Quantum computing and proof-of-memory

Learn quantum computing and maintain your memory for a month to earn $100 in BTC.


Andy Matuschak

Apr 10, 2021 • 5 min read

Say you went to your local electronics store and bought the most powerful computer on sale. You could play the latest VR games with smooth framerates. You could train neural nets with millions of parameters. But one of the most fascinating results in computer science is this: Babbage's analytical engine—invented in 1837 and made of rods and gears—could perform all the same operations, if you gave it enough time and spare storage capacity. Generations of new technology have certainly made it practical to run more complex programs. Theoretically speaking, though, almost every computer that's ever been made can perform exactly the same tasks.

I write "almost" because scientists have discovered a new kind of computer: a quantum computer. For the first time since Babbage, these machines expand the theoretical range of operations a computer can perform. Thus far, quantum computers have mostly been confined to theoretician's notebooks and research labs. But over the past decade, a few labs have built real machines based on the principles of quantum computing. Some of these machines are even available via cloud services, as if they were just another AWS API. Many startups now hope to use quantum computers to accelerate tasks like drug discovery.

So if you're interested in the frontiers of technology—and in better understanding the nature of reality—it's a great time to learn quantum computing.

Quantum Country and the mnemonic medium

Quantum computing and quantum mechanics are infamously difficult subjects to learn. Many people associate the subject with paradoxes and mind-bending abstraction. But there's no significant technical barrier to learning the core ideas. A typical computer science undergraduate has the necessary background.

Part of the trouble seems to be that quantum computing involves a lot of new terms, notations, and concepts. They're introduced in rapid succession, too quickly to properly internalize. Then when the text tries to build on that material, you may have trouble understanding because it's hard to keep enough of those foreign ideas loaded into your head simultaneously.

Michael Nielsen and I (Andy Matuschak) created Quantum Country to help readers overcome this problem. It's a small textbook, written in a new mnemonic medium intended to make it easy for people to remember what they read, instead of rapidly forgetting all the but the gist as usual.

The mnemonic medium works through periodic review. The text breaks every few hundred words for some quick review questions. Then a few days later, when you're likely to start forgetting the answers, you'll receive an email prompting you to review some of those questions again. Each time you review, your memory becomes more durable, so you'll next be asked to review after several weeks, then months, and so on, in an efficiently expanding schedule. If you struggle with a question, the schedule contracts so that you can reinforce it more regularly.

It feels almost like auto-pilot: if you spend a few minutes reviewing each time you're asked, you'll almost inevitably end up durably internalizing all the core details of quantum computing. And you'll remember those details as long as you keep reviewing, which will eventually take only a few minutes per year.

Signaling with memory systems

When someone says that they read a particular book last year, you can't assume very much about their understanding of the material. You've probably experienced the frustrating experience of finishing a book, only to find a few months later that you remember practically nothing from it.

On the other hand, if someone says that they read Quantum Country last year and that they're up-to-date on their reviews, that means something very different. You know that they can answer (with high probability) hundreds of fine-grained questions about the details of quantum computing. Of course, that doesn't mean that they have a nuanced understanding of quantum computing or that they can perform creative problem solving. There's more to learning quantum computing than memory. But this person likely knows more than most people who have taken a university quantum computing course.

So reading something written in the mnemonic medium produces an interesting signal. It signals a degree of interest and diligence: staying current with reviews for the first year takes typical readers around 50% of the original reading time. But at least in principle, it also signals that you've deeply internalized a specific set of material, an approximate proof-of-memory.

The signal is only "in principle" because of Goodhart's law: if these systems produce public signals, and those signals confer a benefit, then people will game the system. They'll rush through the reviews without engaging with the questions, or they'll write bots to complete their reviews. Even if they play fair, people may start to stockpile these signals like meaningless trophies, diluting their meaning.

But if we can find some way to preserve the value of these signals, they could be quite useful. A thousand page views doesn't mean much; a thousand "likes" doesn't mean much; but a thousand readers with proof-of-memory means a great deal, particularly if you can see that people you respect are among those readers.

So in the interest of experimentation, let's see what happens with a private signal.

📋 Task: Earn $10 + $100 in BTC

1. Read "Quantum computing for the very curious".

The first essay on Quantum Country, "Quantum computing for the very curious", covers the basic principles of quantum computing. If you are in fact very curious about quantum computing, then read through the essay and answer all the embedded questions.

Enter your email address below once you've finished, and we'll award $10 in Bitcoin to the first 50 and last 50 valid submissions.

2. Then keep up with reviews for at least a month.

Then, for the following month, keep up with the review emails you receive (say: complete them within three days). We'll award $100 in Bitcoin to the first 5 and last 5 of the submitted accounts which persist for a month.

This task is open to readers who create an account on Quantum Country on or after April 9, 2021, and will run until May 15, 2021. Quantum Country is self-graded, but we can tell if reviewing behavior is very unusual. Please don't cheat: you'll just be wasting everyone's time.


PS: If you're interested in writing in the mnemonic medium yourself, I've been developing a platform called Orbit for that purpose. You can check out the preliminary documentation here.‌

🏆 Winners: Quantum Country

One hundred winners received $10 in BTC for reading through the essay and answering all the embedded questions their posts. Ten received $100 in BTC for also completing the review emails for a month. Congratulations to the winners!

Comments

Comments are for members only.
Please subscribe or sign in to join the conversation!