Comments on Bill the Lizard: Getting a Fair Toss From a Biased Coin

Check out blog.eqnets.com/2009/08/24/dynamical-bia...

2009-09-17T23:34:36.711-04:00

Check out blog.eqnets.com/2009/08/24/dynamical-bias-in-the-dice-roll/

Pradeepg, That's a good question. I think the...

2009-09-11T08:56:10.595-04:00

Pradeepg,
That's a good question. I think the results would change in a different spot in the universe. The reason a coin toss isn't a perfect 50/50 proposition is because of a slight wobble in the spin of the coin introduced by imperfections in it's own balance. This wobble causes one side of the coin to be pointing up for a longer duration than the other side during the coin's spin. If you flip a coin on the moon, the wobble should change, meaning the difference in the durations should increase or decrease (I don't know which) due to the change in gravity. So if a coin has a 51/49 bias on Earth, it might have a 52/48 bias (or it could be higher or lower) on the moon, but I would expect it to be different.

I was just wondering whether performing this exper...

2009-09-11T01:43:49.488-04:00

I was just wondering whether performing this experiment on the Moon or in Universe will have any impact on results?

Troy, I know I have a 100-sided die around here so...

2009-09-08T14:49:33.797-04:00

Troy,
I know I have a 100-sided die around here somewhere. :)

If you have unfair dice, you can do something simi...

2009-09-08T14:11:57.072-04:00

If you have unfair dice, you can do something similar to remove the bias. However, it's going to take a lot longer than with a coin.

If you have an N-sided die (N is 6 for most dice), roll it N times. Repeat until you get a set of N rolls with each possible result present once. Use the first result rolled.

You could speed it up some by using a sliding window. If you roll 1,2,3,2, throw out the 1 and first 2, since they can't form a set of unique results. Continue rolling with 3,2 as a partial set. Since the rolls are independent, the initial 1 and 2 won't have an effect on the next four rolls.

The proof that it works is similar to the proof in the main article. Each roll is independent of the others, so all the orderings have the same probability. Picking the first roll (or any predetermined) roll gives a random result.

The bad news is it will take a while to get an acceptable set of rolls. With a fair coin, you need to flip twice on average. With a fair, six-sided die, you need to roll about 65 times on average. I recommend not trying this technique with a 20-sided die. :-)

Aruni RC, I never watched 21. I guess I ha...

2009-09-08T10:52:14.551-04:00

Aruni RC,
I never watched 21. I guess I have to rent it now.

Changing the initial conditions change the whole p...

2009-09-08T10:03:09.387-04:00

Changing the initial conditions change the whole probabilistic situation.
Just a thought - how many of us saw this in the movie "21"? :D

Anon, Beyond Colin's information about affecti...

2009-09-04T10:40:28.563-04:00

Anon,
Beyond Colin's information about affecting the outcome of a roll of the dice, I don't know if you can do it with fair dice.

Some cheap dice aren't fair, though. If you look at the face of some 6-sided dice, you'll see that the dots (or "pips") are small holes filled with paint. This means that more of the dice material is removed from the side with six pips than from the side with only one pip (and all the others). This causes a very slight imbalance.

Casino dice don't have this problem. The pips on casino dice are filled with material of the same density as the material the dice are made from so that each side weighs as exactly the same as all the others as possible.

There are methods you can use to throw off the balance of a die, just like most coins aren't perfectly balanced. An off-center weight can make one side more likely to come up than others. Shaving the edges of a face so that it rolls more freely away from that face is another way, but it is easily detected by visually inspecting the die. You can also shave an entire face down by a tiny amount so the die is no longer cubical.

"Can bias be introduced for rolling dice"...

2009-09-04T05:56:56.848-04:00

"Can bias be introduced for rolling dice"

I once watched a documentary about a new technique for manipulating dice in a Casino. Essentially the person rolling the dice at a craps table was spinning the dice (around the vertical axis) as oppose to rolling them as anyone else would. After lots of practice they could ensure with high probability what number would come up. They did this by throwing and spinning in such a way to make the dice stop dead when they hit the wall of the table. The number they had face up in their hand would be the number that was face up on the table. The documentary showed them doing this with quite high regularity though it didn't go in depth into the actual probability.

More alarming (or perhaps more revealing) is that the Casino refused to accept it as a form as cheating as they didn't believe it were possible.

I'm sure there's more information out there on the web.

Can bias be introduced for rolling dice?

2009-09-04T02:00:24.185-04:00

Can bias be introduced for rolling dice?

Jakob, Yes, that's a good point. You need to ...

2009-09-02T20:43:58.022-04:00

Jakob,
Yes, that's a good point. You need to make sure the initial conditions are the same for both tosses. Otherwise, the tosser could give himself a slight advantage by influencing the outcome towards either HT or TH.

Agreed. Obviously you need to pay attention to the...

2009-09-02T20:00:09.929-04:00

Agreed. Obviously you need to pay attention to the initial conditions before for each toss, to avoid bias introduced by:

1. If the coin is tossed and caught, it has about a 51% chance of landing on the same face it was launched. (If it starts out as heads, there's a 51% chance it will end as heads).