There's probably more than one variety of tea in the replicator's databanks, and of course there's more than one way to drink tea, so Picard specifies with the order. "Tea." He's specifying the type of beverage he wants. "Earl Gray." He's specifying the variety of the beverage he wants. "Hot." The only really vague command, but specific enough to tell he doesn't want it cold or room temperature. If you're talking about ordering Earl Gray at a Starbuck's you'd have point, but a replicator is not a Cafe and it's brain is not that of barista, so it wouldn't have the common sense to know that "of course you serve Earl Gray hot," It has machine intelligence, and unless part of its function is to read people's minds, no part of Picard's order is unnecessary if he wants to get the tea he wants. Should he have presets? Presets require identification, so whenever he wanted tea he would have to say something like "Picard, beverage choice one-A" or "This is Captain Picard. Replicate tea according to standard parameters" or the replicator would have to have biometric technology in order to determine it's Picard ordering the tea and some schmuck from the lower decks. How is any of that more efficient than just speaking four one-syllable words into the machine to get the same result? I hope you're joking about the boycott/apology thing, because it would be the height of idiocy if you were serious.