thinking changes the original idea. When we remove the opportunity for that thinking the output is affected.

Student: Can you explain again what a free module is? Me: Well, it's a S: And don't say "it's a module that doesn't cost anything" Me: Look I know my jokes are bad but I wouldn't go for one that obvious S: OK. So what is it? Me: It's a module that has time to help you right now

“For a long time, the tech industry liked to think of itself as a meritocracy: it rewarded qualities like intelligence, competence, and expertise. But all that barely matters anymore.” Sam Kriss reports on the next generation of AI technologists.

Researchers have shown how the frequencies and bandwidths of individual photons can be tuned over a wide range inside a short length of standard optical fiber. They anticipate that their technique will be useful in future quantum computing and communications networks.

John Baez wants to wrangle one of the most abstract branches of math to better model the natural world. It sounds like a pipe dream. But is it? Read the latest Qualia column:

Masteful sequence of quotes in Mike and Ike which I never noticed before

No greater praise for a textbook has ever been put down in words

Mathematician Georg Cantor was born #OTD in 1845. He is the founder of set theory and developed the concepts of cardinal and transfinite numbers. These shocking results led his former professor Kronecker to call him a "corrupter of youth." Image: Oberwolfach Photo Collection

Absolutely love this! www.youtube.com/watch?v=T4Up... Kudos to the Norwegian Consumer Council for their efforts to fight enshittification! www.forbrukerradet.no/news-in-engl... Read their report here: storage02.forbrukerradet.no/media/2026/0...

A great overview of the evolution of #container technology via #Docker. It also provides a quick peek into the various technology bits that enable containers work. If you want to learn more about how containers work, then consider starting with wizardzines.com/zines/contai... by @b0rk.jvns.ca.

Interesting format for a review article. 24 brief chapters, each offering perspective from a small subset of the 69 authors on a particular topic. quantumfrontiers.com/2026/02/11/q...

My PhD student Wen Fan (fanweneddie.github.io) is looking for summer internships! ✨His research interests are at the intersection of AI and formal methods, and his work with me is exploring how LLMs can be used to build more usable and scalable verification tools.

New blog post! Autodiff through function types: Categorical semantics the ultimate backpropagator https://julesh.com/posts/2026-02-20-categorical-semantics-ultimate-backpropagator.html

Quantum computing is usually framed in complex linear algebra. This paper asks: what is the least you need to add to reversible classical computing to recover quantum computation? www.pnas.org/doi/10.1073/... @manchegobaby.bsky.social

We are at an airport restaurant. A robot just drove up to us with our food, said, "Hi! Here I am!" and then drove away with our food.

What I’m seeing in some of the Epstein revelations about well known scientists who maintained contact after his conviction, is folks who always thought they were the smartest person in any room, sure they could shrewdly navigate a route that gets them what they want without getting dirty themselves.