Quantized, Data, and Money

The Quantized Knowledge Project has a single goal: to increase the rate of progress of humanity by making learning more efficient. In other words, we want to maximize the product of "how many people learn something via QKP" and "how much they learn."

This post talks about how this QKP interacts with your data and your money to achieve that goal.

Your Money

Quantized will use a dual model for deck pricing.

Community decks will work like large open source projects. They'll be hosted on GitHub and free for everyone. Community members will collectively manage pull requests and improve the deck.

Paid decks will be closed source, and maintained by the individual or group that created them. That group may charge a reasonable price for access. Quantized will take a 30% cut of those fees.

Quantized will also sell a small number of addons, like a finely-tuned spacing algorithm. This ensures that we can support a large number of free users without needing to change our model. In other words, access to free decks will be truly free, forever. But, if you want to leverage some of our data to decrease your review count by 10 or 15%, we'll charge you a couple of bucks a month for that. That also means that we don't have an unhealthy incentive to make decks paid.

Why have paid decks at all? Because having cash flow lets you buy things like professional voice talent to voice language decks and lets you dramatically increase the quality of the decks you provide. (Think about the quality of Core6k; that's the kind of quality that's available if you have money.) It also lets us compensate creators whose work appears in our decks, hire professionals where necessary, and more. Put more bluntly, making good content requires money.

The downside is that charging money might stop people who can't afford a deck from using it.

The goal of the QKP is to accelerate human progress by creating lots of knowledgeable people. That means we need high quality decks (costs money) and widely available decks (don't charge much). These two things are in tension, and I think the above is a good balance.

Your Data

There are two types of data that Quantized deals with: spacing data and card data. What's the difference? "You reviewed card #572947 for 3 seconds last Tuesday and got it right" is spacing data, whereas "What's the capital of Honduras // Tegucigalpa" is card data.

Obviously, your spacing data is yours. Spaced repetition platforms should always allow you to download your spacing data and leave at any time, and Quantized is no exception.

For community decks, you can easily download the card data from GitHub.

Paid cards are a bit trickier. If we want to allow a paid deck ecosystem to thrive, we should at least try to nudge people in the direction of actually paying for paid content. That means we should try to keep it on-platform. In other words, we probably won't let you download card data for paid decks.

Instead, if you want to switch platforms, you can swap out the cards for open source alternatives while keeping your spacing data. So perhaps a paid deck has a professional voice actor read a sentence with the word "それ". Then you can download your spacing data and graft it onto an open source deck. So in three months, instead of getting the professionally read それ sentence, you'll get a community sourced それ sentence without audio.

A parting word

I have a thousand hours of Anki under my belt, and I know how important it is that your spacing data not be locked into any given project. If you have any ideas for ways to handle money and data that would make you more comfortable, please drop me a line. Trust around data and money is the foundation for a solid ecosystem.