Outposts can be upgraded. The question is, are people actually using this and if so, does it make a difference to the outcome of the game?
I build outposts with towers because they let me connect resources. I consider this at the time I place the outpost itself.
I don't build the one that gives food, however I changed my copy to give +2 instead of +1, which makes them better. I also tried making them +2 gildar however it never paid off, and growth was better.
The 2 buildings that give 25% attack/defense need changed, if they gave 33% they would be great (not for me, for the AI when defending itself against each other, and to help vs monsters as a reward for having the outpost there). Otherwise, they should be combined into one building that gives 25% attack and defense inside borders.
The Stable is wonderful, it allows me mobility on large maps, especially Anthys.
Reducing tactical spells is useful during defense and on offense you can sack an enemy monolith or outpost and rush buy this improvement. You can also buy an Arcane Monolith or settle a pioneer on the go, and rush buy it to have reduced tactical spellcasting everywhere except enemy borders.
I would like an improvement that gives 2 militia if attacked (like a city). Enabling this auto-protect without garrison would make players + AI have another defense from each other and monsters. This isn't overpowered because the units couldn't leave the tile to defend a nearby tile in it's zone, it can only protect itself.
I never use Caltrops because it comes too late. If it came as one of the first 3 or so improvements, I would definitely use it.
Warden conflicts with any ability which is good for the tower (except stale) 80% of the time. However, it does give a choice because you probably don't plan on building the others if you do build a warden.
Something (not) funny is happening. Comparing my development vs AI development (Huge Map), the AI either spams early cities and wakes up all the monsters or fails to expand at all and stays at 20 points or less all game. There are large barren swaths of land all over the place, and i spend so much energy defending myself (Porcipinee, or any non-fire user) that I don't have time for pioneers. Instead, I send out my troops plopping Arcane Monoliths all over the place. By the time I have 5 cities (1 Fortress, 2 City, 2 Conclave is my norm), I have 12 towers all over the place and the AI has super swarmed.
What balances me is high frequency of events, rampaging monsters, distance between each target until it expands close to me and the need to acquire resources and empty a spot quickly.
Mid-Late game around turn 200, I have around 1/5 of the map covered in outposts and 5 cities or less. The rest of my cities are taken from AI.
In the game I won with Porcipinee, fortresses were my lifeblood. Eventually the areas I had popped forts around to capture and connect a path of resources, I found areas I could settle with monsters in my killing range. I had occasional setbacks with quests that caused monsters to appear inside of my undefended fortress zone on SE Anthys (West Continent). Skeletons raged throughout my land and forced me to direct my rear armies to the south mid-invasion of Yithril to the north. I had taken 2 cities, and was going to swap drained units for fresh ones, but was forced to accept the loss of those 2 cities to get down there and repair the damage. This was a fun, unexpected challenge because it effected every player, and the army coming at the cities I captured were distracted by skeletons for a few turns which gave me time to get the gold to buy peace. Right as my army mopped up the last skeleton south, a blood moon came and the "story unfolded".
In the end, I was able to win by conquest. I had a late-game explosion in power which allowed me to demand and receive surrenders 10 or so turns into a war.
I also found a bug: You can make an AI pay for peace then on the same turn, declare war and if you drain their coffers you can make them surrender. The AI placement of outposts was good but they weren't able to make good build queues to consider the importance of a given outpost or priority to its defense.