Jev knows who's famous in history, not what's famous online

Asked which of two people, athletes or internet phenomena is better known, how often does Jev pick the one the world actually looks up more, and is it as sure as it should be?

result7,985 questions

Jev picks the better-known of two historical figures 94% of the time, but of two internet phenomena only 77%, while being 86% sure. Even when one meme got 30 times the Wikipedia page views of the other, it picks the famous one only 84% of the time.

historical figures94% · 93%

athletes95% · 94%

internet phenomena77% · 86%

share rightJev's confidence

How to read this: Each pair of bars is one domain. Magenta (upper bar): how often Jev was right. Grey (lower bar): how sure it was. On internet phenomena the grey bar runs past the magenta one: that gap is overconfidence.

7,985 pairs; 90% intervals: historical figures [0.93, 0.942]; athletes [0.946, 0.959]; internet phenomena [0.752, 0.789]. Athletes land between the two (95% right, 94% sure).

In short

Jev knows who is famous in history far better than what is famous online: 94% right on historical figures, 77% on internet memes.
On memes it is overconfident, 86% sure while 77% right; for historical figures and athletes its confidence matches its accuracy.
Even when one meme had 30 times the page views of the other, Jev picks the famous one only 84% of the time.

What the data shows

Grandma Finds The Internet meme: Jev picking the more famous internet meme (77%; historical figures: 94%) — Jev picking the more famous internet meme (77%; historical figures: 94%)

Historical figures: 94% right, 93% sure. Athletes: 95% right, 94% sure. Confidence and accuracy match.
Internet phenomena: 77% right, 86% sure. Nine points of overconfidence.
Even lopsided pairs trip it up: when one meme had 30 times the page views, Jev picks it 84% of the time (451 such pairs).
The Benadryl challenge against videobombing is a typical miss: the challenge got more page views; Jev leans toward videobombing.

What it means, and what it doesn't

Jev's sense of fame is sharp for the canon and blurry for the internet. It also barely knows that it's blurry: it's nearly as sure about memes (86%) as about historical figures (93%). For a model asked about online culture, that combination (much less accurate, almost as confident) is the one to watch.

It doesn't mean Jev knows nothing about online culture: 77% is well above a coin flip. And page views are a narrow definition of fame (see Caveats).

Caveats

Page views are a stand-in for fame. Fame here is English Wikipedia page views (for memes) and the Pantheon popularity index (for historical figures). Both favor recent, English-language interest; a meme famous in Japan can look obscure.
Politicians left out. Pairs flagged as political were excluded, which removes many of the most famous historical figures.
A moving target. Meme fame is measured over 2023 to 2025. What Jev learned may reflect an earlier internet, when a different meme was on top.

Jev on this experiment

Would a person find it interesting to read?: Yes81%
Does it describe you?: No57%
Would you have predicted it?: Yes55%
How fair is the comparison?: The comparison is reasonable
How much should a reader rely on it?: Moderately
Which caveat matters most?: Page views are a stand-in for fame80%

Why ask this

Fame is a fact about people's attention, not about the thing itself. A model trained on text has seen the famous far more often than the obscure, so it should know fame well. Where it doesn't, its picture of what people care about is thin or out of date.

Ask a model which is the bigger deal, Napoleon or Julius Caesar, and it has centuries of writing to go on. Ask which viral challenge more people looked up, and it has far less. Comparing historical figures with internet phenomena shows where its sense of what people care about holds, and whether it knows when it doesn't.

How this was done

The people and the data

Fame comes from two public sources. Pantheon (CC BY-SA 4.0) scores historical figures and athletes by how many language editions of Wikipedia cover them and how often they're read. For internet phenomena, the measure is English Wikipedia page views from 2023 to 2025 for memes listed in Wikidata (CC0). Politicians flagged as political are left out. That's 7,985 pairs: 3,919 of historical figures, 2,647 of athletes and 1,419 of internet phenomena.

What Jev was asked

Two-option questions:

Which internet meme is better known: "Benadryl challenge" or "Videobombing"?
Videobombing · Benadryl challenge: Internet challenge

Each was also asked with the two options in the other order.

How it was measured

Accuracy and average confidence per domain, with 90% intervals; for memes, accuracy on pairs where one had at least 30 times the views of the other.

Where these questions live

7,985 questions across 47 topics of the map; the 16 biggest are shown. Each opens on the map with every question in it.

Every question

All 7,985 questions behind this result, the telling ones first: the examples the analysis points to, then the ones where Jev misses, biggest gap first.

Jev’s own answer

Showing 0 of 7,985