8 min read3 hours ago

–

Press enter or click to view image in full size

Waldhotel Arosa. A birthplace of quantum games.

10 years ago I was a postdoc. I was supposed to be spending my time writing papers about quantum error correction. Instead, I created and ran a citizen science project.

The seed of this idea came in early 2015 when I attended the NCCR QSIT meeting in Arosa, the annual conference for the consortium that funded my postdoctoral position at the University of Basel. At the time I was working on decoders for topological quantum error correction codes: a time when thinking about anything with less than 1000 qubits was not worth my time. The number of total qubits in the whole world was probably less than 1000 back then, but they were mostly on completely different devices scattered around the world. Getting a device with 5 qubits was a big achievement in those days.

Given this big gap between my work and reality, it was natural that I felt a big disconnect between quantum hardware and quantum software at the time. Though I had proposed a five qubit experiment in a 2015 paper, it was just to test a tiny chunk of a much bigger picture. Given this context, it was no wonder that I almost laughed in the face of the technology transfer officer of the NCCR QSIT when she asked me if my research might spin out a business. The idea seemed absurd!

Nevertheless, it planted a seed. And the discussions at the meeting of how the research of the consortium could better impact the Swiss taxpayers nurtured that seed. So while I was playing a quick puzzle game at the back of the lecture hall, an idea emerged: I could make a game out of quantum error correction!

This was obviously one of the best quantum ideas that anyone has ever had in Arosa. It’s also where Schrödinger came up with the wave equation. I’ll leave you to judge which was better.

Over the next year I pitched for and won funding from the consortium, and 2016 was set to become the year of my citizen science puzzle game. Since it was based on the problem of decoding, and was a puzzle based on a grid of numbers (like Sudoku), I called it Decodoku.

If the 2015 Arosa meeting was the conception of Decodoku, then the 2016 meeting was the birth. By then I had a first prototype which I took and gave to anyone I could for feedback. Soon it was released to the world, first on iOS, then on Android, and then on the web. The apps disappeared long ago. But the web version is still around.

The aim of the game was to solve the decoding problem of quantum error correction: given the hints that we can extract about what errors occurred, how can we try to correct them? In some cases we can find highly accurate algorithms to do this for us. In other cases, heuristics can be just as good. It was one of the latter cases that I turned into a game, allowing the players to become the decoder. Then I invited the players to become scientists themselves by reflecting on what their algorithm was and to share it with the world. This knowledge was then distilled into a new decoding algorithm, which was then implemented into the game as a helpful hint bot. It was also distilled into a paper (though it was a while until I got it on the arXiv).

Of course, Decodoku was not the most historical moment of 2016 for quantum computing. That came when IBM launched their Quantum Experience, allowing anyone with an internet connection to run quantum circuits on a five qubit device. I was initially sceptical, since my 1000 qubit bias made me wonder what use just five qubits could be. But then I remembered that I had proposed an experiment for exactly such a device. By the end of the day, I had done the experiment. The age of cloud quantum computing had begun!

But let’s get back to my reminiscences about long-gone Arosa meetings. I’m sure you are finding it exhilarating!

The highlight of the 2017 Arosa meeting was nothing to do with Decodoku, though I do remember using a break to code up the hint bot and I had a call with one of the participants (it was on Skype: ask your parents). The most notable event was instead the presentation by Project Q. This was the first quantum SDK, created at ETH before Qiskit, Cirq and all the others that we know so well today. Until then, the only way to run circuits on IBM’s cloud quantum computer was via their GUI interface, creating circuits with a drag-and-drop interface. But Project Q showed a live demo of how it could be done purely via Python. Finally we could actually write software for cloud quantum computers.

And so I did. Though the end of 2016 meant the end of the Decodoku project, and that I needed to get back to work on non-citizen science, the idea of combining games with quantum computers did not go away. So I made the first game to run on a quantum computer.

It was just a simple rock-paper-scissors type game. It was made more for the accompanying blog post which explained quantum gates than for the fun of playing the game itself. The same was true for the next quantum game, which was a version of Battleships.

Even so, it was playable even in its lo-fi command line form. I took it to the Quantum Game Cafe held in Aarhus where participants could come and play whenever they wanted. In a limited way, this made it the first game running on a quantum computer to be accessible by the playing public rather than just interested programmers.

I also made another purely educational game, this time in collaboration with IBM.

Those who remember the Qiskit blog may be interested to know that it was in fact created for these posts. By me. A person who didn’t even work for IBM at the time!

I soon did join IBM though, thanks to this collaboration. Once there, my focus shifted from what games could do for quantum (via education and outreach) to what quantum could do for games: the tasks that a quantum computer is good at could actually serve game developers to create interesting new games.

So what can they do? Obviously nothing that has to be done every frame. And personalized quantum jobs for each individual player don’t seem very realistic either, especially in the short term. The idea of ‘running a game on a quantum computer’ is not the way to think about it. But running a quantum job during the game development process, or even during a loading screen, is much more reasonable. This means that we can look to procedural content generation: the algorithmic generation of content for games.

I started looking into this in 2019 and 2020, putting out a couple of papers presented at procedural generation and games conferences.

My first step into this area, Quantum Blur, also found a whole other life as a tool for visual artists. The Berlin-based artist Roman Lipski in particular has now created a vast collection of Quantum Blur art over the last five years. So suddenly I wasn’t just someone who looked at the intersection of quantum and games, but also quantum and art more broadly.

This has continued and strengthened as I left IBM to join MOTH. A company whose founding was inspired by my work, and that of other similar quantum pioneers. Like Spencer Topel and Eduardo Miranda who brought quantum to music. So in some sense, Arosa was one of the birthplaces of MOTH too. If the NCCR QSIT technology transfer officer hadn’t asked me the absurd question of whether my work might become a business, MOTH might not exist. At least in its current form.

At MOTH we’ve moved beyond the limited demos and academic papers of the past. Quantum procedural content generation is now something that everyone can experience with our game Space Moths. For this we build the software infrastructure required to connect the quantum computers of VTT, IQM and IBM with Roblox, the world’s biggest gaming platform. This allows anyone to go and play a racing game whose tracks are procedurally generated on real quantum computers. While players are enjoying one track, our platform is collaborating with the QPUs to create the next one.

This is just our first step into bringing quantum computing to the creative industries. Starting with projects like this during the NISQ era, and scaling up to serving the quantum advantages of the FTQC era to the giants of media and entertainment.

But that’s the story of the next 10 years. It’s not something for this self-indulgent retrospective of the last 10 years, written on the way to the 2026 Arosa meeting†. Heading to the birthplace of Decodoku 10 years on. Back then, I didn’t realize that what I was doing would spark a decade-long endeavour to bring quantum computing to creativity. But transformational things can happen in the Swiss alps. At least for me and Schrödinger.

† Or at least I would be if it was still running. In fact the NCCR SPIN wound up after a successful 12 year run a few years ago, but the meeting lives on in the form of Swiss Quantum Days. It’s no longer tied to Arosa, so I am actually heading to Engelberg. But these are all just details. Everyone knows it’s the same thing!