Now, pay close attention.
By reducing the number of participants in each tier, we equalize the strength of the tiers.
I can name 16 players who are approximately equal in strength for the captain slot. Now look. We have 16 captains of equal strength (the difference in skill is minimal).
Now, we repeat your scheme, and we will see that captain 1 will pick first 2 times and last 2 times, and captain 16 will pick first 2 times and last 2 times, and at the same time, the composition of participants in the tiers will be more equal due to the reduced variance.

The conclusion that the captains are equal is nonsense. Accordingly, everything else is incorrect.

Silens wrote a lot there, but I'm going to Poland; I'll be happy to look into his arguments when I return.

The experience of past tournaments has clearly shown that 4 players per team is much more balanced than 3.
And the probability that one of the 5 participants will drop out (not participate in the tournament) is definitely true.

Sanin has a desire to experiment – good luck. It would have been better to conduct a poll before registration.

I spent 5 minutes discussing it with Silens. Yes, mathematically, he correctly proved that an even number is correct. And he confirmed LuckyF's argument that the smaller the tiers, the better, i.e., in an ideal scenario, the most balanced compositions are achieved with two teams! All conclusions are correct! I don't see a problem with draws. In the playoffs, there are at least 2 games, and I think Silens can prove that they can end 3-2 and 2-3 (or 4-1 and 1-4) even with an odd number of participants, resulting in the same overall draw, and something else will need to be done.