"How many jellybeans in the jar?" (Each was wrong, but all were right)

[JGeropoulas]

I just finished watching an interesting 3-part series on Netflix entitled "The Code". The series features mathematics professor Marcus du Sautoy, who demonstrates how "significant numbers appear throughout every aspect of the world, governing all of life".

One of his demonstrations was particularly fascinating to me: He filled a large jar with jellybeans, which he'd counted to arrive at a total of 4,510. He then went around town asking 100 random people to guess how many jellybeans were in the jar. He tallied their guesses, which ranged from 400 to 50,000. When he averaged these 100 guesses, the average guess was 4,515 !

This made me think of the forum principle of needing many observers to establish objective truth. But I can't understand how this applies exactly in this particular experience when these 100 people were guessing apart from and independent of the other 99 (i.e. no networking)? It's a simple experiment, but I can't really wrap my mind around the outcome. What do y'all think about this?

 

[Mark]

I remember reading something about this in the past. I forget where, maybe here. Anyway, it could be that our collective subconscious knows a close approximate to truth.

 

[nicklebleu]

Thanks for posting this, JG. Fascinating - if that could be replicated across a wide range of situations it could be scientif proof for the superiority of networking vs individual action. Well, not that much proof is needed here, where the technique has been demonstrated effective over and over again.

 

[DougEE]

Doesn't this display the Elephant Principle where each observer contributes one aspect of the elephant in the room?

 

[JGeropoulas]

... it could be that our collective subconscious knows a close approximate to truth.

...proof for the superiority of networking vs individual action...

...the Elephant Principle where each observer contributes one aspect of the elephant in the room?

Those were the first thoughts that came to me, but these principles fall apart as explanations of the jellybean experiment.

For example, if each individual's unconscious was networking with the others, it would have to know how many other participants there were, who the other participants were and what their estimates had been, so that their own estimate would mathematically contribute to the final correct average estimate.

But this begs the question: if an individual's unconscious was capable of knowing all those data points I just pointed out, why wouldn't it just know the correct answer? Otherwise one's unconscious would have to know that several others had underestimated 400, in order to come up with the average-maintaining overestimation of 50,000.

As for the Elephant Principle, each man was correct--but about only one part--of the elephant, Only when their individual observations were added did it yield the right answer about the elephant.

In contrast, with the jellybeans, each person was wrong. Only when their individual answers were added--then divided--did it yield the right (averaged) answer.

 

[sedenion]

One possible answer: the unconscious process know the answer (the unconscious process is like insight, intuition) but the concious process will procede as a new layer over the first unconscious process, with its "rational thinking", so the conscious process will try to formulate an answere that appear "good" to some "aparently logical" facts. For example, the first insight is the good answere, but the concsious part will think "hum, maybe i have a tendency to overestimate number, this jar is not so big" or "hum, maybe there is many more than it appear, the jellybeans are small", etc..

From that point, everybody will give an answer around the exact answer, according its own "conscious biases", but if you take sufficient count of answer, then you make an average, you finaly get a very good aproximation, because the fist insight is the good answer: It is like trying to aim at a target, the target center still the same for everybody, but nobody perfectly aim, so you have many dart scattered in the target, and if you make some average of all dart position, you obviously have a good approximation about where is the target's center.

Did you never experienced how you have remarkable ability to aim, to trow a paper in a trash for example, if you don't even try to aim, by just doing it like an automatism ? You just throw the paper, without trying to "rationalize" your aim, and damn ! 3 point !... but if you begin to rationalize (thinking how to make a good shoot, etc), you are unable to reproduce this little miracle...

 

[Chu]

Those were the first thoughts that came to me, but these principles fall apart as explanations of the jellybean experiment.

For example, if each individual's unconscious was networking with the others, it would have to know how many other participants there were, who the other participants were and what their estimates had been, so that their own estimate would mathematically contribute to the final correct average estimate.

I forget what it's called, but there is also a phenomenon through which knowing what others estimate, can make one get it even worse than if they were guessing without any previous knowledge. If they know, for example, that the two previous people guessed 40,000 and 50,000, they might say "30,000" even though originally, they would have guessed 1,000. So, it may be better in experiments like this NOT to know what others guessed.

But this begs the question: if an individual's unconscious was capable of knowing all those data points I just pointed out, why wouldn't it just know the correct answer? Otherwise one's unconscious would have to know that several others had underestimated 400, in order to come up with the average-maintaining overestimation of 50,000.

Is this just a one-time experiment, or was is replicated?

I think that if there is anything to it, it's more about the COLLECTIVE subconscious, rather than the individual one. So, one individual's subconscious wouldn't need to "know" much of anything, OSIT. And in reality, IMO it shows more than anything that most people must have been dead wrong! That their ability to connect visual input with a specific quantity is not very well developed.

As for the Elephant Principle, each man was correct--but about only one part--of the elephant, Only when their individual observations were added did it yield the right answer about the elephant.

In contrast, with the jellybeans, each person was wrong. Only when their individual answers were added--then divided--did it yield the right (averaged) answer.

That may be why when you compare it to networking, it doesn't add up. Take this network. If someone would see X (a topic being discussed) as 400 or 50,000, they would not be very colinear, to say the least. The aim is to get close to the mark in terms of seeing reality as it is, but it's also not random guessing. It involves study, careful observation, working through our own biases, etc. So, it's more about the "collective conscious", I think.

 

[Janek]

It's so called "wisdom of the crowd". From Wikipedia:
[quote author=Wikipedia]The wisdom of the crowd is the collective opinion of a group of individuals rather than that of a single expert.

A large group's aggregated answers to questions involving quantity estimation, general world knowledge, and spatial reasoning has generally been found to be as good as, and often better than, the answer given by any of the individuals within the group. An explanation for this phenomenon is that there is idiosyncratic noise associated with each individual judgment, and taking the average over a large number of responses will go some way toward canceling the effect of this noise.
[...]
The classic wisdom-of-the-crowds finding involves point estimation of a continuous quantity. At a 1906 country fair in Plymouth, 800 people participated in a contest to estimate the weight of a slaughtered and dressed ox. Statistician Francis Galton observed that the median guess, 1207 pounds, was accurate within 1% of the true weight of 1198 pounds. This has contributed to the insight in cognitive science that a crowd's individual judgments can be modeled as a probability distribution of responses with the median centered near the true value of the quantity to be estimated.
[/quote]

Polish Wikipedia also gave this comment:

[quote author=Wikipedia]
Wisdom of the crowd must meet certain conditions - crowd must represent various points of view and their answers can't be coordinated (they have to be given independantly), in other case we are dealing not with wisdom of the crowd, but with stampede
[/quote]
So basically, as far as I understand it, it works when some random guys give a solution to the problem they are not specialist at. It's interesting phenomenon I think, but I'm not sure how to apply it in practice.

 

[Alada]

One of his demonstrations was particularly fascinating to me: He filled a large jar with jellybeans, which he'd counted to arrive at a total of 4,510. He then went around town asking 100 random people to guess how many jellybeans were in the jar. He tallied their guesses, which ranged from 400 to 50,000. When he averaged these 100 guesses, the average guess was 4,515 !

Fascinating! I think the average then can show that our initial 'blink' of a situation can often be very accurate. From experience we know roughly how big a jelly bean is, roughly how many would make a handful, roughly how many handfuls would make a jar – as long as the jar is not exceptionally large, or some unusual shape, or in some other way outside of the range of experience, we can get pretty close on average in that way.

Having said that though, it would be good to know how often the experiment was replicated. Because it also must be possible that the averaged answer was exceptionally wrong. If it is repeatably within a very small margin of error, that would be remarkable. It would be interesting to know what the statistics are.

This made me think of the forum principle of needing many observers to establish objective truth. But I can't understand how this applies exactly in this particular experience when these 100 people were guessing apart from and independent of the other 99 (i.e. no networking)? It's a simple experiment, but I can't really wrap my mind around the outcome. What do y'all think about this?

The comparison fits I think, because sharing and comparing our observations helps us firstly to understand the concept itself, that by gathering many measurements we can achieve far greater results an average, where before maybe we did not know it.

Then we have the problem that many of the things we are trying to perceive are quite large in scale, or in some other way their full and true extent are hidden from us. We each can only bring our observations of a small piece of the puzzle, yet we don’t know until we compare/regress to the mean, what the likelihood is of our readings being accurate or not. Perhaps over time then if we improve the quality of our readings and the system improves, the margin of error get smaller then.

But then sometimes we will still find that we are very wrong, or sometimes we find that everyone was very wrong, up until the point where one piece of vital data was added. For example, you can’t very accurately guess how many jelly beans are in some kind of trick jar, unless you already know through experience or research that it is a trick jar. So everyone is guessing and happily discussing until someone comes along and says "oh it’s one of those trick jars! You should know that the middle is solid, it’s only got jelly beans around the edge pressed up against the glass." Everyone can revise their readings then, and be mindful to watch for similar errors or traps next time. Nothing is certain!

Of course it helps if those in the system are open to changing their minds and re-calibrating their reading instruments based on new data, which is not always a pleasant or straight forward thing to do – see Clinton supporters for evidence of that!

 

[sedenion]

but I'm not sure how to apply it in practice.

By using a good balancing of intuition (unconscious process) and rational thinking... https://youtu.be/5oyZJDBUzWc?t=4m11s This is not about the "Elephant" problem, it is about individual insight disturbed by rationnal thinking (the so called "linear thinking" and "inner dialog" if you want). https://www.youtube.com/watch?v=o2we_B6hDrY

 

[luc]

That may be why when you compare it to networking, it doesn't add up. Take this network. If someone would see X (a topic being discussed) as 400 or 50,000, they would not be very colinear, to say the least. The aim is to get close to the mark in terms of seeing reality as it is, but it's also not random guessing. It involves study, careful observation, working through our own biases, etc. So, it's more about the "collective conscious", I think.

I agree. I read a book years ago by Jaron Lanier called "You Are Not a Gadget". If I remember correctly, he argued against the popular notion of 'crowd wisdom', especially on the internet. I think it's true - there is no such thing. Or as Gurdjieff put it, it doesn't matter how many machines you have, the result will still be mechanical. Gustave LeBon also wrote an excellent critique of the so-called crowd wisdom over a century ago (thread).

Maybe it's more about strengthening the individual (knowledge, being etc.), while at the same time having individuals working together (giving up self-importance) that leads to the 'magic' of networking.

As for the experiment, if true, I don't have an explanation either - maybe it's just a statistical phenomenon? Maybe on average, people are not so bad at guessing such things, but the spectrum is very wide (i.e. there are many people still that get it completely wrong)? That way, when you have a very large sample, you would approximate the right number, but with only a small sample, chances are you get a completely wrong answer?

 

[Gawan]

That may be why when you compare it to networking, it doesn't add up. Take this network. If someone would see X (a topic being discussed) as 400 or 50,000, they would not be very colinear, to say the least. The aim is to get close to the mark in terms of seeing reality as it is, but it's also not random guessing. It involves study, careful observation, working through our own biases, etc. So, it's more about the "collective conscious", I think.

I agree. I read a book years ago by Jaron Lanier called "You Are Not a Gadget". If I remember correctly, he argued against the popular notion of 'crowd wisdom', especially on the internet. I think it's true - there is no such thing. Or as Gurdjieff put it, it doesn't matter how many machines you have, the result will still be mechanical. Gustave LeBon also wrote an excellent critique of the so-called crowd wisdom over a century ago (thread).

Maybe it's more about strengthening the individual (knowledge, being etc.), while at the same time having individuals working together (giving up self-importance) that leads to the 'magic' of networking.

What is important here is to openly network and sharing observations where people could get as close as possible to the truth. And I would also guess that it is more about collective unconscious or information field. Still not a single got it right only the average of all people.

 

[JGeropoulas]

Thanks for all the great comments and insights! So many exciting aspects to the "holiness of true existence"!

 

[Phill4]

One possible answer: the unconscious process know the answer (the unconscious process is like insight, intuition) but the concious process will procede as a new layer over the first unconscious process, with its "rational thinking", so the conscious process will try to formulate an answere that appear "good" to some "aparently logical" facts. For example, the first insight is the good answere, but the concsious part will think "hum, maybe i have a tendency to overestimate number, this jar is not so big" or "hum, maybe there is many more than it appear, the jellybeans are small", etc..

It reminds me of the savant people who have the abilities to pick on this kind of infromation, their mind is wired in such way they associate things differently than a regular brain.
We pick this kind of information at some level to which we have no access to in consious awareness


I heard also about this experimen somewhere , Although, it is true the experimen needs to be done with people being separated and repeated to draw more accurate data.

 

[sedenion]

It reminds me of the savant people who have the abilities to pick on this kind of infromation, their mind is wired in such way they associate things differently than a regular brain.
We pick this kind of information at some level to which we have no access to in consious awareness

You can have access. What you call "consious awareness" (i.e. rationnal thinking -> linear thinking) only "obliterate" it, because the "rationnal thinking" is noisy and don't "understand" the "logic" of the so-called "unconscious" part... To train yourself, and see how your "rationnal thinking" is dominating, you can train yourself by creating totally non-sens sentences but grammatically correct, the fast as possible (i.e. without rationalizing). It's like poetry, something between poetry and Exquisite corpse. The more it is quick and absurde, the better it is... if the sentence have a sens, or come not quickly, this is not good: this mean you rationalized. All that is related to creativity...