Why is this number everywhere?

TL;DR

People choosing a “random” number from 1 to 100 show a strong, repeatable bias toward 37, even in a 200,000-response survey.

Briefing Cornell Notes

Briefing

People asked to pick a “random” number between 1 and 100 overwhelmingly land on 37—so consistently that it stops looking like coincidence and starts looking like a shared mental shortcut. Across dozens of cultures, psychologists have long noted a similar bias in simpler choices: people reliably choose blue and 7. For two-digit numbers, 37 has been the suggested analog. To test that idea at scale, researchers behind the investigation surveyed hundreds of people directly and then ran a much larger experiment, collecting 200,000 responses after posting a community prompt asking for a random number from 1 to 100.

The results barely budged as the sample grew, indicating the distribution isn’t just a quirk of a small group. Even when the analysis set aside obvious artifacts—like the question’s anchor at 1 and 100, and the fact that 42 and 69 are culturally “not random”—a short list still rose above the rest. The most frequently chosen numbers were 7, 73, 77, and especially 37. When the prompt was changed to ask people to name the number they thought the fewest others would pick (a way to reduce favorite or lucky-number effects), the pattern sharpened further: 73 and 37 became the top picks, nearly tied.

That leaves a paradox. If 37 and 73 were truly random, they wouldn’t dominate. Yet the data suggest they do. One explanation is psychological: humans don’t perceive randomness the way probability theory does. People tend to find certain digits and structures “more random,” such as odd numbers and especially sequences built from 3s and 7s. The survey’s digit-level results matched that intuition—3 and 7 were the most selected digits across both questions.

A second explanation is mathematical, tying 37 to how primes behave. Primes feel random because they’re hard to predict from a simple formula and because they appear less often in everyday composite structures. But 37 gets extra attention: it emerges as the median value for the “second smallest prime factor” across all integers. In other words, if every number is factored and the second-smallest prime factor is tracked, 37 sits at the balancing point—half of integers have a second prime factor of 37 or less. The argument also notes that 37 is an “irregular prime” and appears in multiple named prime categories, though mathematicians sometimes treat those labels as partly playful.

Beyond why 37 is chosen, the transcript links the number to decision-making. In the classic “secretary problem,” the optimal strategy for selecting the best option among sequential candidates is to reject the first fraction 1/e of choices, then pick the next candidate better than all seen so far. Since 1/e is about 37%, the “37% rule” becomes a practical decision heuristic—whether the candidates are job applicants, life partners, or any situation where you must commit without seeing the future.

Finally, the investigation broadens into cultural accumulation: 37 appears on products, measurements, lottery figures, serial numbers, and even in a long-running personal collection of “37” sightings. The through-line is that 37 functions as both a social magnet and a cognitive tool—an “everywhere” number that people treat as random, even as probability and perception drift apart.

Cornell Notes

Across large surveys, people asked to choose a “random” number from 1 to 100 repeatedly select 37, along with a small set of related values like 7, 73, and 77. The pattern persists even when the question is redesigned to reduce personal favorites by asking for the number fewest others would pick. One reason is psychological: humans judge randomness using heuristics that favor oddness and especially digits like 3 and 7. A mathematical thread also elevates 37: it is the median of the second-smallest prime factor across all integers, making it a natural “balance point” in prime-factor behavior. The number’s practical relevance shows up in the secretary problem, where the optimal stopping rule rejects about 37% of options before choosing the next best one.

What evidence shows that 37 dominates “random” choices, and how was the test scaled?

The investigation began with small, on-the-spot prompts and then moved to a large online survey. After posting a community request asking for a random number between 1 and 100, it collected about 200,000 responses. The distribution stayed remarkably consistent as response counts grew (from roughly 10,000 up to 200,000). Even after accounting for artifacts—like the anchoring effect of including 1 and 100 in the prompt, and excluding culturally “special” non-random numbers such as 42 and 69—the most selected numbers still included 7, 73, 77, and 37. A follow-up prompt asking people to pick the number they thought the fewest others would choose produced an even clearer result, with 73 and 37 nearly tied for the top spot.

Why does asking for the “fewest others would pick” still produce 37 and 73?

That second prompt is designed to counteract personal lucky numbers and preferences by shifting the goal from “what feels random” to “what feels least common.” Yet the results still concentrated on 73 and 37. The implication is that the bias isn’t just favoritism; it reflects shared intuitions about what looks random or what other people will avoid. The transcript links this to digit-level heuristics: 3 and 7 were the most selected digits across both questions.

How does the transcript connect perceived randomness to prime numbers and to 37 specifically?

Primes are argued to feel random because they appear less frequently in everyday life (most things are composite products of smaller factors) and because there’s no simple exact formula for where the next prime occurs. The transcript then adds a sharper claim: if every integer is factored and the second-smallest prime factor is tracked, 37 becomes the median value. That means half of integers have a second prime factor of 37 or less. The reasoning uses probability over divisibility patterns (e.g., how often a prime is the second smallest factor) and shows the cumulative probability approaches a balancing point at 37.

What is the “37% rule,” and where does it come from?

The 37% rule comes from the secretary problem (also called the marriage problem). Candidates arrive sequentially, and once someone is rejected, they can’t be chosen later. The optimal strategy is to reject the first S candidates to learn the range, then select the first later candidate who beats all previous ones. Maximizing the probability of selecting the best candidate yields S/N ≈ 1/e, which is about 37%. So the best chance comes from rejecting roughly 37% of options and then choosing the next record-breaker.

What cultural or practical “everywhere” examples reinforce the idea that 37 is a social magnet?

The transcript includes a long-running personal collection of 37 sightings: product quantities (like Nutri-Grain granola bars with 37 grams), measurements (a 37-inch yardstick), lottery references (a Texas state lottery figure described as $37 million), and other mundane artifacts such as serial numbers, jersey numbers, and even a staircase with 37 steps. It also mentions a “37 Force” magic trick that relies on audience members selecting 37 after a structured prompt.

How does the transcript address the possibility that 37 is just a learned or “expected” answer?

It acknowledges that people might guess 7 because they expect a one-digit range, and it notes that the most common single-digit answer in earlier work is 7. But the two-digit survey results still elevate 37 even when the task is explicitly “between 1 and 100.” The persistence of 37 across large samples and across a prompt designed to reduce common-choice effects argues against it being only a narrow expectation about the question format.

Review Questions

What two survey prompts were used, and how did the top answers differ between them?
Explain the intuition behind why primes can feel random, then summarize the specific mathematical reason 37 is singled out.
In the secretary problem, why does the optimal strategy reject about 37% of candidates before making a selection?

Key Points

1
People choosing a “random” number from 1 to 100 show a strong, repeatable bias toward 37, even in a 200,000-response survey.
2
The bias persists when the task changes to “pick the number you think the fewest others will pick,” suggesting it’s driven by shared intuition rather than personal luck.
3
Human judgments of randomness favor certain structures—especially odd numbers and digits like 3 and 7—matching the survey’s digit-level results.
4
A mathematical mechanism elevates 37: it is the median of the second-smallest prime factor across all integers.
5
The secretary problem turns 37 into a practical decision rule: reject about 37% of options, then choose the next candidate better than all previous ones.
6
Cultural artifacts and “37” collections reinforce how quickly the number becomes salient, making it feel both random and familiar at once.

Highlights

A 200,000-person survey found that 37 remains near the top even after removing obvious non-random anchors and culturally special numbers.

When asked to choose the number fewest others would pick, 73 and 37 nearly tied for first—tightening the case that the bias is shared.

37’s mathematical role is tied to prime factor structure: it becomes the median second-smallest prime factor across all integers.

In the secretary problem, the optimal stopping point is about 1/e ≈ 37%, turning a “random number” into a decision-making threshold.

Topics

Random Number Bias
Prime Factors
Secretary Problem
Perceived Randomness
Cultural Numerology