For this question you will have to make a few assumptions: are the probability of people meeting up (and then getting married/having a kid) the same for all people?
I think that this problem is a lot more complicated than you think because the way that people meet and marry is not as random as the above suggestion would have.
People tend to mingle in their own social circles which makes this problem very difficult since you now have to classify people according to their "circle" and do so in a specific way.
All these sort of things need to be factored into the model to make it even half useful as a prediction tool.
How badly do you need this answered?