This link is now dead; the Wayback Machine says the text was this:

Very little of what goes on among human beings, very little of what goes on in so limited an activity as a game, is merely conventional (done solely for convenience). In baseball, it is merely conventional for the home team to take the field first or for an umpire to stand behind the catcher rather than behind the pitcher (which might be safer). In the former instance it is convenient to have such a matter routinely settled one way or the other; in the latter instance it must have been more convenient for the task at hand, e.g., it permits greater accuracy in calling pitches, and positions an official so that he is on top of the plays at home plate and faces him so that his line of sight crosses those of the other umpires. More or less analogous advantages will recommend, say, the Gerber convention in bridge. But it can seem that really all of the rules of a game, each act it consists of, is conventional. There is no necessity in permitting three strikes instead of two or four; in dealing thirteen cards rather than twelve or fifteen. -- What would one have in mind here? That two or four are just as good? Meaning what? That it would not alter the essence of the game to have it so? But from what position is this supposed to be claimed? By someone who does or does not know what "the essence of the game" is? -- e.g., that it contains passages which are duels between pitcher and batter, that "getting a hit," "drawing a walk," and "striking a batter out" must have certain ranges of difficulty. It is such matters that the "convention" of permitting three strikes is in service of. So a justification for saying that a different practice is "just as good" or "better" is that it is found just as good or better (by those who know and care about the activity). But is the whole game in service of anything? I think one may say : It is in service of the human capacity, or necessity, for play; because what can be played, and what play can be watched with that avidity, while not determinable a priori, is contingent upon the given capacities for human play, and for avidity. (It should not be surprising that what is necessary is contingent upon something. Necessaries are means.) It is perhaps not derivable from the measurements of a baseball diamond and of the average velocities of batted basseballs and of the average times human beings can run various short distances, that 90 feet is the best distance for setting up an essential recurrent crisis in the structure of a baseball game, e.g., at which the run and the throw to first take long enough to be followed lucidly, and are often completed within a familiar split second of one another; but seeing what happens at just these distances will sometimes strike one as a discovery of the a priori. But also of the utterly contingent. There is no necessity that human capacities should train to just these proportions; but just these proportions reveal the limits of those capacities. Without those limits, we would not have knowsn the possibilities.

Stanley Cavell, The Claim of Reason. pp 119-120