I do think selecting kingdoms in some manner is a good addition to the Championship match. While I certainly think there are improvements that should (and hopefully can) be made to the process, I believe the ultimate goal behind the selection process is to really test the skill of the players involved. I agree the using random boards would match the league structure better, but I would be disappointed if a BM/Cultist, or Rebuild board came up. Beyond those extreme examples, there is also a spectrum of games that don't produce a high quantity of difficult decisions. I feel that using random boards would marginally reduce the competitiveness of the match. I also think we need to figure out a way to increase the diversity of the boards selected to increase the competitiveness of the match. Right now, the heavy engine skew of the boards marginalizes the competitiveness as well.
It would also be nice to have more information on how kingdoms are selected.
I've been involved in this process since season 10.
We receive around 20-35 submissions from players during the season. I try to test all of them, but it requires a lot of dominion over the course of about 2-3 weeks, so we don't always get everyone through all the boards. I try to rate kingdoms primarily based on the difficulty of the play decisions that are involved to challenge the championship match players. drsteelhammer outlined the process pretty well, so I won't talk much about that.
There are a number weaknesses in our system.
The first is probably that the vast majority of boards submitted are "engine boards". I'm not sure what the best solution to this issue is, but we could have the test group generate some additional boards as a part of testing in attempt to fill in the strategic space. I think the "engine meta game" promoted around here lends to this pretty strongly. I have seen boards that I thought were very interesting get rated poorly by the other testers, perhaps because they were considered boring.
Another issue is personal bias. I won't name any names, but some of my fellow playtesters have a particular love for Procession and Black Market, and boards that contain them tend to end up with consistently high ratings. I certainly won't claim that I am not also part of this issue because I'm sure I have my own biases.
We also suffer from the problem of not being one of the few players who actually play at the level of the players we are picking these for. I'm not a bad player, but A board with competing strategies that isn't obvious to me, might be obvious to the top players in A. Then again, we have seen the players on rare occasion pass on the strategy that seemed dominant in testing. I think it would be great if some of the players who have competed at the A level had time to give at least some basic impressions on the boards after we have narrowed the selection down a bit.
I would support trying a different selection process out if it is preferred. My suggestion for a more random set would be to generate random boards prior to the match, and remove any that contain a unopposed monolithic strategy like Cultist/Rebuild/Embassy+BM. This would help maintain competitiveness, and allow commentators to familiarize themselves with the boards.
On the topic of commentary, I don't think having board the commentators haven't seen would improve things. I think being able to answer questions about what works or doesn't work and why is better than just guessing at things. I suppose not having knowledge to speak from would force commentary to stick more to what is happening with the players, which might be preferred by some viewers. I know from my experiences doing it that it is hard to follow the game closely and discuss general strategy at the same time.