Circa 1994: Josh Dever would be sitting on a couch in the philosophy graduate student lounge at U.C. Berkeley. I would propose to him a definition of "dessert" (e.g.: "a sweet food eaten after the main meal is complete"). He would shoot it down (e.g.: "but then it would be a priori that you couldn't eat dessert first!"). Later he would propose a definition to me, which I would shoot down. Over time, the definitions became ever more baroque. Other graduate students participated too.
Eventually Josh decided that he would define a dessert as anything served on a dessert plate. Asked what is a dessert plate is, he would say it was intuitively obvious. Presented with an objection ("So you couldn't eat Oreos right out of the bag for dessert?") he would simply state that he was willing to "bite the bullet" and accept a certain amount of revision of our pre-theoretical opinions.
At the time it seemed like cheating. In retrospect, I think Josh saw right to the core.