Duolingo's Incentives Prevent Meaningful Language Acquisition

Duolingo is bad, and it isn't allowed to make meaningful change to improve itself. Its public-facing goal of helping you learn is obstructed by hidden economic forces.

I run into people that use the app, and I always bring up the same points about why it should be avoided.

I hope the following serves as a reference to point people towards the next time the topic comes up.

#Maximizing engagement and earnings

As a publicly traded company and popular app, Duolingo's primary incentive is maximizing engagement and earnings. Tangible metrics like subscription revenue, daily active users, and in-app purchases drive important decisions.

Focusing on these metrics presents two problems in relation to language acquisition:

1. The user's true acquisition is hard to map. Lesson completion may just signal usage, not learning.
2. In general, language acquisition is monotonous, slow, and not easy.

The former is not immediately important to the company's market success, and the latter describes a state incompatible with trending apps^*. Because of this, Duolingo won't optimize for your language acquisition.

To use its 2024 Q3 earnings report as an example [1], here's what's at stake:

• 8.6 million paid subscribers at quarter end
• 37.2 million daily active users
• A net income of $23.4 million

Prioritizing your acquisition means poor profitability and failure to compete in the language learning industry.

Any meaningful learning is just a bonus, and certainly not the priority the app has in mind.

#Unchallenging lessons

This incentive accurately predicts a setting in line with the attention economy, not with learning.

On one hand, you have things designed to arrest your attention like:

• Experience points
• A day streak
• Unlocking new levels

On the other hand, you get lessons that result in some acquisition but are low quality.

I'm not against gamified learning, but the app is closer to what I'll call learnified gaming: amusement at its center, with learning subordinate.

It appeals to someone who’s not serious about studying or the novice who doesn't know what's best.

#The app might not want you fluent

Another conflict arising from the incentive structure is that the app theoretically benefits from delaying your fluency.

Maybe if you reached fluency, you'd have little reason to stick around.

It's why the medical industry won't use lifestyle-oriented treatment to cure the chronic diseases: if you cure the patient, you lose a customer.

It's why you shouldn't ask where the bathroom is as an excuse to start a flirtly conversation with a girl. Once she tells you where it is, she's given you a valid reason to go over there and not bug her.

If you used the bathroom line, you'd lose out on the girl. It's the same thing Duolingo is afraid of. If the app did its job, your newfound fluency might be your signal to stop using it.

Owned problems like these are a problem in industries where it is disadvantageous to empower customers with the tools to make themselves obsolete.

#Short lessons, small benefits

Still, you might believe you lack the time to fit in proper learning. An undeniable advantage of the app is its small lessons.

Even then, it may be better to take a barbell strategy approach: instead of spending scattered time every day using Duolingo, dedicate two days a week to many uninterrupted hours of focused study using Comprehensible Input.

Ultimately, it’s not like Duolingo is magically bad for no reason, and it's not like it can't suddenly be great. It just can’t allow itself to change because its business is tied to its app-engagement economics.

As a result, there's a divergence between the utility of the app to the learner and the needs of the publicly traded company.

A misalignment between what you and your tools are optimizing for foreshadows suboptimal results^†.

#What to do instead

Duolingo has no place in the learner's toolbox.

Don't focus on grammar, speaking, or memorization either. Use a natural approach oriented around storytelling and listening. This is the basis of Comprehensible Input^‡, which appears to be more effective than traditional means of learning [2, 3, 4].

Visit the Comprehensible Input Wiki to find content in over 60 languages.

Ask an LLM to tell you about any topic that interests you, at a difficulty you can manage.

Try a delivering it a prompt like this, tweaking as needed:

I'm an absolute beginner at <language>, and I want to learn using a Comprehensible Input

Use basic, A0 vocabulary. Use emojis next to words to signal their meaning

Tell me about <topic>

Aim low when it comes to the difficulty of your input. You should understand at least 95 percent of the words in the content you consume.

Sit at a desk and focus; it's a hard strategy to beat.