Getting Anki

Anki can be found at https://apps.ankiweb.net/. Website screenshot shown below.

The Android app is called AnkiDroid (free) and the iOS app is Anki (it’s $25 one time and goes to support the sole developer working on it)

You can sync between your mobile app and desktop application by using your AnkiWeb account.

The theory behind Anki: what is spaced repetition?

If you learn a fact, your memory of that fact will deteriorate rather quickly. If you review that fact a day, your memory will degrade less slowly. This means that reviewing things at specific increasing intervals will ensure that you always remember that fact. Here’s the cool part – increasing intervals! After each progressive review, your memory will degrade less slowly, which means that eventually, you’ll be reviewing your fact once per year, once per 3 years, or even less, while still retaining the information.

This is why spaced repetition and Anki are different from flashcards. You only review a fact when you’re about to forget it, which means you don’t waste time reviewing cards you already know. Let’s do some napkin math — 

Traditional flashcards vs Anki

Let me break this down, comparing traditional flashcards vs Anki’s spaced repetition system (SRS).

Assumptions:

Each card review takes ~5 seconds

Traditional flashcards: Reviewing all cards daily

Anki: Following typical SRS intervals that grow exponentially

Traditional Flashcards (100 cards):

100 cards × 5 seconds = 500 seconds (8.3 minutes) per day

180 days (6 months, an arbitrary period of time) × 8.3 minutes = 1,494 minutes (≈25 hours total)

Anki (100 cards):

Initial few days: ~100 cards/day (like traditional)

By week 2: ~40 cards/day

By month 1: ~20 cards/day

By month 3: ~10 cards/day

By month 6: ~5 cards/day

Rough Anki calculation:

Month 1: (~60 cards/day avg) × 30 days × 5 sec = 150 minutes

Month 2-3: (~15 cards/day avg) × 60 days × 5 sec = 75 minutes

Month 4-6: (~7 cards/day avg) × 90 days × 5 sec = 52.5 minutes

Total Anki time: ~277.5 minutes (≈4.6 hours total)

So Anki takes roughly 1/5 the time (4.6 vs 25 hours) while typically providing better retention due to optimal spacing of reviews. This is a simplified model – actual results vary based on card difficulty and individual memory patterns.

This means that you can have a deck of 100,000 cards and still keep up with it! You can memorize an insane amount of facts and retain them indefinitely.

How to use Anki?

Let’s look at the desktop app. Most of the apps work rather similarly.

Once you’ve downloaded the app, you can create a deck, which is a collection of cards.

Let’s add a card to this deck now.

Now we can study this card

Notice how you can select the difficulty you had in recalling the card. “Again” and “Good” should be self explanatory. My headcanon for when I select “Hard” is when I get something almost right (like one letter off) and “Easy” is for when I instantly recall the answer without even thinking about it.

The numbers above each button are how soon until you see that card again.

What can I do with Anki?

Reverse cards

The abstract form of a card in Anki is called a note. You can have a note that turns into multiple cards, for example.

One common use case for this is the “Basic (and reversed card)” — this is when you turn one note into two cards, with the fields swapped. This is useful when you decide to learn a foreign language and need to train both recognition and recall.

There’s more to this than you may think. Let’s say you speak English and are trying to learn Spanish. You’ll need to remember hello -> hola in order to speak the language. On the other hand, you’ll also need to understand the opposite direction of hola -> hello. This is so that when someone says that word, you’ll be able to comprehend them. 

With the reverse card feature, you can skip creating two cards when one will do, plus you won’t need to update the card twice when you need to change it.

Cloze deletion

Let’s say you’re trying to learn grammar in Spanish. In your textbook you’ll see this familiar table:

Yo quiero

Tú quieres

Él/Ella/Usted quiere

Nosotros queremos

Vosotros queréis

Ellos/Ellas/Ustedes quieren

You’ll be tempted to create the following card:

Front: Yo (querer)

Back: quiero

Stop! Don’t do it! There’s a better way!

The best way to learn this is by using context. We learn through pattern recognition, so it’s tough to apply facts memorized in isolation.

Create cards in this format instead:

Front: Yo ____ (querer) comer una manzana

​​Back:  Yo quiero comer una manzana

You’ll remember the conjugation in context and you’ll also get some free practice with the surrounding words. 

You can create these cards by using the Cloze card type and clicking here to delete a certain part of the word.

Image occlusion

This is basically cloze deletion but for images — you can black out certain parts of an image and then reveal it.

Audio

You can add audio cards! I’ve never done this, but it is an option, say for training listening comprehension.

Card design

The Fundamental Trade-off: there’s always a trade-off between creation time and review time. 

More effort during creation = Easier reviews later

Quick creation now = More challenging reviews

The beauty is that you get to choose where on this spectrum you want to be. If you don’t make this choice explicitly, it’ll be made for you implicitly.

Having a card in Anki is far more important than having a perfect card in Anki. Sometimes, I intentionally create “bad” cards just to get the information into the system. Later, when I have more time or when the card starts bothering me during reviews, I’ll clean it up. This approach keeps the momentum going while allowing for future improvements.

The best Anki cards share two key characteristics:

They’re extremely short (1-2 sentences)

They have minimal ambiguity

If you’re looking at paragraphs in your cards, that’s usually a red flag. While I have a few cards like that (intentionally), they’re the exception, not the rule.

Let’s look at what works and what doesn’t:

Good examples:

Vocabulary terms

Specific dates or events (like offices in the Cursus Honorum)

Clear, single-concept questions

Challenging examples:

Long, opinion-based content

Complex historical events (like the fall of the Western Roman Empire)

Detailed explanations that require multiple concepts

When dealing with complex topics (like system design), I use a different approach. Instead of trying to recall exact answers, I aim for conceptual understanding. For instance, with a question like “Why is caching useful?”, I’ll count it as correct if I can articulate the main concepts, even if I don’t recite the answer verbatim.

Don’t let perfect be the enemy of good. Start creating cards, use them, and let your review experience guide your improvements. The time you spend perfecting cards is time you could spend reviewing them – find the balance that works for you.

Sticking with it

Something that doesn’t get enough attention in the Anki community is how to stick with it long-term. After starting and stopping Anki at least 8 different times, I’ve learned some valuable lessons about maintaining a sustainable practice.

The most crucial piece of advice I can give you: don’t delete your decks. Ever. I’ve made this mistake too many times

• Always keep backups
• Store them even when you’re not actively reviewing
• Think of them as a knowledge investment

Returning to Anki after a break can be intimidating. Coming back to 1,000+ reviews is enough to make anyone want to quit before starting. This is fundamentally a UI/human psychology problem. The interface isn’t well-suited to handling breaks in usage. Fortunately there’s a hack:

• Adjust your daily review limit (in Options)
• Start with just 10 cards per day
• Gradually work your way back up
• Remember: any review is better than no review

Rather than treating Anki as a formal study session, I use it as a gap filler:

• During elevator rides
• On the train
• Waiting for food
• Any spare moment where I might normally check Twitter

Even minimal use is incredibly powerful. Reviewing 5 cards is better than 0. Some days, I only do 1 card. The goal is maintaining the habit, not hitting perfect numbers.

Make your own decks

This is important enough to warrant its own section. Don’t download huge decks from the internet, make your own. I’d recommend downloading a deck only if it’s less than 50 cards (let’s say learning an alphabet, or something like that).

The problem is that Anki is not for learning, it’s for reviewing. Learning on Anki is incredibly painful since you’re learning without context. If you use someone else’s cards, you’ll hit a brick wall.

Using Anki for language

I primarily use Anki for language learning. Most of my cards are straight vocab, with a few being cloze deletions (not enough I’d say).

A few quick notes

• Shove it into Anki (even if you don’t think you need it). I regret not having more cards. I only spent around 25 hours!!!! this year reviewing Anki.
• If you have multiple words with the same meaning (eg wealthy, affluent), generate two different images and then put them on the front of each card. You’ll associate the word with the image so you’ll remember both individually
• If a word isn’t sticking, create a mnemonic for it and put an image reminder of the mnemonic on the front of the card.
• My Russian teacher gives me a long list of vocab words after a lesson. I have a virtual assistant input them into Anki — I highly recommend this if you’re falling behind on card creation! You can also input words via .csv.

Using Anki for Leetcode

I also use Anki for Leetcode practice. Instead of endless random practice, I focused on mastering a core set of problems. Traditional LeetCode practice assumes you’ll naturally absorb patterns through repetition. Since I have a poor memory, this empirically wasn’t effective — I’d do a problem, retry it a week later and have completely forgotten how I had done it. 

The process:
Selected about 30 fundamental LeetCode problems

Created simple Anki cards for each problem

Front of card: Problem name/link

Back of card: Just time/space complexity

When a card comes up, I:

Open LeetCode

Solve the complete problem

Review complexity

Mark the card as done

As you can see I’m using Anki purely as a scheduling tool. 

The goal is not to memorize solutions per se, but rather common patterns. For example: converting adjacency matrix to adjacency list is now a pattern I recognize instantly. I wanted to focus on building “muscle memory” for common coding patterns.

The most common criticism I get is about overfitting – the worry that I’ll only learn to solve these specific problems. I’m not really worried because – 

• My memory isn’t great (ironically, this helps)
• The goal is pattern recognition, not memorization
• These patterns transfer well to similar problems

So far, I can say it’s been working quite well. I’m far more comfortable with Python syntax in a Leetcode context, and I can apply chunks of what I’ve memorized elsewhere in new problems.

I used to panic during coding interviews. Every interview felt like starting from zero and I had a significant fear of blanking out.

Memorization transformed my entire interview mindset:

• Instead of “solve this impossible problem,” it became “identify the right pattern”
• Having a solid foundation of memorized patterns gives confidence
• The interview feels more like pattern matching than pure problem-solving
• I became way, way calmer during interviews

What decks do I have?

Active

• Early Christianity in the WRE
• Leetcode V2
• Maximum NYC
• NATO Phonetic Alphabet
• Northeast trees
• NYC City Council Members
• Robert’s Rules of Order
• Russian
• Russian Cursive Alphabet
• Sys Design
• The New Colossus Poem
• Vim

Archived

• Clojure
• Graduate Intro to Operating Systems
• Leetcode
• Spanish

What should you memorize and why?

Let’s start with the obvious ones:

Academic Learning

The canonical use case for Anki is coursework, and for good reason. Beyond just acing tests, memorizing specific facts creates a mental scaffolding that helps you grasp bigger concepts. I’ve found that having these little details firmly in mind gives you something concrete to anchor those abstract theories to. If you’re interested in this check out how med students use Anki to memorize stuff for their coursework.

Language Learning

This one’s a no-brainer. Anki feels like it was purpose-built specifically for vocabulary and language learning. It’s perfect for that steady accumulation of words and phrases that language mastery requires.

I’ve also been experimenting with some less conventional uses:

Nature Knowledge

I’m currently using Anki to memorize trees in my area. It’s part of my goal to be more outdoorsy in New York, and having this knowledge makes every walk more engaging.

Literature

I’ve started creating cards for poetry & quote memorization. This is tough since I find memorizing ordered lists difficult.

Book Retention

I’m experimenting with transforming entire nonfiction books into Anki decks. For example:

• “Through the Eye of the Needle” by Peter Brown (about early Christianity)
• “Designing Data-Intensive Applications” (a technical book)

The time investment for experimentation is surprisingly low. I spent 25 hours total this past year on Anki, so a failed experiment might only cost me 5 hours across an entire year. That’s a pretty low-risk way to potentially discover a game-changing use case.

I don’t have definitive advice about what you should be memorizing. What I do know is that experimentation is key. The time cost is low enough that trying new approaches is almost always worth it.