Peer learning vs. online courses

Online courses are very good at getting you from zero to understanding. Somewhere between intermediate and advanced, they stop being the best tool, and that is where many learners stall. This is not an argument against courses. It is an argument for knowing when to switch, and what to switch to.

The short version: use courses for foundations, then learn alongside other practitioners, and teach what you know as early as you can. The research on learning by teaching explains why the last part matters so much.

What are online courses good at?

Structure. A good course gives you concepts in the right order, clear explanations, worked examples and exercises with known answers. When a field is well mapped and the problems are standard, that is exactly what you need.

Courses are also cheap, available at any hour and patient. You can replay a lecture as many times as it takes. For getting to "I understand how this works", they remain one of the most efficient tools there is.

Why do so few people finish them?

Because a course on its own asks a lot of your motivation and gives little back in the way of accountability.

The data is stark. In their study of Harvard and MIT courses on edX, Reich and Ruipérez-Valiente found that low completion rates had not improved over six years. As reported by Inside Higher Ed, 3.13% of all participants completed their course in 2017-18. Even among learners who paid for a verified certificate, only 46% finished.

Not every enrolment is a serious attempt, so the raw rate overstates the problem. But even committed learners drop off. What keeps people going is usually other people: a deadline someone else is waiting on, a question someone else asked, a session someone else booked.

Where do courses run out?

At the edge of the syllabus. The problems experienced practitioners face are rarely standard. They are specific to a codebase, a team or a product. A course can teach you how a message queue works. It cannot sit with you while you work out why messages in your system arrive out of order.

At that stage, progress tends to come from working through a real problem alongside someone who has seen something like it before: live, in your context, with room to ask the question you did not know you had.

What does research say about learning by teaching?

That preparing to teach helps you understand, and actually teaching helps you remember.

In a set of four experiments, Fiorella and Mayer (2013) asked students to study a lesson either expecting a test or expecting to teach. Those who prepared to teach did better on an immediate test. One week later, the students who had actually taught the material, by recording a short lecture, did clearly better than those who had only prepared. The authors suggest that explaining pushes you to organise and connect ideas, not just recognise them.

A related study points the same way. Nestojko, Bui, Kornell and Bjork (2014) found that people who expected to teach a passage recalled more of it, and recalled it in a better organised way, than people who expected a test.

The practical lesson: teach early. You do not need to be an expert to explain something to a person a few steps behind you, and doing it will show you exactly where your own understanding is thin.

How do the two approaches compare?

The rows that matter most change as you progress. Early on, structure wins. Later, feedback and teaching do.

How do you know it is time to switch?

A few signs suggest a course has taken you as far as it can:

• You can follow the lessons but stall on your own projects. Recognising a solution is different from producing one.
• Your questions are about your situation, not the topic. "Why does my query plan do this?" is not in the syllabus.
• You keep starting new courses on the same subject. That is often a sign you need practice, not more explanation.
• Someone asks you a question and you cannot explain the answer. That is the gap teaching would expose, and close.

How do you find a good learning partner?

Look for overlap and contrast at the same time. A good partner knows something you want to learn, wants something you can teach, and is close enough to your level that both of you can follow the conversation.

• Shared context helps. Someone working in the same language or domain can engage with your actual problems, not toy versions of them.
• A small gap is better than a large one. An expert can be a great teacher, but a peer a little ahead of you often remembers what was confusing.
• Reliability matters more than brilliance. The partner who turns up every week will teach you more than the one who cancels.
• Swap roles. If you only ever learn from one person, you miss the benefit of teaching.

What makes a peer session productive?

Preparation and a narrow focus. Sessions that try to cover a whole topic drift. Sessions built around one problem rarely do.

1. Bring a real problem. A failing test, a design you are unsure of, a concept you cannot explain. Share it before the session if you can.
2. Agree the goal in the first two minutes. "By the end, I want to understand why this query is slow" keeps both of you on track.
3. Drive, then switch. The person learning should be at the keyboard. Explaining while someone else types is slower and teaches less.
4. End with a summary in your own words. If you cannot say what you learned in two sentences, you have found next week's topic.

How do you combine them in practice?

Keep courses for new ground and use peers for depth. A simple weekly rhythm works for most people:

1. Learn a concept from a course or documentation.
2. Apply it to a real problem of your own.
3. Talk it through with another practitioner, especially where you got stuck.
4. Teach it to someone a step behind you.

If you want the sessions to count for more than memory, record them. Lemma sessions are live one-to-one sessions on a single skill, where either person can teach, and each completed session is added to that skill's record on both people's Skills Passports. Teaching counts for more than learning there too. Between sessions, the AI tutor covers drills and explanations.

If you are looking at apps for this, is there a skill swap app that proves your skills? sets out what to check before you commit. If you are changing careers, this mix is also how you build evidence a certificate cannot show; our guide for career switchers covers that path.