A reading of 1.4 million scores

I built a page with thirteen sliders and a single number. People used it. I went back to the dump to see what they actually did.

The extract that ends May 8, 2026 holds 1,441,570 scored rows.¹ A live counter on the homepage has kept climbing since then. I did not infer anyone's 2016 or 2020 vote. Those fields do not exist. What I have is the sliders, the timestamps, and a small optional survey in which 14,037 people described themselves as Left, Moderate, or Conservative.

Here is what is worth remembering.

It was thirteen sliders and one number. Almost everyone showed up in 2018 and 2019.

The inputs were race, sex, gender, ability, age, richpoor, christian, muslim, jewish, native, english, education, and orient. Not fourteen. The output was one integer from 0 to 100.

77.9% of the scores landed in 2018 and 2019. The site did not go quiet after that. It kept reading, smaller, through 2026. If you came here expecting a 2020-2022 experiment, the clock is wrong. This is a late-2018 instrument with a long tail.

The typical visitor was told they were the privileged one.

Among 1,154,230 rows with a computed score, the median was 26 and the mode was 12. 44.1% of those scores sat between 5 and 25. The page did not flatter the room that showed up. Most people left with a low-twenties number and a sentence that said most other people were more marginalized than they were.

Almost nobody left more favorable toward intersectionality.

14,037 people took the optional three-question survey, about 1% of the score rows, most of them in 2019. They were asked whether they had heard of intersectionality, how they felt about it after the test, and whether they considered themselves Left, Moderate, or Conservative. That is self-report after a polarizing page. It is not a model of the electorate.

Of the 13,606 who answered the feeling question: 33.8% less favorable, 54.5% about the same, 11.7% more favorable. Conservatives: 58.8% less favorable. Liberals: 63.8% unchanged. The thermometer measured the prior, then handed it back.

Left and Conservative described different lives. Race was not the gap.

Two tribes used the same thirteen sliders and did not report the same person. Left versus Conservative means: sex 59.8 vs 32.4, orient 54.7 vs 24.1, christian 20.9 vs 52.0. Race was 32.6 vs 26.4. The loud slider on the homepage is not the loud difference between the people who filled it in.

Treat this as a 1% sample of people who opted into a survey after the calculator, not as America.

Peterson week filled the room. Floyd left a pulse. The election was silent.

The week after Jordan Peterson shared the page, November 30 to December 7, 2018, produced 140,307 scores. The peak day, December 1, was 39,125. The George Floyd fortnight, May 25 to June 8, 2020, produced 18,009: a real pulse, an order of magnitude smaller than launch week. The 2020 election fortnight produced 1,989. Election day itself: 129.

Unrest left a fingerprint. The election did not. The instrument's real audience arrived because a tweet did.

I built this to see what both sides did with the same vocabulary. They showed up. They moved the sliders. They mostly did not change their minds. They were not the same person on sex, orientation, or Christianity. They were closer on race than the page design would make you expect.

Take the calculator Old statistics page