The method
The science of the melting word.
Why we built a language app that looks like a stock market.
Berlin, 1885
A scientist forgot things on purpose.
His name was Hermann Ebbinghaus. He memorised lists of nonsense words, waited, and tested himself again and again. He found something uncomfortable: memory doesn't fade slowly, like a battery. It crashes. After one day, most of what you learn is gone. After a week — almost all of it.
Scientists call this the forgetting curve. It is one of the oldest and most repeated results in psychology. And it means one simple thing for language learners: the word you “learned” yesterday is not learned. It is leaving.
Honest numbers
Why a checkmark is a lie.
Most vocabulary apps mark a word as “done” the moment you answer it once. But your brain has no “done”. It only has right now. A word you knew perfectly last Tuesday may be half-gone today — and the app still shows the same green checkmark.
So we removed the checkmark and gave every word a price.
A word's price is our honest answer to one question: how likely are you to remember this word right now? Answer a card well — the price rises. Do nothing — the price melts, day by day, following the same curve Ebbinghaus drew by hand 140 years ago. The chart you see in your portfolio is not a game. It is your memory, drawn in real time.
The chart is not a game. It is your memory, drawn in real time.
Two numbers per word
Not all words melt at the same speed.
Here is the beautiful part. Every time you successfully recall a word, something changes in your brain: the memory becomes more stable. Modern memory models — the same family of algorithms used by the most serious learning tools in the world — track two numbers for every word: how strong the memory is, and how fresh it is.
A new word is fragile. Miss two days, and its price crashes. But a word you have recalled five times over growing gaps becomes what traders would call a blue chip: it can sit untouched for weeks and barely lose value. This is the real goal of learning — not to answer a card today, but to make the memory so stable that you almost never need to review it again.
And the fastest way to build that stability is surprisingly specific: recall the word right when it starts to fade. Too early is a waste of your time. Too late, and it's gone. Psychologists call this spaced repetition, and the effort you feel in that moment — that little “ugh, I almost know it” — is called desirable difficulty. It's not a bug. That strain is the learning.
The daily queue
You don't choose your session. Your memory does.
When you press Study, we don't show you a random pile of cards. We look at every price in your portfolio and pick the words that are melting right now — the ones at the exact point where one recall gives the biggest gain in stability. Melting words first. New words next. Fresh words wait, because touching them today would teach you almost nothing.
And a session is never more than 20 words. Not because we are lazy — because attention is a limited resource, and twenty focused recalls beat a hundred tired ones. Five minutes. Portfolio green. Done.
Passive gains
Even listening moves the market.
For a long time, apps only counted one kind of learning: the flashcard. But your brain doesn't stop learning when your hands are busy. When you listen to your personal podcast — built from your own words — and an episode ends, every word in it just got refreshed in your memory.
We count that. Honestly, though: hearing a word is not the same as recalling it. Recognition is a weaker signal than retrieval — the science is clear on this. So a finished episode gives your studied words a light touch: their prices get fresher, but your review plan doesn't change. A walk with headphones will never replace a session. But it is no longer invisible. It counts, because it does count — in your brain.
The bottom line
A portfolio of living assets.
Your vocabulary is a portfolio of living assets. Each one grows when you use it and melts when you don't. We watch every one of them, every day, with a hundred years of memory science running under a very simple screen — so that the only thing you have to do is show up for five minutes and keep it green.
Further reading: Ebbinghaus (1885) on the forgetting curve · Roediger & Karpicke (2006) on the testing effect · Bjork on desirable difficulties