On November 22 2004 01:00 gg2w wrote:
Sorry to keep harping on this question as it seems most people are past caring. Seems like all my old SA magazines have been thrown out, schucks.
The flaw in the reasoning is assuming that the choosing of the first or second girl occurs with the same probability in the GG case as it would in a BG or GB case. Certainly we can all agree that GG, BG and GB all occur with equal probabilities. It is very important to work from this perspective and you will see why in a moment.
Let us compare GG and BG. In the case of GB, the second sibling (the only girl), is selected for %100 of the time. However in the GG case, G1 is selected for %50 of the time and G2 %50 of the time (if you wish, only require the percentages to add up to %100).
"Since you know a girl opened it, it could be G1 from GG or G2." True, but the selecting of G1 from GG only happens half the time I have a GG, while selecting G1 from GB happens %100 of the time I have a GB. This is what I meant before when saying that the statement "one must be a girl" acts differently on GB/BG as opposed to GG. The argument that I can differentiate G1 and G2 in GG isn't the point, it's that those occur in a split percentage since being a girl doesn't single out the first sibling over the other.
"If she is girl 1 then boy 2 or girl 2 are possible. If she is girl 2 then boy 1 or girl 1 are possible." This is very convincing, even had me fooled for a while. Problem is that you are looking at it from the wrong perspective, you are trying to decide the probabilities from the point of view of having originally selected a girl (1 or 2). Look at it this way, let's assume we have a boy 2 case, then we MUST have a girl 1 case. So this has a "relatively" high possibility. Now let's assume we have a girl 1 case at door, with the additionally possible girl 2 case (i.e. G1 at door, G2 unknown). Why should this have the same probability as the boy 2, girl 1 case? All I stipulated was that a girl appear at the door. Couldn't it just as well have been girl 2 who shows up at the door? So while G1/G2 is possible, it certainly shouldn't happen as frequently relative to GG as G1/B2 case happens relative to GB (%100). So what you end up doing by making all the G1/G2, G2/G1, G1/B2, and G2/B1 cases all equally likely is that you end up counting GG twice.
Here's a vague probability tree to show you what's going on.
1. GB = 1/3
2. BG = 1/3
3. GG = 1/3
Given 1. G1 shows at door %100 of the time (since it must be a girl).
Given 2. G2 shows at door %100 of the time (since it must be a girl).
Given 3. G1 shows at door %x of the time and G2 shows at door %(100-x) of the time (since all I require is a girl hence it can be either with the probability of both of them (not necessarily at the same time) adding up to %100).
Sorry to keep harping on this question as it seems most people are past caring. Seems like all my old SA magazines have been thrown out, schucks.
The flaw in the reasoning is assuming that the choosing of the first or second girl occurs with the same probability in the GG case as it would in a BG or GB case. Certainly we can all agree that GG, BG and GB all occur with equal probabilities. It is very important to work from this perspective and you will see why in a moment.
Let us compare GG and BG. In the case of GB, the second sibling (the only girl), is selected for %100 of the time. However in the GG case, G1 is selected for %50 of the time and G2 %50 of the time (if you wish, only require the percentages to add up to %100).
"Since you know a girl opened it, it could be G1 from GG or G2." True, but the selecting of G1 from GG only happens half the time I have a GG, while selecting G1 from GB happens %100 of the time I have a GB. This is what I meant before when saying that the statement "one must be a girl" acts differently on GB/BG as opposed to GG. The argument that I can differentiate G1 and G2 in GG isn't the point, it's that those occur in a split percentage since being a girl doesn't single out the first sibling over the other.
"If she is girl 1 then boy 2 or girl 2 are possible. If she is girl 2 then boy 1 or girl 1 are possible." This is very convincing, even had me fooled for a while. Problem is that you are looking at it from the wrong perspective, you are trying to decide the probabilities from the point of view of having originally selected a girl (1 or 2). Look at it this way, let's assume we have a boy 2 case, then we MUST have a girl 1 case. So this has a "relatively" high possibility. Now let's assume we have a girl 1 case at door, with the additionally possible girl 2 case (i.e. G1 at door, G2 unknown). Why should this have the same probability as the boy 2, girl 1 case? All I stipulated was that a girl appear at the door. Couldn't it just as well have been girl 2 who shows up at the door? So while G1/G2 is possible, it certainly shouldn't happen as frequently relative to GG as G1/B2 case happens relative to GB (%100). So what you end up doing by making all the G1/G2, G2/G1, G1/B2, and G2/B1 cases all equally likely is that you end up counting GG twice.
Here's a vague probability tree to show you what's going on.
1. GB = 1/3
2. BG = 1/3
3. GG = 1/3
Given 1. G1 shows at door %100 of the time (since it must be a girl).
Given 2. G2 shows at door %100 of the time (since it must be a girl).
Given 3. G1 shows at door %x of the time and G2 shows at door %(100-x) of the time (since all I require is a girl hence it can be either with the probability of both of them (not necessarily at the same time) adding up to %100).
Let me put this simply.
Odds that Girl 1 shows up at the door: 50%
Odds that Girl 2 shows up at the door: 50%
Assuming the girl in question is Girl 1, the child 2 can be Boy 50% or Girl 50%
Assuming the girl in question is Girl 2, the child 1 can be Boy 50% or Girl 50%
Therefore it all adds up to 50/50.
If you are NOT originally selecting, then the inversion of GB and BG counts as one unit. You HAVE to select, because if we take
GB
GG
BG
BB
as the only possibilities, then the girl you see must eliminate either GB or BG. You can say that it doesnt eliminate either because as long as one is a girl, both are safe. However, if you take the liberty of saying that the girl you see can represent either the girl in GB or the girl in BG, GB can be inverted to = BG. Therefore making one case.
If you have a Boy2 Case (50% probability) there is a Girl 1 case (100% probability)
Therefore its still 50%

