Pascal's Wager and Diversification of Risk.Marginal Revolution reports on a Justin Wolfers' variant on Pascal's Wager: If you have five children, train each to be a disciple of a different religion. That way, at least someone in the family will get into Heaven.

This is an argument worth thinking about. It is not an argument about yourself, but about helping your children. Here, perhaps, is the fallacy. If I am selfish, I care about my own salvation, which this project will surely doom. If I care about my children, I should choose the best estimated outcome for each of them-- which is the religion which best meets the conditions of Pascal's Wager (a high up-side gain and a low down-side cost). The Wolfers project makes sense only in a mixed case where my utility is concave in my children's utility. That's not unrealistic, but it's what we think of as our duty towards the children either-- and remember, if we're thinking about realism, my own salvation is going to weigh very heavily.

Of course, Pascal said that he didn't think his Wager was sound, since only God can give saving faith-- cold-blooded calculation won't make you love God. But the Wager does work as a way to avoid punishment for sin, if not for adoption into God's family.