I'd almost prefer the 6 seed to the 5 seed. The 5/12 seed matchup seems like it ends up in an upset a lot of the time.
Just out of curiosity, I looked up the # of wins per decade for the 11, 12, and 13 seeds. The "there's a 12-5 upset every year" doctrine is something I grew up with and has been remarkably consistent — only 2000, 2007, 2015, and 2018 haven't had one. Years without 11-6 upsets, though, are also pretty rare — only 1992, 1997, 1999, 2004. So actually, in terms of "always pulling off an upset", one could argue that 11 is the new 12. Case in point:
Years | 13-seed wins | 12-seed wins | 11-seed wins |
1990 – 1999 | 7 | 14 | 9 |
2000 – 2009 | 9 | 16 | 13 |
2010 – 2019 | 8 | 15 | 21 |
20 years ago, the last three or four at-large bids were almost always 12 seeds. These days most of the 12s are auto-bids. Not sure if that's due to more conferences than before, or just fewer multi-bid conferences, but whatever the reason, the average 12 seed in 2020 is not as good as in 2010 or 2000 (although I am too lazy to pull the KenPom ratings to prove this).
To make matters worse, 4 samples per year is very few, so the odds of "12 seeds are great at upsets" seeming reasonable just by chance, even over a decade of tournaments, is pretty high. In other words, the 14 to 9 discrepancy in the 90s is not that unlikely even if 12 and 11 seeds were exactly equal in those years.
At any rate, I would expect that over the next 20 years, the most likely outcome is (# of 11 seed wins) > (# of 12 seed wins) with much higher probability than when 11s and 12s were all bubble teams.
Data source (someone should probably check my counting)