There are two popular language learning software platforms: Anki and Duolingo. Anki is hard, free and effective. Duolingo is easy, commercial and ineffective.

The number of Duolingo users far outstrips the number of Anki users. Duolingo has 8 million downloads on the Play Store. Anki has 40 thousand. So there are 200 Duolingo users for every Anki user[1]. If you ask a random language learner what software to use they'll probably suggest Duolingo. If you ask a random successful language learner what software to use they'll probably suggest Anki. Most language learners are unsuccessful.

It should be no surprise that most language learners use an ineffective product. Learning a language is hard. Duolingo is designed to attract as many customers as it can. Therefore Duolingo must be easy to use. Anki is designed to work. Therefore Anki must be hard to use[2]. Effectiveness and mass-adoption are mutually exclusive.

On average, you'll have to ask 200 language learners what software they use before one of them tells you about Anki. In practice, you are unlikely to stumble across Anki at all. You can only find out about Anki if you go looking for it and you'll only go looking for it if you already know it exists.

No one ever told me about most of the software I use. Usually I'll infer that it must exist and then go looking for it. This is how I discovered qutebrowser, Spacemacs, tmux, i3 and—of course—Anki.

This principle isn't limited to open source software. Everything I've ever taken seriously exhibits a similar pattern. There's a common way most people do it and there's a cheaper, better, skill-intensive way a tiny minority do it.

Minority opinions are inherently controversial. Here are a few (relatively) uncontroversal examples to illustrate the trend:

  • Vim keybindings vs. conventional hotkeys
  • Lisps vs. programming languages without syntactic macros
  • Good history books vs. the news
  • Powerlifting vs. popular exercise de jour
  • Tiling window managers vs. the desktop metaphor

If you disagree with one of the examples above that's okay but I won't reply to you in the comments.

I don't mind choosing to be part of the majority. I watch mainstream movies. I eat at popular restaurants. I perform the oldest magic tricks in the book. I'm okay with these choices I know about.

I make most of my choices unknowingly, by default, from the options presented to me. Most of the time I pick Duolingo over Anki without ever knowing I've made a choice.


  1. I'm ignoring the Anki desktop app, the med student userbase, the paywalling of the iPhone Anki app and differences in churn to simplify this example and because I don't have the relevant numbers. ↩︎

  2. I mean that effective spaced repetition is inherently difficult. The Anki onboarding experience could be improved without loss of utility. ↩︎

New Comment
9 comments, sorted by Click to highlight new comments since:

Usually, my invisible choices come in the form of a) being provided with or finding an easily accessible solution to a larger problem and b) ignoring ways in which that solution turns out to be insufficient for my goals. I've tended to solve these by

1. Reflecting on whether I am doing things that do not fully result in the outcomes I want for them

2. Spend five minutes thinking about what might solve the issue

3. Spend five minutes trying to implement any solution I came up with

Some things that this process has helped me resolve:

-I listen to music while I shower but sometimes water gets under my phone case, rendering my phone mildly inconvenient to operate for >30 minutes. Solved by getting a new, waterproof case and turning my phone's screen to face down instead of up while I shower

-I tried learning to bike early this year and my parents let me borrow an old bike. I was able to ride, but only very awkwardly. Solved by visiting a bike shop and test-riding the first bike they recommended. It turned out the bike I was riding was just too small.

-My Pandora music station was tolerable but I kept hearing a lot of stuff I didn't enjoy. Solved by actively down-voting songs I wouldn't be excited to hear again.

-Flossing my teeth is mildly annoying and wrapping dental floss around my fingers would tend to reduce circulation a perturbing amount. Solved by buying dental floss picks.

Also tangentially, something similar to this also happened to me with graduate school. For a long time, I saw graduate school as the logical extension of me being a person interested in and good at computational research. Turned out that my day-to-day enjoyment of professional work is basically identical to my enjoyment of research, except paying more.

Out of curiosity, how is Anki better than Duolingo? I used Duolingo to learn the basics, then learned from actually reading material in the language. Is Anki suited for anything other than vocabulary extension? I find Duolingo way better for getting a feel for grammar.

Anki might not be suitable for anything other than vocabulary extension (and medical school). I'm not really sure. I've only ever successfully used it for vocabulary extension.

I may be biased because I use Anki to study Chinese, an unusually difficult language. The relative merits of Duolingo vs. Anki may be less clear-cut for easier languages like Romance languages where it's not necessary to learn the language so systematically. In the case of Chinese, grammar is so simple and vocabulary is so hard that vocabulary extension is pretty much the entire game, so optimizing for anything other than vocabulary acquisition can cause you to fail in the long intermediate slog.

What language(s) are you studying?

I find Duolingo way better for getting a feel for grammar.

This makes sense. I prioritize vocabulary over grammer for a couple reasons. [1] You can communicate effectively with vocabulary and without grammar (but not vice-versa) and [2] the difficulty of learning a language's grammar is far outweighed, in the long run, by the difficulty of learning vocabulary.

It may make sense for someone to start learning with Duolingo and then transfer to Anki (for hard languages) or just reading material in the language (for easy languages). I'm happy to hear that Duolingo works for you.

Starting with Duolingo might be OK, but it’s very bare-bones and unlike the real deal of having thoughts in a language and then expressing yourself. In fact, I recommend not using it after you have a rudimentary grasp of the language.

Try entering full sentence-to-sentence cards in Anki. At first, you won’t be able to write good sentences on your own, so use a sentence-search service for that (I don’t remember what they’re called, but at the least you can use the example wordreference sentences). Over the course of a year, I input all the thoughts I’d had that day which I couldn’t translate. By learning to produce words in-context instead of memorizing the multilingual dictionary, I began to run out of additions to the Anki deck. That exercise gave me near-fluency, which has persisted to this day.

So, I think Anki is suited for nearly everything except oral skills. Plus, spaced repetition saves time.

Also, you might try an app like Tandem. If you’re single, try learning to be interesting in the foreign language - it’s naturally motivating, and both of you can chalk up any awkwardness to mistranslation. :)

I've found Anki really terrible for learning, even for simple things like vocabulary; what it does is help me remember things I've already at least half-learned.

How is effective spaced repetition inherently difficult? I thought the entire point of Anki was to make it easy, i.e. automate everything about it that can be automated. All the user has to do is turn it on and do the work every day, but presumably that's true of Duolingo also (which I have no experience of).

I have actually not found Anki effective for language learning. My experience has been that my practice with flashcards has not transferred to the situation of trying to either utter a sentence in the target language, or understand one when I hear it. That is a hurdle I have never surmounted, however I go about the task.

Spaced repetition is almost designed to be difficult. The method works best when you are presented with questions that are at the limit of of your abilities. If you see a lot of cards that are easy to answer you are probably wasting a lot of time. Therefore, when it works well you are seeing a lot of cards that are hard for you to answer. I think that feels bad for a lot of people because only being presented with the hard cards makes it seem like they aren't good or improving at the thing.

Actually making the cards is what stops me.

Spaced repetition is inherently difficult because it requires a period of intense focus. It you're doing it efficiently it's hard on a minute-to-minute level. In my personal experience, it's the most intense form of study I can scale up. When I'm using Anki, it's the hardest 10-15 minutes of my day.