Last month I watched a 50-minute conference talk on React Server Components, nodded along the whole way, and closed the tab feeling smart. Three days later a coworker asked me when you'd actually reach for a server component over a client one. I had nothing. I could picture the speaker's slides. I couldn't tell you a single thing on them.
If you learn programming from YouTube, conference talks, newsletters and the odd podcast on your commute, you've probably had the same moment. The material was good. You were paying attention. It still didn't stick.
I've spent a while trying to fix this for myself, partly because I ended up building a tool around it. Here's what has actually worked, with no claims that I've found a magic trick.
Watching feels like learning, which is the problem
A well-made tutorial is smooth. The instructor types, the code runs, the explanation lands. That smoothness tricks your brain into thinking you understand, because recognizing something is much easier than producing it.
Psychologists call the fix "retrieval practice." The best-known study is Roediger and Karpicke's 2006 paper, where students who read a passage and then tried to recall it remembered more a week later than students who simply reread it. Rereading felt more productive to the students. It wasn't.
For developers, the equivalent of rereading is rewatching the tutorial, or scrolling back through your bookmarks and thinking "oh right, that one." The equivalent of recall is closing the video and trying to write the thing from memory.
Take notes after, not during
I used to pause a tutorial every 30 seconds to copy code into a notes file. The result was a perfect transcript I never opened again, and I barely followed the video because I was busy typing.
Now I watch a section straight through, then pause and write two or three sentences in my own words. What problem does this solve? When would I use it? What's the one line of code that matters? If I can't answer, I rewind. That rewind is the useful part, because now I know exactly which bit I missed.
A note like "use useTransition when a state update would block typing in an input" is worth more than 40 lines of copied code. You can test yourself on it later.
Build the smallest possible thing, the same day
The gap between "I watched it" and "I can do it" closes fastest when you write code without the video open. It doesn't need to be a project. For the Server Components talk, I eventually made a tiny Next.js page that fetched a list on the server and had one client component for a like button. Fifteen minutes. That little file taught me more than the whole talk had.
The rule I try to follow: if a tutorial is worth an hour of watching, it's worth fifteen minutes of building something with nobody showing me how.
Come back to it on a schedule
This is the step almost everyone skips, me included for years. You learn something, use it once, and then don't touch it for two months. When you need it again, it's gone.
Spaced repetition solves that by bringing an idea back at growing intervals: a day later, then a few days, then a week or two, then a month. Each time you successfully recall it, the next gap gets longer. Anki is the classic tool for this, and plenty of developers use it for things like Git commands, regex syntax or language quirks. (I wrote up how the SM-2 scheduling algorithm works in JavaScript if you're curious what happens under the hood.)
The catch with flashcard apps is that someone has to write the cards. For a 50-minute talk, turning it into four or five good questions takes real effort, and after a long day I'd rather open another tab than write flashcards.
The bookmark graveyard
Here's the honest version of my learning setup before I changed it. A Watch Later playlist with 200-plus videos. A browser folder called "read later" that I never read. Screenshots of tweets about TypeScript tricks. Saved Instagram reels of someone explaining Big O notation in 60 seconds. A podcast app full of half-finished episodes.
None of it was organized and none of it came back to me. Saving things felt like progress, and I never actually touched them again.
That frustration is the reason I started building Recl. You save videos, reels, podcasts or articles into it, it writes a summary of each one, sorts them into topics, and then uses spaced repetition to bring the key points back before you forget them. It's the same "come back to it on a schedule" step from above, except the review material comes from what you already saved, so the card-writing step mostly disappears.
Recl's feed, where saved videos and articles show up as short summaries.
I still do the other steps by hand. No app is going to build the tiny Next.js page for me, and it shouldn't. But having the talks and tutorials resurface on their own means the stuff I save doesn't just vanish into a playlist anymore.
A simple routine you can start this week
Pick one tutorial or talk you've already watched recently. Without rewatching it, write down three things you remember and one question you can't answer. Look up only the answer to that question. Then build the smallest piece of code that uses what you learned.
Put a reminder in your calendar for three days from now and another for ten days out. On each one, try to explain the idea out loud or in a short note before you look anything up. If you blank, that's fine. Blanking and then checking is exactly the practice that makes it stick.
Do that for a few weeks and you'll notice the difference. You'll forget less, and you'll stop saving so many videos you were never going to watch.
I'm Spartan Rich, and I build Recl, an AI learning app that saves your videos, reels, podcasts and articles, summarizes them, and uses spaced repetition so you remember what you learned.
Top comments (0)