I don't think its a good idea to vary far from the default magitech HP style because most mods are balanced with magitech and people will want to play other mods with megapack.
There's going to be balance issues no matter what. Even if the HP of most units aren't changed, the much higher than average HP of the buildings is a big difference. Not to mention that units are going to be stronger, since we're aiming for 3-4 hits to kill themselves, right? So the way I see it, balance with non-megapack factions is going to be broken no matter what, so may as well choose a standard we think will best fit the game.
While we're discussing HP, though, we need to take a look at armour. As a reminder, the damage formula is
(attack strength - armour value) * damage multiplier. Thus, armour can make a huge difference. However, if the average attack is something like 300 damage, then 20 armour is almost nothing (7%). Armour would have to be more in the 50 (17%) to 150 range (50%) to make an impact on most units. However, I'd personally rather see a focus towards the attack type rather than the armour value. The armour value is more for making some types of metal armour better than others, so I think that armour should be based at zero. So a swordman (PS: can we please rename it to "swordsman"?) would have zero armour, as it's pretty much the most basic leather armour unit. On the other hand, the horseman might get something like 50 armour. Long story short, we shouldn't be using armour values very much, instead focusing on the attack and armour types. If you've seen the Google Doc, you'll notice some multipliers as low as 0.2, so attack and armour types really matter, especially with specialty units.
However, I'd like to point out we can't really do that much planning about HP, armour, and attack strength in discussion alone. We need to make a testing faction and give the values a try. I again think we should base the HP around 1000, and have all units' HP and attack strengths within multiples of 50 or maybe 25. This makes the numbers simpler. Easier to remember that a guard has 1500 HP than 675 and easier to remember that his attack strength is 400 instead of 225. But honestly, let's actually give the values a try before we knock 'em.
Although I'm not sure about the fact that a unit should be able to kill itself in 2-3 moves. First of all, there's the attack speed issue. Some units might have really fast attacks, which should be weaker. Other units are powerhouses with lots of HP, like the airship. If the airship has 2000 HP, to kill itself in 2-3 moves would have an attack strength that's one hitting most units, which is too much in my opinion. The attack strength should be relative to the average unit, not the unit using the attack. However, even if we're killing the average unit in 2-3 attacks, that still seems a bit fast too me. Units with splash attacks would dominate, since they could kill a unit before the unit can reach the attacker. I think the average unit against the average unit should be about 3-4 attacks to kill and weaker units should take longer while stronger units might take less hits. This 3-4 hits should also take into aspect the armour multipliers. Some units would have natural advantages over others while some are at severe disadvantages, which really changes the number of hits needed to kill the unit.
Finally, buildings. Please remember that buildings have their own armour types, against which the majority of attacks are rather ineffective against. Thus, we don't need
too high HP values for the buildings. After all, if it's too high, even siege units won't be able to do much damage as a percentage. Percentage damage is what's important, not absolute.
So, if you haven't seen it yet,
here's the attack and armour types doc. Also, anyone want to volunteer to create a testing faction for testing appropriate values for HP, armour, and attack strength? Bear in mind we need to test attacks with splash damage, buildings, and battles with large numbers of participants.
