Slitherine

This has been mentioned many times already, but I was wondering if the developers have anything to say about the AI's bad habit of spamming hundreds of low quality colonial troopers in the mid/late game. Sometimes they have so many of them that the monthly upkeep and the disband money alone would make it worthwhile to disband all of them, then use the cash and extra income per turn to purchase a few updated units (obviously this might take a few turns). Not only is it smart money-wise, but operations become very powerful against the hordes of weaklings. An orbital bombardment on a stack of 10 units of 2-power-each is way more effective than the same bombardment on a single unit of power 20. Plus, healing per turn is a % of a given unit's life which means that a more powerful unit can recuperate much more power in a turn than a weaker one (not to mention that the weaker one will likely die and the investment is gone altogether). The only advantage I can think of for having lots of crappy units is that it could take many turns for an enemy to capture a city because they have to kill every last garrison, but this strategy is flawed if their power is low enough for operations to destroy them.

I know that these units are often just leftovers from the previous era, but it's still a problem. It seems the only way to keep this under control is to periodically wage war with every faction in order to cut down their stockpiles of worthless units, but even so, I've seen AI purchase crappy colonial troopers (in time of need) when they have the ability to produce something much more formidable. Granted, I understand that in a time of need, you simply need purchase the best you can afford and if you have little cash on hand you must settle for something less than ideal. Regardless of the reasoning, it shouldn't be the human player's responsibility to make sure that all the AI stay engaged in war enough such that the obsolete units get destroyed (that's lame, and doesn't even work especially well).

Ideas:
1) If I recall correctly, the AI is programmed to pretty much spend all it's money all the time. If this is true, make unit upgrades the highest priority. I know that it's nice to speed up the production, but the AI already gets a morale bonus which boosts production. Plus, what's the point of purchasing stuff if it's obsolete?

AND/OR

2) The AI typically spends down to 0 credits each turn. Change this so that they spend down to X credits, not zero. X is a value that would change throughout the game and it would be based on the amount of money it takes to purchase a good unit. X will be different for each AI because "good" is based on that faction's current tech. When an "emergency" situation triggers, they can spend this money. An example is when a city is besieged and there's no garrison. Side note: I think this would also help the AI against the aliens in the beginning which they currently struggle with.

AND/OR

3) Every Y turns, the AI considers the most obsolete unit (omitting formers and colonizers) and decides to either disband it or upgrade it depending on these factors: current money balance, rank of the unit, and faction military strength. If this idea were implemented, there would have to be a way that the AI only considers obsolete units so that you don't have AI's disbanding perfectly good units. Perhaps disbanding should only occur for units that are very obsolete and very low rank.

Good post, I totally agree with you jdmillard and I would like to add my two cents :

I realized recently that all my games follow the 2 same patterns ( i play on Hard / very hard) :

- First one is an early aggression from the AI, typically from the religious/imperial/solar dynasty before you are able to reach mid game level units. Those games are tough, often turning into a desperate fight for survival with all your cities under siege. If your aggressor has conquered another faction before attacking you its GG most of the time, there is very little chance for you to come back on that, especially if AI gets to nuke long before you get the third tier city defense building. Even if you survive that first war, you will most of the time discover that another faction has run away with the game while you were struggling and coming back on them is really tough.

- Second scenario game : I survive the colonial era and reach the mid mecha era with a medium size modern army. From there everything get's very easy. Because of the poor AI decision in unit creations, winning war is really easy by wiping huge stacks of crap infantry with artillery, nukes and strenght 20+ units. It usually takes ages for the AI to respond by producing better unit, and by the time they do, I am in the process of getting my first transcendance units, so keeping a large military advantage is not hard. The only difficulty you may have is to let the AI tech to black holes and mass build them.

First scenario game are often frustrating, second scenario game gets old very quickly after you have been trough them a few time, sweeping trough the map with unmatched armies.

I think scenario one problem comes from a mix of problem between poor diplomacy system, too powerful faction bonus (especially religious one and solar dynasty) and absolutely no penalty for conquering another faction early in the game. Those things were discussed a lot in other thread.

Scenario 2 problems boredom comes from AI army composition. I agree with the solutions propose by Jdmillard and would like to add to those what was discussed in this thread http://slitherine.com/forum/viewtopic.php?f=173&t=47273 with basically the following conclusions

- Either lower the costs in credit of refitting
- Either implement the option to rebuild trough cities obsolete units ( my personal favorite)

This will be beneficial for both human player and AI, keeping the general level of armies on the map higher. It would also benefit high teching faction be enabling them to keep a better army than aggressive faction and survive more easily against the more aggressive factions.

Considering that every single 4X game I played I can think of suffers from the "AI builds tons of crappy units" syndrome, this issue doesn't seem something that can be easily solved.

BlueTemplar wrote:Considering that every single 4X game I played I can think of suffers from the "AI builds tons of crappy units" syndrome, this issue doesn't seem something that can be easily solved.

True. It appears that the AI does not build units, it buys units and therefore, it buys whatever it can afford and that usually is troopers. I've made some modifications that has helped this. In the units files there is the strategy line of code, if you make this 0.0 the AI will not build or buy the unit (most times, sometimes it will anyway), if you make it 1.0 it will always build that unit and it will build a lot of them. I've also done the same with some of the weapons making it outfit units with better weapon load out. Again, it doesn't always work but for the most part the AI has a much stronger army and not hundreds of poorly armed troopers.

I even gave the AI sub=1.0 (in the factions XML's) for refit but the AI does not appear to refit, so that was a bust. The best thing that worked for me was editing the strategy line in the units and weapons this gives the AI a percentage of what units and weapons to buy.

Yeah I think I'll change that the AI partitions its budget for purchases and refits, with the ratio depending on their troop numbers; so if they got a ton of units, they rather refit than purchase.

void wrote:Yeah I think I'll change that the AI partitions its budget for purchases and refits, with the ratio depending on their troop numbers; so if they got a ton of units, they rather refit than purchase.

I think that would make a big difference.

void wrote:Yeah I think I'll change that the AI partitions its budget for purchases and refits, with the ratio depending on their troop numbers; so if they got a ton of units, they rather refit than purchase.

Sounds good!