I was considering recently the implication of having an absolute cost (say +10 cost to an unit so it can have more attack rating) or a relative benefits (say an outpost gives +25% defence rating).
Is it really what the designers wanted? To favor each time the strongest units, in a kind of snowballing effect aka Darwinesque approach 'the strong becomes stronger, the weak will be weaker' etc. ?
Because think about that. For lousy unit you build, like some club equipped militia, even a +5 cost is significant. But for a knight wearing a plate, a +10 cost is not that significant, when you factor the cost of his equipment. Bottom line, the weak units will never have perks generally, the ones which are expensive will have 3... Min-max if you want.
Let's take the other point, giving a bonus as a %. Typically outpost bonus. +25% defence means nothing if you have low quality units, because you'll get between 0 and 2 bonus point in defence. For a strong unit, that's a very big bonus on the contrary. So again here, a gameplay consisting of a few super units is again privileged...
Is it what the designers wanted in the first hand? I'm not too sure, because we all know about a game 'lock' called 'stack of doom', basically, a stack strong enough to crush anything in its path without loss.
Or alternatively, we know that players have a tendency to prefer high quality over quantity, because this is a way to win against AI with production advantages, as you want to avoid a war of attrition against such AI. You want to lose at most 1 unit when the AI loses 10... I think you get what I mean. The natural tendency is to try having the units that will allow 0 loss most of the time.
And so, if the game mechanics themselves lend to this style of gameplay, then chances are that the 'stack of doom' syndrom appears quite easily.
So... conclusion, should we not have costs with % (like 'strong: +25% cost') and bonus with absolute values (like: outpost defence: +3 defence to all units). And not the reverse?