iiinova — where your digital self keeps evolving

Shout out to Delfosse and the team for shouting this from the abstract earlier this year.

040

I'm used to asking for 10k shots, which is a week for a job. Even if we throw as many qubits as we can at the problem to get parallel factories, seconds per shot is still a day for a job. I've long said that QCs are the tortoise, not the hare. A day for a job makes that very clear!

330

Obviously this depends on architecture for gate times and error rates. If error rates were zero then we'd have no need for the overheads of QEC. But for the first FTQC devices, it seems we are looking at minutes for superconducting qubits and hours for trapped ions. For one shot!

130

James Wootton @decodoku.bsky.social · 03/09/2026

We now have specific architecture plans for fault-tolerant quantum computers. Finally we can ask to within an order of magnitude: how much time will a megaquop take? Assume 10^-6 logical error rate and 1 shot. Basically a Hello World worthy of actually running on a QPU. What's the lower bound?

141

Mathematicians would argue that this is every physics paper. But what's wrong with proof by vibes?

backrooms.mothquantum.com

We use a 120 qubit process on IBM Quantum's new Nighthawk devices to make the levels. So if you want to see what it would be like to get stuck in a horror version of Miami (or the QPU ibm_miami at least) check it out! backrooms.mothquantum.com

Quantum Backrooms | A Quantum Computing Horror Game

Quantum Backrooms is an open-access game, inviting anyone and everyone to explore a virtual world generated by real quantum computers.

000

Horror video game gets its creepiness from a quantum computer

"But what can you do with it?" I hear you cry, for no-one will care a jot if it is useless. Well obviously we will make a game! We used it to generate levels for a backrooms-inspired dungeon crawler. You can read all about it in @newscientist.com www.newscientist.com/article/2528...

newscientist.com

Quantum Backrooms is a horror game in which the player explores eerie rooms. The twist is that the rooms have been generated by a quantum computer

A new way to think about quantum programming

If you want more detail than this thread but less than the paper, I wrote a blog post. mothquantum.substack.com/p/a-new-way-...

mothquantum.substack.com

By James Wootton

We formalize it as a new model of quantum computation (to make it sound very fancy and us sound very clever). We also show look at what conditions need to be met for it to be equivalent to the circuit model, and what the overheads are for our 'proof by hack'.

100

By this I mean building up circuits by an iterative process. A k-local Pauli tomography is taken at every step, and gates are added by declaring how you want those values to change. It's based on similar previous approaches. Notably a thing I did once, and also Motta et al's approach to QITE.

100

Cute Cat Plays Whack-A-Mole Game

I have a few things to plug so I guess this needs to be a thread. First: new paper! What if instead of designing quantum software with H's in boxs and of weird things, what if you just play whack-a-mole with a managable set of variables. Turns out it works! scirate.com/arxiv/2605.2...

static.klipy.com

ALT: Cute Cat Plays Whack-A-Mole Game

242

James Wootton @decodoku.bsky.social · 04/05/2026

It's 10 years since IBM kicked off the quantum cloud revolution. It ruined my ability to care about anything that I couldn't just go off and run. And long may it continue to do so!

170

James Wootton @decodoku.bsky.social · 01/05/2026

Since we have three offices, I guess you are about to get three sales!

James Wootton @decodoku.bsky.social · 10/03/2026

I guess I am for quantum what you are for horses. So if that's not too niche, I'd be down to help out.

010

Reposted by James Wootton

Bob Coecke @coecke.bsky.social · 05/03/2026

Trailer for our concerts at Oxford's Quantum Centenary with "the Kazakh Mozart", virtuoso musician and composer Rakhat-Bi Abdyssagin. youtu.be/4e_Gg-BZwo0

youtu.be

Oxford Quantum Centenary Conference Concerts Highlights: Rakhat-Bi Abdyssagin and Bob Coecke

YouTube video by Oxford Philosophy of Physics

021

James Wootton @decodoku.bsky.social · 13/02/2026

Next week I am hoping for a paper with an algorithm that helps me to proofread bluesky posts, but I think that might be QMA hard.

060

James Wootton @decodoku.bsky.social · 13/02/2026

And then another end-to-end architecture, this time looking at how the best way to use qLDPC codes for the one big application we all know about: breaking current crypto systems. The team at Iceberg Quantum show that it can be done with less that 100k physical qubits. arxiv.org/abs/2602.11457

160

James Wootton @decodoku.bsky.social · 13/02/2026

Two great papers on the arXiv today! One has a title echoing that of Kitaev's "Fault-tolerant quantum computation by anyons" almost 30 years ago. Finally Anasuya Lyons and @benbrown.bsky.social shown how to make it work from encoding to decoding. arxiv.org/abs/2602.11258

2153

Reposted by James Wootton

Zoe Holmes @qzoeholmes.bsky.social · 09/02/2026

We've studied iterative strategies for finding ground states We use the ground state of an easy Hamiltonian to `warm start' finding the ground states for some harder Hamiltonians We derive gradient guarantees that (as for adiabatic methods) depend on the gap and a few other natural properties 🧵👇

081

Reposted by James Wootton

Manuel Rudolph @quantummanuel.bsky.social · 05/02/2026

Did you know you can simulate quantum states with Pauli and Majorana propagation? In short: High-temperature states are provably and practically sparse, and we can use imaginary time evolution to get there starting from the infinite-temperature state. scirate.com/arxiv/2602.0...

2173

James Wootton @decodoku.bsky.social · 04/02/2026

If we are comparing to 1st century Galilean miracle workers, perhaps Majoranas are more like Shimon bar Yochai: Spends years naked in a cave, but expected to suddenly emerge with superpowers.

020

James Wootton @decodoku.bsky.social · 04/02/2026

How do we figure out the circuit required to prepare a given state, and what's the best circuit? With help from MPS, these guys get it down to logarithmic depth. arxiv.org/abs/2602.02698

1101

James Wootton @decodoku.bsky.social · 03/02/2026

Banishing 512 might be easier.

000

James Wootton @decodoku.bsky.social · 02/02/2026

I have 1024 followers! Nobody else follow me now. I like powers of 2.

1110

James Wootton @decodoku.bsky.social · 30/01/2026

It's a 17 qubit rotated surface code.

100

James Wootton @decodoku.bsky.social · 30/01/2026

Logical qubits encoded in the fusion space of non-Abelian anyons in a generalized surface code. Not the most efficient way to build a quantum computer, but definitely the coolest! Quantinuum and co show how it's done on one of their H2 devices. arxiv.org/abs/2601.20956

083

James Wootton @decodoku.bsky.social · 30/01/2026

The background did indeed highlight the experience

040

James Wootton @decodoku.bsky.social · 29/01/2026

Yesterday I did two things: * Wrote a self-indulgent blog post reflecting on the last 10 years in which I went crazy and started doing quantum games. * Held a surface code in my hand. Here I share both with you, the denizens of blue sky open.substack.com/pub/mothquan...

3120

James Wootton @decodoku.bsky.social · 29/01/2026

Today we get the computer scientists' view on the transition into fault-tolerance. There will indeed be a whole lot of classical computing involved in getting early fault-tolerance running, and the better it is, the more quantum we can use. arxiv.org/abs/2601.20247

0122

James Wootton @decodoku.bsky.social · 28/01/2026

Yeah, that would be an interesting and novel test. Especially now that @andreasateth.bsky.social at ETH has finally provided us some loophole free Bell tests.

020

James Wootton @decodoku.bsky.social · 28/01/2026

Always nice to see that quantum foundations have survived the wave of capitalism that has swept through quantum. Today Hausmann and Renner from ETH show that quantum states are more than just a few probability distributions in a trenchcoat. arxiv.org/abs/2601.18872

1181

James Wootton @decodoku.bsky.social · 27/01/2026

In terms of the 'early fault tolerance' naming debate. This seems a bit more on the NISQ with FTQC sprinkles side, but could also help out in the early days of full FTQC. bsky.app/profile/deco...

000

James Wootton @decodoku.bsky.social · 27/01/2026

I got out of the habit of posting half-baked thoughts about papers when Twitter died, But I think I should bring it back! This paper combines error mitigation inside QEC to cancel leading logical errors and effectively giving a distance boost. arxiv.org/abs/2601.18384

2160

Reposted by James Wootton

Dulwich Quantum Computing @dulwichquantum.bsky.social · 23/01/2026

"We are a group of quantum scientists who oppose the ongoing militarization of our societies and the use of quantum research for military purposes." disarmquantum.com Manifesto: arxiv.org/abs/2601.14282 Here's just one example: "Quantum technology for military applications" doi.org/10.1140/epjq...

Sketch of quantum warfare utilising various quantum technology systems

3224

James Wootton @decodoku.bsky.social · 23/01/2026

Quantum computing wouldn't be where it is today without the cryptographic applications of Shor's algorithm. We've been militarized for 30 years, and just pretending otherwise. But I definitely support finding ways to fill up QPUs with more inspirational applications.

070

James Wootton @decodoku.bsky.social · 20/01/2026

Yes, I agree that this approach will definitely be a thing on the journey to fault-tolerance. But it should should it's own term, and "partially fault-tolerant" seems good to me.

020

James Wootton @decodoku.bsky.social · 20/01/2026

That's a definition I can get behind!

000

Is there clear taxonomy that I don't know about. For the noble purpose of avoiding misleading press releases, there probably should be!

220

To me these seem pretty different. One is about finding and correcting all the errors, and so is proper FTQC. Even if it is early days. The other is more about detecting and avoiding errors, and probably paying a sampling overhead to ignore them once you find them. So still a form of mitigation.

150

The other is to have elements of QC interleaved into NISQ. So maybe you have a low distance code mostly (or entirely) doing the job of error correction. Maybe you use a distillation style process to give you a high quality initial state to kick off a NISQ algorithm.

120