It just goes to show that these so called "academics" are willing to pour millions of dollars into whatever silly little project they can think of.

Posted By: Adinsx

But this isn't just anyone, it's a machine.

You have to remember that random number generators that take into account many variables would be very difficult to create. It's more an experiment in programming than one in probability.

oooo... that's a really good point!

I've spent a little bit more time researching this subject, and I've learned an important point about just "guessing" doing better... the reason this is an important subject is because the computer isn't just "guessing," it's making a deliberate choice. If they improve this "silicon coin shark", building up its ability to correctly select by three or four percent every year, pretty soon we'll have a machine that can correctly predict a coin flip 100% of the time with just a "glance." People don't realize that scientific progress doesn't just happen overnight, you have to build to it. Anyway, it should go without saying that this technology would be applicable in other situations, such as giving satellites the ability to detect fires in realtime, preventing auto collisions, or even in directing postal traffic. I was very wrong when I made this post:

Posted By: legatissimo

It just goes to show that these so called "academics" are willing to pour millions of dollars into whatever silly little project they can think of.

This project isn't silly at all.

I think you're going to be hard pressed to find anyone around here who thinks that science is a joke.

All I'm saying is that a mashine that picks one out of two things correctly 50% of the time is NOT a scientific breakthrough. As soon as the machine is able to predict 51% however, then I will be impressed. Untill then, it's nothing more than a Magic 8 Ball to me.

Posted By: Udoboy

I know that Stanford's computer science department has been working on the same experiment to no avail, and they're supposed to be the best in the USA.

A colleague of mine worked on the team you're talking about and he told me all about the troubles they were having trying to get the program to predict correctly. He almost left the project when it was in its infancy - the success rate was closer to 41% then, and at one point it even dropped down to 37%. They're making progress, but MIT is clearly succeeding where Stanford is floundering.

seems like to me a computer programed to respond heads every time would be correct 50% of the time and it would only require one line of code. hmm...

People are getting hung up on the "nearly 50 percent" because they aren't getting the rest of the story.

From what I gathered, part of the "nearly 50 percent" is explained by a positioning laser sensor getting a false reading due to cosmic alpha particles. These particles bombard the earth's atmosphere at extremely high energy, resulting in "unknown" test results because of the muons (remnents of alpha collisions with the upper atmosphere) that interfere with the extreme precision sensors. This means that there are an unknown number of erroneous tests (neither true nor false). So legally, they cannot claim "5e+9 correct predictions out of 10e+9 tosses" but "nearly 5e+9 correct predictions out of nearly 10e+9 tosses".

They have a plan, that I figure requires relocating the equipment to a salt dome deeper underground, but such a plan hasn't been released yet. Again, I'm just guessing. And even if it gets released, we won't be able to get much out of it due to classified sections marked out by the dept of homeland security.

Reading through some old discussions.

If they can get this thing to work, and then make it small enough so that a guy (say, perhaps, a quarterback) could carry it , then I bet the odds are better than 50-50 that the NFL teams would pay BIG bucks to have one of these babies, because the winner of the coin toss at the beginning of a football game wins nearly 50% of the games.

Posted By: Udoboy

So far, we can only predict that it will bounce.

I'd be impressed except you neglected to include what degree of accuracy you have attained with your prediction.

Posted By: ImDero

There's a 50% chance of heads. There's a 50% chance of tails.

Probability states that ANYONE can correctly call heads or tails 50% of the time.

This is untrue. It depends upon the coin. I know that of American coins, the North Dakota Quarter will hand heads over 63% of the time, due to the weight of the tails side. I also know that a dime is the American coin most likely to land on it's edge, due to its smaller realative surface area on the heads and tails sides, combined with the fact that there are more ridges on the edge for the small circumference of the edge, making more likely to stand on edge once falling on it.