okay, lets go.
It seems normal to me that upgrading is less efficient in gold per hammer than buying directly.
indeed, there is too much advantages of upgrading compared with direct buying.
-keep a lot of low upkeep units... keep the gold, upgrade only when needed
-need only 1city able to do the high-tech unit to upgrade all units.
-keep the xp gained
-can upgrade multiples units per turn with one city instead of only 1 building per turn.
-can reduce the time the unit occupies the city building list
-build high xp (lvl3) low tech, low upkeep units in far away developped city, have them move to front, upgrade to high tech units when front is joined, gaining gold on the upkeep difference during the moving time.
upgrading is already a very rewarding system. increasing the efficiency of gold/hammer for upgrading increases the rewards for upgrading.
first I wanted to note that it seems you are focusing on gold cost for upgrade instead of "gold efficiency" for upgrade.
Indeed, it seems you are forgetting that between VEM and GK (ergo then GEM), unit cost increased ... so it seems normal that upgrade costs increase.
your reasonning is akin to the following one (which I know you are not making) :
-in VEM, longsword is 170 hammers and it costs 170*2 (to simplify the calculus) 340 gold to buy it with gold.
-in GEM, longswords are 230 h, costing 460gold to buy...
"this difference of 120gold is HUGE .... please reduce it because it becomes harded to rush build units when attacked...bla bla."
however, the 120gold difference is normal ; hammers cost increased so gold cost too.
strangely, you don't react like that;
why ?
is it that when buying you are not comparing GEM vs VEM as you are comparing gold vs hammers : ie : efficiency of gold ?
but then, when upgrading, you are comparing cost between GEM and VEM ? why that ?
why ?
further, upgrading is less costly then building anew.
you need only 250g (80*3 +10) to upgrade a highly promoted sword to a high promoted longsword
instead of spending 460g (230*2) to buy a low xp longsword from scratch.
this is almost a factor of 2.
clearly upgrading is more interesting.
further, using :
Swords were 110 in VEM and are 150 in GEM. Longswords are 170 in VEM and 230 in GEM.
VEM would be 60*3+10=190. GEM would (now) be 80*2.5+10=210. (Obviously it was 250, which I think you can imagine is a noteworthy change per unit).
Using the GK upgrade formula would put them at 170. Which is just as close to VEM but on the other side of the fence.
the difference in hammers between the sword-longsword upgrade is 20hammers (80 instead of 60)
When advocating the 2.5 (or smaller effect),
you say you want a reduced upgrade cost but still with same effect than in VEM.
in reality what happens instead that you will obtain this :
same units, costs increased for building each, but the cost for upgrade doesn't move.) which seems highly unlogical and increases the efficiency of gold for upgrades in GEM compared to VEM.
with 2.5 :
Do you really want that an upgrade costing +20 hammers compared to VEM to cost only 20 more gold than in VEM. (210GEM vs 190VEM)
well, it seems a very high profitability of gold for upgrading units from VEM to GEM
(1 more hammer bought with 1 more gold)
thus, you want GEM to favor upgrading compared to VEM ?
If that is the case :
When focusing on apparent costs you are forgetting that you might go into a direction where it becomes more interesting to buy low-tech units in many cities and upgrade them into high tech units instead buying directly high tech units.
and if it doesn't become "more interesting" on a per-unit transaction, at least it might become more interesting as a global empire strategy as you will have less specialization needed, less military building needed...etc
And it is not realist : swordmen trying to convert to rifles need to forget all their hard-won reflex and rewrite new ones. it is harder than training new units from scratch.