EDIT: This is in officially, as testing went well enough.
This is an experimental test of changes to the Science Cost for most upgrades and how much a units cap increases with each upgrade. It also includes bumped up initial caps for Mark 1 units (by around 30%), with Starships, Forcefields, etc starting with 2 units instead, but have slightly reduced stats to bring them back in line.
Why the Tech Costs?
Playtesting and community interaction showed that the current Tech costs are all over the place, with the higher marks having incredible efficiency, and the middle ones having horrendous. Essentially, it was always better to pick a generalist unit and max it out instead of diversifying, picking what units to upgrade carefully, etc. That feels pretty bad, so here they're smoothed out to be much more equivalent. Mark 4 in particular was a really bad return on investment.
This should hopefully also give more of a sense of "freedom" with your Science.
Why the Mark Caps?
Upgrades had ludicrous scaling and power spikes, with even Mark 2 making a units overall power 4x what it was, then at Mark 3, 9x what it was, and so on. This meant that the later Marks were absurdly powerful, capable of demolishing almost everything by itself, and with Fabricators seemingly returning in the future, that posed a future balance disaster. To that end, Caps increase by only 0.5x their original value instead of 0.8 per upgrade. This means for instance, Mark 2 makes a units overall power 3x what it was, Mark 3 makes it 6x what it was, and so on. It's more like what it was in Classic, essentially - smaller "jumps" in power between them.
An oddity: You start with a single Starship per type, but it can't increase by 0.8x, so it rounds up. This meant Starships had even better scaling than almost anything else, so they and any other units affected have been bumped to start with 2 units, still gaining an additional 1 per upgrade. Combined with the above change, this brings their scaling into line with everything else. Unfortunately, they've had to have some damage output nerfs to bring their Cap damage into line, but on the upside their Cap health is now more appropriate, more or less. In addition, you start with 2 Forcefields, which is bad for performance, but seeing as this only adds 1 per planet no matter how much you upgrade, this seems fine.
Why the bumped up initial caps?
Mark 1 units were pretty "meh" in their Cap power. Looking at saves people had, they tended to be un-built and ignored, which feels very against the idea of trying to find a use for everything. In addition, Schematic Servers were entirely ignored in several cases, and even I myself didn't find it enticing (especially with the current upgrade scaling) to capture any. So this is a hopefully decent power boost to Mark 1, and a smaller one to Mark 2.
In addition, energy costs have went down for affected Fleetships, Turrets, and Starships. Starships are also a bit cheaper to build individually, making their Caps more expensive to build than a Fleetship Cap, making up for the general lack of attrition (hopefully). Metal costs have not changed currently, but seeing as there still seems to be a surplus of metal later on, this doesn't seem too bad for now (though easily changed).
Installing
Download the file and unpack it. There should be 3 folders in there. Then follow this folder path:
AIWar2 -> GameData -> Configuration
You should see folders in there with the same names as in the download. Take everything in the download folders, and copy it into the equivalent folder in the game. So for instance, copy everything in GameEntity, then paste it into the GameEntity folder in your game.
To remove, just verify your game files and it'll revert. This will not break saves, though if you save then remove it, you may have a brownout and your unit caps will look a bit funny, being in the negatives suddenly.
Ending
This is something that has bugged me for a good while, so here it is as a publicly available test for anyone curious. If it goes well or people don't have problems, I'll possibly put it in officially. I haven't tried it in practice much yet, but math wise it seems to work.
I hope I've explained it well enough...I keep re-reading it and am really unsure, so see what testing and opinion is like.
There's an attached Notepad file with well...notes, on the Tech Costs. The list at the very bottom is the one in use with this experiment.
