After getting a bit frustrated with Tim Harford last week for his inaccurate discussion about altruism, yesterday he offered an insightful post about probabilities and clustering.  So, how likely do you think it is that you will get 6 two yolk eggs in your half-dozen egg box? Well, Harford was told that getting one two yolk egg occurs with a probability of 1 in 1000.  So, if you get six of them (and if, BIG IF, they are independent events) you simply multiply 1 in 1000 by itself 6 times to determine the likelihood of getting 6 two yolk eggs in one box. From that multiplication you get a really small number: 1 in one trillion.  But, what if they aren't independently determined? Here's where the sociology of egg-packing comes in, supposedly egg-packers can tell the difference between one yolk and two yolk eggs, and they set aside the two yolk eggs and pack them together later.  Fantastic! What's the moral? Don't count your independent events before they've hatched? Or something like that...