Research and Cafes

Those who drink decafs can only churn out lemmas, not theorems. To be a true churning machine of theorems, you have to consume coffees.

Nearly every research result I have had in Italy has come from being in the proximity of Cafe Salza in Pisa, an institution from late 1800's. I seem to take a table there for an afternoon and order every pastry on display. From my first visit there, Paolo Ferragina and I published multimethod dispatching algorithms in STOC 99; later, the 2d string indexing in PODS 01; from a visit two years ago, we got the succinct tree encoding result in FOCS 05; and finally, during my visit there last summer, Gianni Franceschini and I solved the suffix selection problem to be presented in STOC 07. I hear that the famous Salza has changed hands and the new owners are french. Sigh.