You need to compare it with others whose gameplay has either not really changed or has changed predictably. I didn't ask for nothing - has Dungeon become weaker than Castle-Swamp(-Stronghold-Tower)? IMHO it hasn't; on the same map it still stands higher, which is quite enough.
Regarding "why you can't test" - the point is "don't be afraid". For testing purposes, 20-30 are perfectly adequate values, and besides, they are clean numbers (unlike 23-33).
In my opinion, if Dungeon's starting army is to be strengthened, it should be by adding eyes-harpies, in order to lift the other starters (Lora, Gunnar).
Added after 7 minutes