In the spirit of contributing to the economic drain from the NCAA Tournament that Skip noted, I wanted to consider its structure. This year has been the exception for the lack of complaining about who got in and who did not. In my view, that's a good thing because it detracts from the real problem in the Tournament's structure -- the seeding.
Problem 1: there is a huge difference in playing teams seeded 15 and 16 versus those seeded 12 and 13. Teams seeded 1 (and to a lesser extent 2-seeds) get to play one practice game and three real games to reach the final four versus other teams that must play four genuine opponents. That "gimme" game is a huge advantage over the field.
Problem 2: the difference in teams seeded 5-12 is relatively small. Comparisons between the fourth best SEC team and the third best Big 10 team or between a second-tier major conference teams and top a "mid-major" team are are often in the eye of the beholder. They are not very consistent from year to year, and the psuedo-statistics like the RPI only give a feel of science to a huge lack of information.
Problem 3: good teams from bad conferences who always get the 15 or 16 seeds really never have a realistic chance to win a game. Holy Cross takes Kansas and Kentucky to the wire but really gain nothing for it.
Below, I propose a small modification to the tournament that would address these problems while continuing to give preference to the best teams from the best conferences -- a prerequisite for any realistic structure since the tournament is really by and for the major conferences.
Step 1: Assign seeds only for teams in the 1-4 slots. This will be (almost) exclusively the best teams from the major conferences. That way, they do not play each other until deeper into the tournament.
Step 2: For the 48 remaining slots, hold an initial round of 96 teams, where matchups are decided by random draw. This attacks the problem of making imprecise distinctions by treating the rest of the field as essentially equivalent. It also gives good teams from historically bad conferences a realistic chance to win a game.
Step 3: Take the 48 winners from Step 2 and assign them to the remaining slots. I would prefer a random draw. An alternative that gives more preference to the 1 and 2 seeds would be to place the 8 lowest rated teams against the 1 and 2 seeds and then randomize everyone else. (Geographic issues could be worked into the draw also.) Either way, the 1-4 seeds do not have to play each other early, but the 1 seeds do have to play genuinely competitive opponents rather than scrimmage games against overwhelmed opponents.
Even with these flaws, interest in the Tournament is huge (as Skip notes). The lessons from this for other sports leagues is a topic I'll address down the road.