Let me kill the headline before you anchor on it. I did not learn Japanese in twenty hours; nobody does. What I did was spend about twenty tracked hours — every minute is on a graph — getting from useless to functional, over roughly forty days at fifty new words a night. The gap between those two words turned out to be the most interesting thing I've studied in years — interesting enough that I built an app out of it. So this is a learner's note, not a victory lap.
First the receipts, because language-learning content is full of people quietly rounding up. Here is exactly what twenty hours bought:
- About a thousand words — the right thousand, which is the whole trick
- Most of basic grammar, absorbed without ever opening a textbook
- Enough to follow what's happening in a typical anime
- Enough to hold a real conversation — slow, clumsy, mistakes, but real
And here is what it pointedly did not buy: reading novels, writing kanji, debating philosophy. I skipped reading and writing on purpose — more on that below, because it's the single biggest shortcut in the whole thing and most people are too scared to take it. If you want to translate literature, I'm the wrong teacher and I wish you well. If you want to understand and be understood — watch, listen, speak — the rest of this is for you.
Ten years of anime taught me five words
Here's the embarrassing part. Before any of this I watched anime for ten years — a decade of pure immersion, the exact thing the internet swears will teach you a language on its own. Do you know what it taught me? Konnichiwa. Arigatou. Nani — and nani only because of the meme. Five words. That's not a learning story, it's a cautionary tale. Which is weird: little kids learn whole languages sitting in front of a TV. So why did a decade of the same input do nothing for me? The answer, once I found it, reframed everything:
The flashcards aren't the whole method. The flashcards are the bootstrap. They force-feed you to the threshold where anime stops being noise and becomes a free, unlimited, personal tutor that you were going to watch anyway.— from the notes for a video I keep meaning to make
When you understand nothing, your brain files the whole language under "background noise" and filters it out — you're not failing to learn it, you're not even hearing it. But once you know the first couple hundred high-frequency words, pattern recognition switches on: you catch a word, then a phrase, then your brain fills the gaps from context for free, the way kids do. Immersion isn't the myth. Immersion alone is. Something has to carry you to the threshold first.
Why twenty beats two hundred
First I did what everyone does: I downloaded the owl. A real month, streak going, gems flowing — and at the end I still couldn't order a coffee. It isn't broken; it's working as designed. Duolingo doesn't need you to finish, it needs you back tomorrow, and the two hundred hours the estimates put on basic conversation isn't a bug — it's more streaks, more ads, more subscription months. A streak is not a skill. It's a leash.
So I quit and — being a software engineer with an unhealthy relationship with optimization, lazy enough to spend ten hours finding the path that saves a hundred — asked a different question: what's the actual fastest way a human brain absorbs a language? The answer was three boring ideas doing all the work.
One: you forget on a curve, not a calendar. Every word starts decaying the second you learn it. Review too early, you wasted the rep; too late, it's gone. But review it at the exact moment you're about to forget and it sticks dramatically longer — and the next review can wait longer still. Ten minutes. A day. Four days. Two weeks. A spaced-repetition algorithm predicts that moment per word, per you, so nearly every one of those twenty hours ran at maximum efficiency. Twenty hours of the right thing at the right time beats two hundred hours of vibes.
Two: one new thing per card. Exactly one. Every card was a full sentence where I knew every word but one. Never drowning — that's day one of every textbook. Never coasting — that's month three of the streak apps. You sit permanently one step past your edge, which, having lived it, does seem to be the condition under which a brain actually acquires a language.
Three — the one that explains the ten wasted years: the thousand most common words cover roughly sixty percent of everything said in a typical show. Not the ten-thousand-word vocabulary an app drip-feeds you for a year. The first thousand — if they're the right thousand, in the right order. The owl had me on the word for turtle in week two; I have, to this day, never once needed turtle. The deck was never the destination — it just force-fed me to the line where anime flips from noise into a tutor I was going to watch anyway. The apps try to be the whole journey; this gets you to the part where the journey runs itself.
The shortcut everyone's scared of
Now the part people argue with me about. I skipped reading and writing entirely. No kanji. On purpose. The logic is almost rude in its simplicity: kanji is the genuinely hard part — hundreds of hours — and for my goals it bought nothing. I wanted to follow anime and make friends in Japan, both of which are spoken. Manga has furigana; hiragana and katakana took a few hours; the vocabulary and grammar are identical whether or not I can hand-draw a character. So kanji was pure diminishing returns and I left it on the shelf. That one decision is probably half of why twenty hours was even on the table.
What it actually felt like
I won't pretend it was smooth. Day one is miserable — nothing sticks, you fail the same word four minutes apart and feel stupid. The reframe that saved me: reaching for a memory and missing isn't failure, it's the rep — that is the wiring happening. Fake progress feels great; real progress feels like this. The method's one true enemy isn't difficulty, it's the pile: miss a few days and the reviews stack, and I came back to three hundred due cards and "waited for the right time," which is procrastinator for never. What got me back was lowering the bar to the floor — ignore the number, do ten minutes, leave.
And then, about two weeks in, the thing happened.
A character said a full line and I just… understood it, before the subtitle appeared. I paused the episode and walked a lap around my room. Ten years of watching anime and the language had been a locked door the entire time — and it just, opened.— from the notes for a video I keep meaning to make
From there it stops being studying; every episode is a boss fight you're slowly winning. Which is the honest place to draw the boundary. Twenty hours got me the key and the door. The three months I later spent in Japan — karaoke every night, friends who spoke no English, a singing teacher who only spoke Japanese — that wasn't the learning. That was the cement. The learning was already done, quietly, fifty cards at a time, before I ever landed.
The part that became an app
I started telling people about this, and every one had the same reaction: okay, that's insane, how do I do it? So I'd send the full setup — the app, the deck link, the settings, the scheduling config, the guide I wrote. Twelve people asked how. You know how many made it through the setup?
Zero. Not one. Not my best friend. Not my mom. The most effective learning method I have ever used — proven, free, sitting in plain sight for twenty years — and the setup screen kills a hundred percent of normal people before card one. The method is solved. The access is broken.
That bothered me for months. So, being lazy in a good way, I did the least lazy thing I've ever done and spent six months building the version with no setup: one deck per language, the right thousand words first, one new thing per card, real audio on everything, the algorithm pre-tuned and invisible. That's what Flua is — the whole mechanism above, working silently, so the twenty-hour path is the only thing you have to think about.
Not fluent. Functional. The interesting part was never the result. It was finding that the door had a key the whole time, and that the key was smaller than anyone admits. The same three ideas became the method I reach for first whenever I want to get less bad at something fast.
Keep going How Flua began · How I learn anything in 30 days