Castle is brilliant largely due to the diversity of heroes. My personal starting preferences are as follows:

Jebus Cross - Cassiopeia/Corkes
Diamond - Jeremy/Corkes
8mm6a, 6lm10a - I take Cassiopeia myself, but I fully allow Miriam. Yes, here it's not even about the lategame, but rather the pivotal moments like 126-132 (when every tile of vision during contact with the opponent is worth its weight in gold) she is very strong
8xm8a (and any poor template) - Lina

p.s. On mm6, if the opponent takes Cassiopeia, I also take Lina to balance the tempo with the economy.

p.p.s. Also an excellent hero is Illor. I don't take him as a starter (although sometimes it's a good option), but I gladly switch to him if I start with Lina.

Yep, that's about what I was hoping to get.

There are 3 starters - Cassie, Jeremy, Anabel, and there are 3 swap-in heroes - Corkes, Miriam, Illor. In certain situations, you can start directly with a "swap-in" hero (for example, Corkes on Jebus, not sure about Miriam, I'm skeptical), but there's a risk of not making it out of the early game. I hadn't considered Lina, since I've never played 8xm8, maybe there is some merit to her.

Added 43 minutes later

I repeat, if this is the party policy

The party policy is - we want to balance the power of the towns. Not necessarily to perfect balance, but so that you can normally play any town against any other, maybe with a slight offset (<2000).

The second point of the party policy is - fewer lines in the changelog. To avoid what happened in WT with 100500 minor changes that barely affect anything. This is especially true for standard SoD functionality. If a change isn't necessary, it's better not to make it. For example, the prices for monks-birds-bird dwellings in SoD are clearly too low, so they were raised without much hesitation, since a rollback is unlikely.

The third point of the party policy is - less flip-flopping, introducing and reverting various features, raising and lowering numbers. The strategy here is - first, we introduce all the conceptual, balance-forming features at once, then we test them in tournaments, and based on the results, we slightly adjust numbers up or down (prices, starting quantities). Again, the requirements for standard functionality are much stricter.

Therefore, I'm not claiming that Necro-Conflux will remain unchanged, but we need to determine exactly - which numbers to adjust, where to adjust them, and by how much. For example, the starting quantity of air elementals is also not a constant, but the correct numbers are not obvious here, so we'll wait.

So you're not speaking into a void; all opinions are collected and considered.

Although I think the phrase "only flies appear" doesn't reflect reality; it would be more accurate to say guaranteed flies appear! Is it even worth mentioning that absolutely all neutrals have an incredible attraction to flies, and with a certain amount of experience with the Fortress army, the flies pull out battles that are hard to imagine for any other tier-1 creature.

Well, agree, towns that start with specialists for levels 1-3 benefit much more from having 3 stacks, because it increases the main strike stack, which is still more important than the chaff or flankers.

Question - surely Rampart and Conflux, with guaranteed three stacks and level 2 (especially critical for Conflux, more than any other),

I suggested cutting Rampart's starting army more heavily, but there were many voices against it, so it ended up as it is now. I agree that Rampart gained more from the 3 stacks than Dungeon.

Conflux was released with the comment "needs testing in practice", because predicting the results of changing the game model in your head is difficult. Therefore, everything that didn't require obvious intervention remained as in SoD. Including the starting stacks.

The same applies to Necropolis.

do you really think my opinion is caused by some inner fire?

Through text, emotions are quite visible. This is a case where after stating an opinion, you need to wait, cool down, and then think about the situation again. Categorical opinions are generally rarely correct; the truth is usually somewhere in between.

Try to look at the situation through the eyes of the developers, taking into account the 3 points of the party policy, then all the changes will become quite logical.