Readers beat a 170k-word vocab quiz by always picking the longest answer
The web app guesses how many of English's 170,000 words you know from a 100-question quiz. One reader wrote a six-line console script that just clicks the longest option every time — and scored in the high 70s to mid 80s.
The tool's tell: on multiple choice, the correct definition is usually the wordiest one. People reported 62k 'words known' by ignoring the words entirely and always choosing the longest blurb.
An app developer who'd built the same kind of quiz explained the bug at the source: AI tends to write an over-complete correct answer, making it longer than the distractors and quietly defeating the whole test.
A quiz you can beat without reading the words
The web app gives you 100 multiple-choice words and estimates how many of English's ~170,000 you know. It's a fun toy — readers posted scores of 71k, 74k, 75k — until someone noticed it has a tell.
The correct definition tends to be the longest of the four options. A high-karma reader didn't just assert it; he automated the test:
"You are correct. I tested that hypothesis about a dozen times and it seems that if you always pick the longest you'll get it right somewhere in the high 70s to mid 80s." — latexr
His comment included a six-line browser-console loop that clicks the longest button on each of the 100 questions. Another reader ran the strategy by hand and confirmed it:
"Hahahhaha i got 62k points by just choosing the longest definitions. Great observation!" — a reader who reproduced the exploit manually
Why the longest answer is the right answer
The explanation came from someone who'd shipped the exact same kind of app and hit the exact same wall:
"These were likely all AI generated… AI often wanted to make a very covering answer for the correct one, making it often longer than the others, thus defeating the idea of the quiz in the process." — a developer who built a similar vocabulary app
The machine-written distractors give the game away other ways too. One reader found definitions that reference the word itself — "Lethargic: exhibiting lethargy" — and others that you can rule out by decomposing the word into common parts, which means the obscure word couldn't have meant that.
The deeper problem: writing wrong answers is hard
Even a reader who scored 74k didn't trust it, and explained why the number is inflated:
"The distractor answers weren't convincing enough… if it was [decomposable into common English words], then it wouldn't have been obscure." — a reader who compared it to a notoriously hard medical exam
He contrasted it with Poland's LEK medical-graduate exam, where the distractors are written so well it's near-impossible to vibe-guess above chance. That's the real lesson buried under a fun word game: a vocabulary estimate is only as honest as its wrong answers, and an LLM asked to write four options will quietly make the true one the most complete — and therefore the longest. The score the app hands you measures, in part, your ability to spot a verbose blurb.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
How 'think of the children' becomes real ID for every packet you send
A 2023 essay traces the path from age-verification mandates to identity attached to all internet traffic. The thread's quietest line: in 2023 it read as paranoia, in 2026 it reads as a roadmap.