PaulErdos has (or rather had) an ErdosNumber of zero. Everyone else has an ErdosNumber equal to one plus the lowest ErdosNumber of anyone they have written a mathematical paper with. Consequently, Erdos' co-authors have a number of 1, their co-authors have a number of 2, etc.
- google:erdos-number
ErdosNumber__s are amusing because PaulErdos published papers with so many people - he was the most prolific mathematician of our time - and because mathematicians collaborate so much, hence _every_ mathematician has an ErdosNumber. The idea wouldn't work for other fields where people don't connect as much (such as history, perhaps). It does work for similarly promiscuous fields - for example, the BaconNumber in cinema acting.