06-26-2016, 05:32 AM
no- there's a number in statistics called a standard deviation. it isn't really arbitrary, its a well established and important parameter when talking about these things
let's say there's an apartment of 50 people. none of them know each other well, but they all have opinions on whether to repaint the outside of the building pink or green. you take a vote from all 50 people, and 40 or them say green and 10 say pink.
if you keep asking them, opinions will change. people will think about it more, people will lobby, people will move out and new people with new ideas will move in. i know this, but lets pretend we can go back in time, groundhog day style, to get all the data we need without each "trial" interfering with one another
you would notice that on some iterations, it isn't 40/10. it's 41/9, 42/8, 38/12. this is because there is a "fuzz" around that average, which we call the standard deviation. if one day we hit 41/9, we would go "eh yeah thats about 40/10" and paint it green. no matter which way the "standard" (mean) blows (deviates), we are still green > pink because in this case, the standard deviation is about plus or minus 2. what this means in statistical talk is that it is very likely that the result will be within the bounds of ONE standard deviation, and the "tipping point" (25/25) is well, well, well outside of the bounds of one or two standard deviations
if instead the initial vote was 26/24, it would be a victory for green. but if we were curious enough to relive that day and vote again, and again, and again, the standard deviation would blur at 26/24 and 27/25, a victory for green, but also sometimes 25/25 (a tie?) and even 24/26 which would make it pink
my argument is that this number, which is very much not arbitrary at all, may be ~2 or 3%. and if it is, it means that if superman turns back a few days to save lois lane and we have to do the vote over, it is possible it would come out 49/51 and its decided that the UK stays IN the european union
let's say there's an apartment of 50 people. none of them know each other well, but they all have opinions on whether to repaint the outside of the building pink or green. you take a vote from all 50 people, and 40 or them say green and 10 say pink.
if you keep asking them, opinions will change. people will think about it more, people will lobby, people will move out and new people with new ideas will move in. i know this, but lets pretend we can go back in time, groundhog day style, to get all the data we need without each "trial" interfering with one another
you would notice that on some iterations, it isn't 40/10. it's 41/9, 42/8, 38/12. this is because there is a "fuzz" around that average, which we call the standard deviation. if one day we hit 41/9, we would go "eh yeah thats about 40/10" and paint it green. no matter which way the "standard" (mean) blows (deviates), we are still green > pink because in this case, the standard deviation is about plus or minus 2. what this means in statistical talk is that it is very likely that the result will be within the bounds of ONE standard deviation, and the "tipping point" (25/25) is well, well, well outside of the bounds of one or two standard deviations
if instead the initial vote was 26/24, it would be a victory for green. but if we were curious enough to relive that day and vote again, and again, and again, the standard deviation would blur at 26/24 and 27/25, a victory for green, but also sometimes 25/25 (a tie?) and even 24/26 which would make it pink
my argument is that this number, which is very much not arbitrary at all, may be ~2 or 3%. and if it is, it means that if superman turns back a few days to save lois lane and we have to do the vote over, it is possible it would come out 49/51 and its decided that the UK stays IN the european union
![[Image: Sd9744R.png]](http://i.imgur.com/Sd9744R.png)