How Many Skittles are in this Jar? - YouTube

At a country fair in 1906 in the South of England

They held contest to see who could best guess

the weight of a certain cow they had there.

A lot of people entered the contest.

Over 800, in fact.

What was interesting was that even though their guesses ranged from high to low,

hundreds of pounds off from each other,

A statistician named Sir Francis Galton calculated that the median guess,

so if he ordered the guesses from low to high and looked at the middle one,

was actually within 1% of the actual weight of the cow.

The cow's weight was 1,198 pounds, and the median guess was only 9 pounds too high

He was pretty interested by this phenomenon, and he called it Vox Populi,

which means "The Voice of the People"

And, I'm curious to know if it works, so today we're going to go find out!

I didn't have a cow,

so I did what any other person would do in this situation...

(enter super market)

(browse aisles)

(take skittles)

(checkout)

(cut open skittles)

(pour skittles in jar)

And then I asked a lot of people...

How many skittles...

How many skittles...

How many skittles...

How many skittles...

How many skittles...

How many skittles...

How many skittles...

are in this container?

Here's what they had to say

"I think I'm going to guess 275"

"1,500?"

"Gonna go with 1,353"

"2,500, all the time, keep saying like this"

"2,500, and I'm guessing based on what I feel like is in a bag of skittles"

"maybe... 300"

2,225

"713"

"52"

"3,000"

"2,000?"

"500"

"I'm going to guess 1,750"

"I'm just going to guess 17,000"

"I'm gonna guess 2,000"

"I have no interest in holding it because that actually doesn't do anything for me...

I'm guessing there's 50 per row...

well you have to think in patterns. I think patterns.

it does seem like a lot though...

um...

I'm sorry, is this taking way too long?

'All the time'

1,250 seems too many

um, i'll say 1,250"

'all the time'

Now for some explanation

What I want you to realize is that each of these guesses is independent and individual.

So, no guess is dependent on another guess and that's what makes the Wisdom of the Crowd work supposedly.

The variation that comes between different people

actually helps at the end of the day

to produce an average, accurate, answer.

So, you might think that working together in a group is beneficial,

bouncing ideas off one another.

But that's actually not the case.

What usually happens in a group setting is that

the person with the loudest voice or the highest status

or even just the person who speaks first can sway a whole conversation in the way that they choose

Watch, actually, what happens in this next clip where one of our participants does just that:

bases her judgement off of the two that came before hers

"I'm gonna go with 1,500 because

I want to guess in between the two of them, and that's my answer."

So yes, we can reap the benefits of the crowd's wisdom,

but only if each component and voice of that crowd is independent and individual.

One example of this, exactly, is in the game show "Who Wants to be a Millionaire?"

I'm sure you're familiar with this, but 'Ask the Audience'

is just an example of the Wisdom of the Crowd in action.

"Today, I think this might be a good time to ask the audience."

"I guess I'll probably have to ask the audience."

"I'd like to ask the audience."

"I think, I'm gonna ask the audience."

"Can I ask the audience?"

"Can I ask the audience?"

♩The Next Episode♩

Over the many years that the show has been on air

contests have only gotten the questions right about 65% of the time.

and the audience, on the other hand, has had a 91% success rate.

It's not just in game shows, though, that combined forecasts and average opinions are better than individual ones.

If you go to macroeconomic forecasting, there's one clear example of this too.

One example of this is the Blue Chip Economic Indicators Survey,

which takes the voices of over 70 different economists

and merges them into one aggregate forecast.

It takes voices form academic institutions

to major banks

to consulting firms

and averages their voice out into one consensus.

Over a span of 15 years

this survey has performed better than any one of the individuals who comprise it.

Alright, so I just counted up the number of skittles,

and there were so close to 1,500

and in total

we have 1,499 skittles

and then

four little duds right here.

At this point, you're probably still wondering

did the experiment work today? That's what I want to know.

And, in fact, it did not.

The median guess from our crowd was 1,000 skittles, and as you just saw

the actual amount of skittles was 1,499.

If you asked me why this didn't work, I have a couple ideas.

Perhaps, our samples size isn't large enough.

At 130 people, there's still enough room for variation.

But, 1,000? 1,499? That's still a very big gap.

What went wrong?

Well, one thing that definitely could have been the case is that aside from the median being 1,000,

our mode was also 1,000.

Out of a population of 130, 10% of people interviewed guessed 1,000.

"1,000"

"1,000"

"1,000"

If you think about it, 1,000 isn't an unreasonable guess.

There weren't about 100 skittles in the jar, and there weren't about 10,000 skitlles in the jar.

So, going to 1,000, the closest power of 10 to our answer

is pretty good judgement.

And at that point, I don't know if our

participants had a great enough incentive to refine their guess.

As you see,

sometimes science doesn't work out as expected.

Sometimes you'll find data that confirms a theory, and sometimes you won't.

But, if you think there's a flaw in my experiment, please let me know in the comments below

and I would be very willing to retry this

but otherwise, I hope you enjoy this first video of my channel

If you liked it, like this video.

And if you want to stay in touch and see what's next, click right here to subscribe.

See you then.

'All the time'