http://www.predictablyirrational.com/?page_id=117

A game :) Be sure to hit "More Info" at the end.

Didn't work on me. :P I guess I'm either rational or very stubborn!

Funny, on the way in today NPR was just advertising how they were going to have an article on Predictable Irrationality on All Things Concidered. I thought this was coincidentally about that.

Haha, I was chastised for being rational.... I didn't know that was a bad thing.

hehe, did not read the instructions all that well. I ended up clicking all the doors looking for points, thinking it was a search for the door with points they would just open up, then I happened to double click and there they were. Then, I just stuck with it!

Funny, on the way in today NPR was just advertising how they were going to have an article on Predictable Irrationality on All Things Concidered. I thought this was coincidentally about that.

Haha, I was chastised for being rational.... I didn't know that was a bad thing.
I'll have to look for it. Tierney is writing about it in his next article. I'll link it when he does.

It didn't work for me either. I made a rational choice and the software told me otherwise. I tried it again with the same rational strategy and it told me I was rational after all.

Oh well. Perhaps I shouldn't expect much scientific rigour from an internet game :)

I was irrational, I didn't want the doors to go away. :)

You switched an equal of times when the doors were shrinking.
More frequent switching when the doors are shrinking is most likely a reflection of an irrational tendency to keep your options open.
This tendency cost you points in the game, but it could have much larger implications in life...

This is the message the second time through. I did not switch at all. Just kept clicking door number 1.

I can not figure out the point of this little exercise. Unless they are storing info in a database, like that old 20 questions internet game. Which the programmers/webmasters used the learned algorithms to make a crappy little Simon game, to sell at WalMart.

I think the point of the game was to demonstrate that people make irrational decisions based on things like "keeping their options open" even when doing so would lead to a reduced outcome. I expect Tierney will have a pretty good article on it next Tuesday.

Why is keeping options open, considered irrational?

That is very rational behavior in humans.

The first time I played.

I clicked on closed doors like 20 times before I clicked on an open door to receive my little Skinner Box morsel.

Who decided that was rational or irrational?

Who decided that it was rational or irrational to have to click on an already opened door to get a morsel? Wouldn't it be more rational to have a morsel when you open the door, instead of clicking an already open door?

Rationally speaking, it would be more logical to get your reward when you open the door.

I think the idea is that once you discover that only open doors give you points, clearly the rational choice is to click on the door that gives points (the one that is already open).

It is not a Skinner box because you get a "reward" (points) _every_ time if you make the right choice, except for the very first click of course.

Well, you do get random, or apparently random numbers.

One of the behaviors I noticed when I was chosing, was, if I got a 65, 55, then say a 22(on open doors obviously)...I switched doors immediately, initially. Thinking that the pattern was trending downward.

Later, it appeared random.

And later still, when I chose the single door, during the shrinking door phase, it appeared to me that no small numbers were popping up.

Obviously, many of these observations are like 'my new car syndrome', where when you buy a new car, everyone on the road now seems to be driving the same make and model. If you are looking for it, you will see it more often.

Additionally, by the time you play the second round, you have figured out that you must click on open doors. So any comparison between the two rounds is flawed. Unless that particular bit of information or behavior, is really what is being tested and catalogued.

Right, but a random positive number is always greater than 0 (which is what you get when you click a closed door).

I think the problem is that you don't have the opportunity to experiment with the doors first to see how they work. It is only possible to make the rational choice when you understand what is going on. Until then, you're just guessing.

I think the idea is that once you discover that only open doors give you points, clearly the rational choice is to click on the door that gives points (the one that is already open).

It is not a Skinner box because you get a "reward" (points) _every_ time if you make the right choice, except for the very first click of course.

But they explain up front how it works. You have the knowledge that only open doors give you points. So if you didn't read the instructions then I guess that's predictable incompetence. :p

Switching doors, now that I think on it, is always going to mess you up because when you switch to a new door you waste a click opening the door before any points come out.

I guess that's predictable incompetence.

Predictable indifference.

But they explain up front how it works. You have the knowledge that only open doors give you points. So if you didn't read the instructions then I guess that's predictable incompetence.

Switching doors, now that I think on it, is always going to mess you up because when you switch to a new door you waste a click opening the door before any points come out.

Each door, according to the instructions, has a different distribution of points. So a question arises of how many times you have to sample each door to determine how to use the remaining clicks to get a maximised score. Sampling each door at least once seems appropriate, as you don't know if each score is in tens, hundreds, thousands, whatever. I suspect that as you don't know anything about the distributions, except that they are different, there won't be any simple answer about how you should sample each distribution. Perhaps you would need to find a philosophical solution instead of a mathematical one. Most of us would have to use our intuition, not logic.

Let's assume, rightly or wrongly, that you've sampled each door and decided upon one door to continue using until the end. If your chosen door is the last door you sampled then you don't lose a click re-opening the chosen door. So if you repeat your test many times, even with the same rational strategy, then some tests could be adjudicated more 'predictably irrational' than the others just due to this random feature of the test.

That is a good point. However, it requires no more than two door switches to sample all 3 doors and determine (intuitively) that the expected value of each is roughly the same.

That is a good point. However, it requires no more than two door switches to sample all 3 doors and determine (intuitively) that the expected value of each is roughly the same.

Roughly the same in what way? They can vary by at least 50% and you only spend 2% of your clicks in changing from one door to the next. They only seem intuitively similar in relation to a scale you've arbitrarily introduced to the problem. No upper or lower limit is specified for the distributions.

In fact, you might like to consider how the problem changes if the first number shown was negative.

Whether you encounter a negative number is irrelevant. All you should be trying to determine (approximately) is the expected value from each door. After noticing that the expected value is about 50 from all 3 doors, obviously the rational thing to do is to stick with whatever door is currently open, because 50 > 0.

Whether you encounter a negative number is irrelevant.

If you think the average door value is negative then your strategy changes and you always change doors! You wouldn't want to add more (negative) value to your total.

Why is keeping options open, considered irrational?

That is very rational behavior in humans.

In this case, it was points from a door. Any door. It didn't matter, you got some points.

To my thinking, then, it didn't matter if the other doors went away. There was nothing in the new doors I wasn't going to get from the same door.

However, keeping ones options open isn't irrational. Neither, in this case, is just clicking the one door that is good enough. Perhaps it would be better if the test had higher stakes, but we can't go around killing the test subjects.

Why is keeping options open, considered irrational?
It only is irrational in the context of this particular decision. Sometimes it is a good thing, sometimes you need more information. I think they're just trying to highlight a basic human behavior.

When the value of the consequences of three different options(or paths of behavior) is unknown, I believe that it is very rational to keep all options open. And my experience, and experiences, backs up this observation and belief.

As said, this particular experiment was set up to show the opposite. It obviously had the outcome predetermined before it was set up. To which the end is obviously to set up the promotion of the guy's book, and its sales.

I find the idea of Games Theory much more interesting, anyway. It is more complex and insightful. Allows for better prediction, especially when the outcome, and value of the outcome, is less known.
http://en.wikipedia.org/wiki/Games_theory

And while most women will go "Duh" at the whole concept, or at least most successful women will. As a man, the idea, initially was counter intuitive. So to me, when something is learned, that makes it a superior proposition. I learned nothing from the door experiment.

It only is irrational in the context of this particular decision. Sometimes it is a good thing, sometimes you need more information. I think they're just trying to highlight a basic human behavior.

I don't respect their judgement of when a person is irrational since their testing program, which they presumably designed themselves, seems inadequate to do so. It is irrational to use a test that throws up false results.

I too find Game Theory very interesting. It's a big earner for the mathematicians who understand it as well. Perhaps I should read a good book about it instead of raking over this woeful example on a messageboard :).

The Advantages of Closing a Few Doors (http://www.nytimes.com/2008/02/26/science/26tier.html?_r=1&n=Top%2fNews%2fScience%2fColumns%2fFindings&oref=slogin)

I haven't read it yet.

“We may work more hours at our jobs,” Dr. Ariely writes in his book, “without realizing that the childhood of our sons and daughters is slipping away.

From your last link.

Seems a contradiction. To which there are many in this experiment.

If we are to believe the pre conclusions of the experiment, it should be perfectly ok to let the childhood shrink, for it is only one, in this case, of two options.

If keeping options open is inferior, then why NOT let your kids childhood shrink.

You still have your job, the open door.

I don't think they ever said any where that all doors had equal value. If you value work over children then sure, click the work door and let the children disappear.

So you quit your job to spend time with the kids?

You're a master a missing the bigger picture and getting hung up on irrelevancies.

I agree with you Pan, the point of the door experiment is completely irrelevant.