I prefer winner breaks if the race is long and alternating breaks if the race is short. For me, exactly where the dividing line is between "long" and "short" probably depends on the game, the equipment, and the rules.
What I really want, whether I am playing or watching, is a match that gives both players ample opportunities to compete, to interact, to use their various pool skills. The winner-breaks format can completely eliminate competition from what is supposed to be a competitive event and turn it into a one-person exhibition. Actually, the alternating-breaks format can also fail to produce interaction, but at least it means both players will get to the table. For a match with lots of interaction among the players we either need lousy players, something I don't want to watch, or rules and equipment that force the interaction.
In most pro events, however, lengthy break-and-run packages are fairly rare. So for middle-length races (say, races to 8, 9, 10, 11), I'm fairly indifferent between winner breaks and alternating breaks.
Yes. And good fortune certainly plays a role. A player can look invincible in one match, breaking successfully, running out, etc., and then can't break successfully at all in the next match on the same table; and the stats get turned upside down.
But note that I generally post more than just a couple of break stats (such as successful-break percentages, breaker-won-game percentages, and B&R percentages). In the "Miscellany from the data ..." post in many threads, I report on things like the percentage of games that ended in one inning, the percentage of games that were run out by the player at the table following the break, the percentage of games won by the player who made the first ball after the break, and the frequency of fouls, misses, and safeties. Those help to flesh out the level of performance as well as the degree of difficulty created by the rules and equipment.