Why is this number everywhere?

In Buddhism, the number 37 holds significance as it relates to the “37 factors of enlightenment” or “37 Bodhipakkhiyādhammā” in Pali, which are qualities or practices conducive to awakening or enlightenment.

As per AN7.71 These factors are considered essential for the progress towards and attainment of enlightenment.

In the spirit of the Watercooler section, I share the video above.

It talks about the consistency of 37 as a “random” number choice indicates a potential universal bias. Experiments varying demographics could help uncover the origins - whether mathematical, psychological or other factors shape this bias universally.

The video suggests 37 may feel unpredictably special, but are there rational cognitive processes underlying this, like primacy/scarcity effects? Studying number cognition could provide answers.

Knowing typical human stopping points could influence polling/survey design by suggesting optimal sample sizes. The video argues, mathematically, that 37% may represent the sweet spot where additional data provides diminishing returns.

The inherent interest in finding patterns in randomness, as shown through the number 37 fascination, taps into humankind’s need for explanatory models even where none exist fundamentally. Could this signify an evolutionary advantage?

While the number demonstrated widespread subconscious familiarity, were participants able to articulate reasons for its appeal? Bridging the gulf between implicit and explicit knowledge merits investigation!

The decision-making applications are practical, but do intrinsic biases always serve us? When do they become detrimental?

Larger sociological forces could be at play too. For instance, how might 37 popularity relate to broader cultural phenomena like the appeal of lucky numbers?

More philosophical questions emerge too - what does the perceived significance of 37 say about humanity’s relationship with randomness and patterns?



There’s also 42, of course, which begs the same global questions but not in relation to the suttas. Douglas Adams:

It was a joke. It had to be a number, an ordinary, smallish number, and I chose that one. Binary representations, base thirteen, Tibetan monks are all complete nonsense. I sat at my desk, stared into the garden and thought ‘42 will do.’ I typed it out. End of story.

Which is my way of remembering that speculation about randomness may be a distraction from the plain, obvious realities which explain human suffering and happiness.

But the thought experiment part is cool :sunglasses:!

Namo Buddhaya!

It’d be difficult to set up a signigicant experiment proving that people are biased towards a special number.

The unbiased average is supposed to be 1%. Just imagine how many people you have to poll until people have picked all the numbers 1 to 100 at least once, then imagine how long it’d take for people to pick all numbers at least twice, then how much more for the distributions to average out to be between 0.5-1.5% and how much more to get all frequencies to be 0.9-1.1% with >95% confidence.

Unless the weight is very heavy, a sample of even 10 000 people is meaningless. It’d be like asking only one person and using that as evidence of bias towards that number.

Yes it might be evidence of bias but the uncertainty interval is >99.999%

I think they probably make these videos just for views because people are generally interested in numbers & geometry but do not want to study mathematics, and so someome exploits that for gain.

Confidence intervals aren’t used on every single possible value of the dataset, otherwise how would we have been drawing conclusions from continuous data like people’s weight. You have to wait until many people measure in at exactly 108.83 lbs? and then 108.84? Confidence interval is used with a single generalization like a correlation. Like what percentile is your weight, and how confident are you that it’s in that percentile? Same applies to discrete data with a huge rarely-repeating range like city populations. This isn’t that different of a situation as this if you adapt those into a histogram.

In this case, you can use Kolmogorov–Smirnov test with the expectation that it would be uniform, which it is clearly extremely far from being if you look at the video of 200,000 respondents, match it with a binomial distribution, or use the proportion who “answered 37” VS the expected chance 1/100, which I will do here:

Proportion Comparison Statistic & Binomial Probability

I’m going to estimate to the nearest 100 because the video doesn’t give exact numbers, but that shouldn’t make an important difference.

Respondents: 203,000
Filtered out respondents: 1 5,900 + 2 6,000 + 42 5,000 6911,000 + 99 4,000 + 100 2,000 = 33,900
Note filtering data should not change the result at all, because it should still be a random number no matter which how many are available. Leaving it in would also barely change the result in this process.
Filtered population (n): 169,100
Number of filtered options: 100-6 = 94 numbers
Expected probability of any number (p): 1/94 = .010638
Number of 37 responses: 4,100 (I used a pixel counter in ms paint)
Probability 37 was picked (q): 4,100 / 169,100 = .024246

Null Hypothesis H0: p = 1/94
Alternative Hypothesis Ha: p > 1/94

We have to do some tests to make sure the data are significant: np = 1799 > 10; n(1-p) = 167,301 > 10

Z score (the standard deviations above the mean) = (q-p)/sqrt(p(1-p)/n) = 54.544

For the P-Value, no calculator goes that high, but using some approximation techniques, I got the P-value to be at most 10^-1000, <<<<< 0.01 confidence interval. That’s the most statistically significant result I have ever seen anywhere. It’s like if you picked the same random atom as me out of 12 universes on the first try. Imagine somehow picking the same random one out of a single mustard seed.

Binomial Distribution:
= 10 ^ -527

This data isn’t uniform, in fact you can see the patterns if you sort it by digit oddness. You can practically model it by weighing each number formulaically count(7 → 3 …). There’s jumps every 10 on the 7’s, every 11 on the doubles, and valley’s on the 0’s. It’s an extremely wild yet precise graph.

Combining sin waves to create a model.

Conclusion: The number of people who picked 37 in those data, compared to the expected uniform chance, is +54.5 standard deviations, which is literally unimaginably statistically significant, and this is accounting for how confident we’re allowed to be with the given sample size.

The binomial probability is 10^-525 %, the chance that all those people said 37 if the probability were uniform.

Also, how could they forget the fine-structure constant, 1/137? And there are even more like the smallest magic square has 37 in the middle. To be honest, there’s no deeper meaning of 37 anywhere. People pick it because “odd/prime = random” without thinking straight as the video suggests. Our usage of base 10 numbers is 84% per"cent" arbitrary (84 base 12 = 8 * 12 + 4 = all very good numbers = 100% arbitrary).

1 Like