The question
The Rest Is Science asks: how many words do we know?
A small studio with a big thing for words.
VocaboMeter began as a curious side project at KoDesign. We are a small independent web studio that likes turning good questions into thoughtful, playful websites.
Meet KoDesign
KoDesign · Independent web studio
Made independently, with care
There is no giant assessment company behind VocaboMeter. It is designed, written, tested, and maintained by KoDesign. That means we can keep it focused, transparent, and a little more human.
The real origin story
From one podcast question to four languages.
VocabOwl appears
Someone turns the question into VocabOwl, a playful English vocabulary estimator.
The follow-up
The podcast later talks about VocabOwl. That video is embedded below.
My language
I try VocabOwl and want to make the experience available in Albanian.
Why stop there?
The idea grows into VocaboMeter for Albanian, English, German, French, and eventually more.
The follow-up
The podcast returns to the vocabulary question
This later video discusses the experiment after VocabOwl was created. It matters to the story, but it is not the first discussion that started the chain.
Visit The Rest Is ScienceUnder the hood
How the game works
Choose a language
Each language has its own reviewed question bank, split into five difficulty bands.
Answer 50 words
You receive ten randomly selected words from every band. Pick a meaning or choose “I don’t know yet.”
Get your result
We calculate a competitive skill score and a separate approximate vocabulary range.
The numbers
How scores are measured
Correct answer
+1 pointWrong guess
−⅓ pointI don’t know yet
0 pointsAdjusted accuracy = max(0, (correct − wrong ÷ 3) ÷ 50)With four possible definitions, random guessing succeeds about one quarter of the time. Subtracting one third for a wrong guess makes the expected score from pure guessing approximately zero.
Vocabulary estimate
Random words alone would overrepresent common vocabulary. VocaboMeter instead tests 10 words from each of five difficulty bands, then uses performance in each sample to estimate knowledge across that band.
Leaderboard score
The leaderboard uses the normalized 0–1000 skill score, making results easier to compare across languages. Completion time only breaks ties.
The sample behind the number
Five small samples estimate a much larger vocabulary.
Random words alone would overrepresent common vocabulary. VocaboMeter instead tests 10 words from each of five difficulty bands, then uses performance in each sample to estimate knowledge across that band.
Core basics
3,000wordsIntermediate
7,000wordsAdvanced
10,000wordsExpert
25,000wordsThe obscure
40,000wordsVocabulary estimate
Σ max(0, (correct − wrong ÷ 3) ÷ 10) × band size3,000 + 7,000 + 10,000 + 25,000 + 40,000 = an 85,000-word estimation model. These are transparent modelling buckets, not claims about exact dictionary size.
Example: a 60% chance-adjusted result in the 7,000-word Intermediate band contributes about 4,200 words to the estimate.Open about the work in progress
Our playable question bank is growing while human review continues.
The current words, definitions, and distractors were written with AI and passed an editorial check. We are working closely with people who know these languages to verify every question, improve weak ones, and add many more.
Approved for play does not yet mean verified by a fluent speaker.A different sample next time
Each run selects 10 words from every difficulty band and never repeats a word within that run. Because the bank is much larger than one 50-word test, playing again can produce a substantially different set.
A transparent estimate
≈Useful, playful, not a certificate
VocaboMeter is an educational game. Word knowledge has many dimensions, and language sizes are debated. Treat the result as a well-reasoned estimate and a reason to discover another word.
Credits
With credit and thanks
The idea was inspired by The Rest Is Science with Prof. Hannah Fry and Michael Stevens. The approachable test format also owes a clear debt to VocabOwl.