Imagine the following situation:
You get on a plane in Chicago and fly to London. A few days later, in London, you go to Saturday morning services and the young woman leading the service definitely doesn't sound British. In fact, she sounds American. So after the service you talk to her a bit and discover that you both grew up in the same suburb of Washington, DC. In fact, you went to the same high school. And even at pretty much the same time (three years apart, so you only overlapped by a year). Oh, and your parents' houses were about 150 meters apart.
It's a pretty odd feeling. Trust me.
Posted by bzbarsky at December 31, 2005 1:44 PMPretty cool. In fact, that's way out. What are the chances? Do we have a statistician here?
Posted by: David Naylor on January 1, 2006 6:01 AMBoris
Welcome to London - Where many things can happen
Happy New Year!
To David Naylor:
As someone who has studied statistics, I can tell you that that question often leads to one of the major statistical fallacies. If you see some astoundingly coincidental series of events, and ask "What are the chances of *that* happening?", the answer is generally "incredibly small", so you think how unlikely it was to happen. But before the even happened, you wouldn't have asked "what are the chances of [specific astoundingly coincidental series of events that hasn't happened yet] happening?". If you were to ask in advance "what are the chances of *some* sequence of events occuring that I would find astoundingly coincidental?", the answer is "not particularly uncommon".
A standard example used is that of running into someone you know in a place you don't normally see them; for example, running into one of your professors at the store, or into one of your co-workers walking down the street in another town over. You think, "if I had been 30 seconds behind or ahead, I wouldn't have run into John", and you think the odds are relatively unlikely. However, the relevant odds are not those of running into that particular person at that exact moment in that exact place, but those of running into *someone* you know at *some* point that day and *some* location along your travels for the day; the latter odds are much better.
Applying that to this case, if you were to ask (in advance or not) what the odds were of travelling to London, attending Saturday morning services, and hearing them being given by someone who grew up in the same town, went to the same school at around the same time, and lived within 150 meters of you, the answer would probably be "mind-bogglingly low". However, if you were to ask what the odds were of, on some foreign trip, encountering someone from the same town, who shared some small subset of your experiences (same school, same workplace, same neighborhood, same organization(s), one worked at a place the other frequented, etc), the odds would be low, but not unlikely.
And if you ask "what are the odds of reading the weblog of someone who had such an experience", and consider how many weblogs you read regularly, the odds rapidly approach 1. :)
This is why people hate statisticians sometimes. :)
In any case, it is still awesome to run into someone like that.
This is more or less the same as anon but I like the explanation.
Miracle on Probability Street
"The Law of Large Numbers guarantees that one-in-a-million miracles happen 295 times a day in America"
http://www.sciam.com/article.cfm?chanID=sa006&articleID=00094511-E068-10FA-89FB83414B7F0000&colID=13