Brady's Videos and Benford's Law - Numberphile
video description
It requires a greater percentage increase to change a leading digit from a 1 to 2. just as it takes a greater percentage increase to change a leading digit from a 2 to a 3 and so on. There is no natural growth in terms of a video's length so one wouldn't expect the duration figures to follow the law.
Date: 2022-04-08
Related videos
Comments and reviews: 9
ZER0
The reason 3 comes out top in duration is this. It is the same as singles ie pop records. I haven't cheeked but I would put my life on it, that most pop songs are around 3 minutes. I think its because you are telling a story but in a very short time on just one theme.
I don't know if that is the reason but I promise you if Benfords law is applied to the duration of pop songs (since the format began, then 3 (minutes) WILL be the top number. Im now going to try and find if my hypothesis is correct. It definitely be 1 or 2 or 5.
reply
The reason 3 comes out top in duration is this. It is the same as singles ie pop records. I haven't cheeked but I would put my life on it, that most pop songs are around 3 minutes. I think its because you are telling a story but in a very short time on just one theme.
I don't know if that is the reason but I promise you if Benfords law is applied to the duration of pop songs (since the format began, then 3 (minutes) WILL be the top number. Im now going to try and find if my hypothesis is correct. It definitely be 1 or 2 or 5.
reply
fffffff
People are saying the duration graph doesn't fit for the reason that you make them a specific length. But I noticed that if you cut the graph after leading digit 2, so that the first leading digit in the graph is 3, then it looks pretty benford-esk. So Benford's law looks like it is a measure of preferences. That's why it has to be measured in the context of something, like population, video duration, or distance to a venue. If you grabbed a giant random sampling of numbers then it would fall apart.
reply
People are saying the duration graph doesn't fit for the reason that you make them a specific length. But I noticed that if you cut the graph after leading digit 2, so that the first leading digit in the graph is 3, then it looks pretty benford-esk. So Benford's law looks like it is a measure of preferences. That's why it has to be measured in the context of something, like population, video duration, or distance to a venue. If you grabbed a giant random sampling of numbers then it would fall apart.
reply
GibsnRage
Trying to get my computer to generate numbers like this. and I actually got pretty close, surprisingly. With simple random) functions, though, I always get 11% probabilities for each digit. I did some convolution to the basic random functions, and managed to get a distribution like: 18 / 13 / 12 / 7 / 13 / 9 / 10 / 8 / 6
Still imperfect, but it's beyond me to try and improve that further. Time to look into the vast community of people out there working on improving random number generators!
reply
Trying to get my computer to generate numbers like this. and I actually got pretty close, surprisingly. With simple random) functions, though, I always get 11% probabilities for each digit. I did some convolution to the basic random functions, and managed to get a distribution like: 18 / 13 / 12 / 7 / 13 / 9 / 10 / 8 / 6
Still imperfect, but it's beyond me to try and improve that further. Time to look into the vast community of people out there working on improving random number generators!
reply
Psycho
Benford's law applies to unintentional distributions. Most likely, you have chocen an intentional average duration for each video, one long enough to provide a sufficient understanding but not too long that would make a video tiresome. That distributuion happened to be in the 4s. View counts and comments are unintentional, gathered from random people who don't care about the actual number. The law applies to gathered sums, not intentional choices
reply
Benford's law applies to unintentional distributions. Most likely, you have chocen an intentional average duration for each video, one long enough to provide a sufficient understanding but not too long that would make a video tiresome. That distributuion happened to be in the 4s. View counts and comments are unintentional, gathered from random people who don't care about the actual number. The law applies to gathered sums, not intentional choices
reply
Nienke
although your sample size is quite large, your distribution is quite limited and probably spans at most 2 orders of magnitude, next to that, your format will mean that there is a certain optimal time for a video, too long and people will tune out, but too short means you can't fit the amount of info you'd like into the video, and what you end up with is a distribution centred strongly around a certain duration
reply
although your sample size is quite large, your distribution is quite limited and probably spans at most 2 orders of magnitude, next to that, your format will mean that there is a certain optimal time for a video, too long and people will tune out, but too short means you can't fit the amount of info you'd like into the video, and what you end up with is a distribution centred strongly around a certain duration
reply
Ratstail91
You, as a creator, generally aim for an entertaining video, which generally can't be done with only 1-2 minutes, so naturally, your efforts skewed the benford curve for video duration. However, it's still notable that there -is- a curve, starting from 3, which seems like a decent length for your average video. Therefore, I don't think your video durations violate benford's law at all; it simply skews it.
reply
You, as a creator, generally aim for an entertaining video, which generally can't be done with only 1-2 minutes, so naturally, your efforts skewed the benford curve for video duration. However, it's still notable that there -is- a curve, starting from 3, which seems like a decent length for your average video. Therefore, I don't think your video durations violate benford's law at all; it simply skews it.
reply
gt
Well clearly the duration won't be so accurate as most of your videos last from 5-6 minutes (300 - 360s) so that's why it falls mostly in the '3' category. 1 doesn't come out as often because 100-199s is 1. 6min - 3. 3min and no that's not going to usually happen. 1000-1999s is also unlikely as 16. 6-33. 31min is not probably going to happen. Thus 1 is not top.
This is my reasoning
reply
Well clearly the duration won't be so accurate as most of your videos last from 5-6 minutes (300 - 360s) so that's why it falls mostly in the '3' category. 1 doesn't come out as often because 100-199s is 1. 6min - 3. 3min and no that's not going to usually happen. 1000-1999s is also unlikely as 16. 6-33. 31min is not probably going to happen. Thus 1 is not top.
This is my reasoning
reply
Hendrik
First of all, there are 1249 comments right now (starting with a 1 of course.
Now to the duration.
To start with a 1 your video hast du be longer than 1, 67 minutes and shorter than 3. 3167 minutes, or longer than 16, 67 minutes long.
but only 2 out of the 20 video suggestions I get after viewing this fall into those categories.
Simple as that.
reply
First of all, there are 1249 comments right now (starting with a 1 of course.
Now to the duration.
To start with a 1 your video hast du be longer than 1, 67 minutes and shorter than 3. 3167 minutes, or longer than 16, 67 minutes long.
but only 2 out of the 20 video suggestions I get after viewing this fall into those categories.
Simple as that.
reply
AMRosa10
A few of the users over at Stackoverflow looked at the reputation figure for all of the members on the site (Reputation points are awarded or lost for answering questions or posing questions, and when they filtered out the single-digit reputations, the numbers pretty much plotted exactly to Benford's Law.
reply
A few of the users over at Stackoverflow looked at the reputation figure for all of the members on the site (Reputation points are awarded or lost for answering questions or posing questions, and when they filtered out the single-digit reputations, the numbers pretty much plotted exactly to Benford's Law.
reply
Add a review, comment
Other channel videos
