I consider it a tradeoff- it might not be strictly fair to all the cards, but on the other hand, you have more unpredictable (and to some of us, therefore more interesting) results, as well as a more concrete means of comparing cards. (It's easier to compare X to Y within the context of Z kingdom than it is in a completely general sense) I figure Qvist pretty much has the fair-as-possible group analysis market covered, anyway, but I can certainly understand where some people would prefer that idea to mine.
EDIT: Under the assumption that Sudgy speaks for most of the "needs changes" opinions, allow me to present both my current plan and a double elimination alternative that could mitigate the random factor a bit:
(Each of these has a play-in to go from 205 to 200)
Original plan
-20 Kingdoms, 5 each survive (100 remain)
-10 Kingdoms, 5 each survive (50)
-5 Kingdoms, 4 each survive (20)
-2 Kingdoms, 5 each survive (10)
-1 Championship
39 Matches
Double Elimination
-20 Kingdoms, 4 each win (80 perfect/120 1-loss)
-Winners: 8 Kingdoms, 5 each win
Losers: 12 Kingdoms, 5 each win (40/100)
-Winners: 4 Kingdoms, 5 each win
Losers: 10 Kingdoms, 4 each win (20/60)
-Winners: Bye round
Losers: 6 Kingdoms, 5 each win (20/30, combine brackets to 50)
-5 Kingdoms, 4 each survive (20)
-2 Kingdoms, 5 each survive (10)
-1 Championship
71 Matches
(I was going to work up a format with a wildcard/second chance component but the numbers kept working out weird)