Military Strength Computed Wrong? (TA)

Please move this post - not sure how I posted in this forum.

Any ships w/in range of the military sb when the sb has weapon/defense increasing modules will up the power rating of those ships.

They probably also had a large qty of crappy ships while you had a lower number of superior ships. A pile of crappy ships still adds up.

...but in the end, you still only have a pile of crap.

I've noticed some oddities with this too. Does anyone know what the actual calculation is?

The military calculation is (if I remember correctly) total offense + total defense + 0.1 * total HP.

As you can probably see, that doesn't actually tell you a whole lot about your opponent's military. Ranking itself versus its opponents is one of the major weaknesses of the AI.

The main problem seems to be the way potential military production isn't taken into effect. That makes it way too easy for a player to sandbag the AI, because they don't realize that you can spin your economy on a dime, and crank out a heavy war fleet.

They don't consider mobility of their war machine or loadouts, either. Three or four battleships with high defense and attack may not register much on the charts, but they'll chew up weak, plentiful ships like the Torians and Drengin tend to field without much trouble, even if they're worth more to the ratings. Similarily, defenders sitting in orbit are NOT much use in war, since they're only going to be used to garrison planets. In the same way, military starbases are purely defensive.

Despite all that, I've had AIs cowtow to my "superior" military, despite the fact that it was a bunch of spammed tiny fighters in the overlapping influence of 4 military starbases. Sure, you don't want to ATTACK that, but they're no threat to you. And I can't count the number of times that AIs have attacked my vastly superior warships just because they have thirty thousand super dominators that won't even scratch my paint.

Or the aforementioned "You've taken half of my planets and crushed all military resistance I've sent against you, but I'm still going to act smug and demand your surrender" behaviour.

Yeah that's what's really pissing me off - It used to be I'd always be a mid pack player but now it seems that the AI is so smug - I've reproduced this three times - Enemy has one planet left, but it's military rating is twice mine - even though im using superior firepower and ships. It's bogus because I can't offer a peace treaty.

Peace really needs to work on a system of "How have our ships faired against them" system and perhaps a modifier for "How have our ships faired against them in the past" so that they can better determine if they want to go back to war against you.

The AI should calculate military power not only based on ship's power, but strategy efficiency. If, during the war, it has lost ships in a 10:1 ratio against you, then it should evaluate your military strenght 10 times than it actually is.

That way, the computer will actually compute your strategic genius when considering if to go in a war or not. Makes me think of Rome's reputation in the Mediterranian, where 6 legions was said to be able to beat half a million of barbarians.

So, if you are a badass in strategy/ship design as opposed to the computer, they should take you VERY seriously.

It would be neat if the AI actually understood the concept of skirmishes, using them to test their opponent's strength. It could be done using proxies by sending them ships and then observing the results.

Better yet, the AI could actually run a few simulations pitting their ships against the ship of the line their enemy is fielding, make an estimate as to how many ships the enemy could construct in the amount of time the war is expected to last, and ensure that its own military is adequately prepared for the war before heading out. If it doesn't have enough ships in the right positions to be able to take out the enemy in x amount of time, then it needs to build more/better ships and move them to where they need to be. Declaring war prematurely is foolish.

Of course, all that is way easier said than done.

But the root problem - the "my 0 attack/defense ships give me massive military might" needs to be fixed - It's annoying and causing the AI to repeatedly declare war on me when they have no chance to win.

I agree here this is absolutly annoying owning an ai and it still thinks its more powerfull because of all his bs 2pc fighter spam being useless in orbit only to killed of in a rather repetative boring war. Geez make some average to fast attck fleets AI and who knows this game might be challeging instead of a 2pc crap fleets.

I've been saying this too for as long as I played Galciv2 (bought it when it came out!). Its a bit curious that the developers, who have shown great patience and involvement with the community, have seemingly ignored this issue.

Isn't it odd that if you steamroll the entire enemy fleet (+4attack/ship) with a few handpicked ships of your own (say, +40attack) in a matter of weeks this should have no effect whatsoever on their evaluation of their war effort? I know it would on me!

It doesn't make sense.

Its not hard to implement either, you could modify the AI in a way that it looks not naively on millitary rating, but also the potential ability of its ships to defeat your ships.

In fact, I think I'll refrain from buying TA until these core issues are addressed (spying being the other one, IMO). Improving visuals is all well and good, but I'd like to see more substance added!

Ok, though I think the premise of this thread, that a self interested ai inherently leads to either better gameplay or a better challenge from the human pov, is somewhat simplistic, i guess i'll throw in my 2 cents conscerning how relative military capability should be calculated:

Military Strength Ratio * Tactical Ratio * Weighted Distance Ratio

Military Strength Ratio - a weighted average of the ai's military, industrial, and technological scores over the same average for a given opponent. I have no clue what the weights should be, which is likely an empirical question, but something along the lines of .5/.3/.2 respectivly would probably be appropriate. Given the once ship produced per planet thing, maybe a planetory score should also factor into this equation.

Tactical Ratio - the weighted sum of the rate of fleet to fleet victories and planetary conquests. Again, the weights are likely a touchy thing here.


Weighted Distance Ratio - the least obvious part of this, and one that hasn't been mentioned specifically. The distance of the enemys planets is key in determining to what degree a civ will benefit from a war versus other civs. I'm no expert on spatial analysis, but I think a simple way of factoring in geography would be someting like:

The distance between the average coordinates of the enemy civ's planets (weighted by planetary industrial output) and the average coordinates of the ai's planets (unweighted)/ the distance between the weighted average of the ai's planets' coordinates and the non weighted average coordinates of the enemy civ's planets.

The ai will thus percieve a greater advantage when the civ has many, non industrial planets close to its industrial, ship producing planets, and less of an advantage when the ai has many industrial, ship producing planets near the player's less defensible, non ship producing planets.

A civ will perceive another civ as being equal in strength when the main equation has a value of .5, an advantage when that equation is greater than .5, and a disadvantage when it is less that .5.

An alternative way of factoring in distance would be to simply use the ratio of the average coordinates of the player's planets to the average coordinates of the enemy civ's planets to the average distance between average enemy civ coordinates and all other civ's average coordinates.

Bump for dev view.

The a single game's challenge might also may be mainted more consistently throughout if diplomatic relations suffered as a player got a higher and higher percentage of planets habited by major races. The number of habitable planets is the best proxy for player skill at any level (with the exception of the early part of higher levels given the ai's colony production advantage). It would make sense from the "ai like a players" perspective for the ai to make a more dramatic breaker with the planetary frontrunner than simply planting spies on the planets. It also would seem relativly easy to code, but that is speculation, and I'm sure there are factors that I haven't considered (fun).