AI suggestions

I'll start the ball rolling with a couple of suggestions:

Invasions

My invasion attempts very nearly always succeed because I never invade without a force which I'm confident is large enough to win comfortably. Failed invasions are costly in time and money for invasion tactics and the destroyed transports. I think the AI should check more carefully that its invasions have a realistic chance of success before launching them.

Related to the above is transport design. If I'm going after extremely heavily populated planets, I won't mess about with troop ships carrying 1 or 2B. I'll design a few carrying 5-10B. I've never seen the AI do this.

I always use considerably more troops than I need to take a planet becuase immediately after the invasion, it's the surviving forces that will make up the population. Low population planets can be a serious economy drain. On those occasions where I do end up with a low occupying force, I look to move population there as soon as possible. This is something I think the AI should replicate.

Filling Troop Ships

Very often, I will not fill a troop ship from the same planet it was built, but I will instead send it on a round trip of local planets. As mentioned above, I often have very high capacity ships so this is vital. The AI tends to keep churning out troop ships and fill them from the same planet. This leaves too many of it's worlds drained of population which in turn drives it into an economic hole. If it's going to do this, again it should be looking to replenish that population from one of its more well populated planets at the earliest opportunity.

The AI never seems to use "Pearl Harbour" tactics. It gives loads of time for you to prepare for war. When V1m declares war, you'd better believe that a lot of ships will be destroyed by the end of the following turn.

Building a big ole military base by the home planet of and donating it to a minor race at war with someone you don't like but don't want to fight is another neat tactic. Especially when combined with the sale of a few semi-obselete medium hull destroyers.

+ The AI doesn't seem to build "tailored" fleets. When I'm planning to go to war with a race, I'll research as much defense as I can to match their offense, and research enough weapons that they don't have defense for. Thus my custom fleets can have less offense than they have but still win solidly.

Once I saw the Drengin AI attempt to do so. He had all missile offense with no defense, and declared war on me. I was using beam weapons and also researched missile defense (because I knew that the Terrans and Drengins were both using missiles, and they were my immediate neighbors). Later we negotiated peace, and I noticed he started to build a little bit of beam defense into his ships. I switched to missiles, and when I'd see him in battle with another race, he still had beam defense, and kept making ships with more and more beam defense. When we went to war for the second time, I had only missile-offense, and his miniscule beam defense was useless. Almost every other race was using missiles, so it didn't make sense for him to continue pumping out beam defense ships.


+ I don't think I've ever seen the AI upgrading a ship. I could be wrong though. Upgrading ships that have experience is a good thing if you have the cash to do it, plus it's cheaper than building brand new ships when you need some quick modifications.


+ Another thing I do is when I get ready to attack, I'll pull out my weakest ships so they aren't destroyed. Then as soon as the attack is over, I'll put them right back into the same fleet.


+ Some may call this an "exploit", but I don't believe it is. When I don't need any ships, I will set my planets to build the most expensive ship available. As I research better tech, I will upgrade (and obsolete) the old design. This automatically changes the ship being built by my planets to the new one. When my planets are 1 week away from completing a ship, I will stop them from building. Then when I need ships (when someone declares war on me, for example), I'm one week away from every planet producing my most powerful ship.


+ I "cherry pick" ships from planets. Once the initial onslaught is complete and all ships surrounding a planet are destroyed, I will station a smaller fleet nearby to cherry pick any newly created ships. This gives me time to take out all their defenses before sending vulnerable transports in. Once I've gotten a few nearby planets cleared of defenses, I'll station one fleet in the middle of the planets so that I can easily destroy and ship created. The AI doesn't figure this out and doesn't sent the ships out of orbit immediately after building them, so they're easy pickings.

I redevelop my planets. The AI sometimes builds a farm on a precursor mine after it works out that it can't afford to support a factory there. That would be fine if it built that factory later and made use of the bonus. If there aren't many planets 1 precursor mine can win the game for its owner.

The AI isn't very efficient with its military. It requires far more ships to defend your planets with "defenders" as the AI does than to defend them with roving fleets. Because there are so many more ships with the AI approach, it's also harder for it to keep all its forces at or near your current latest ship models. It ends up spending huge amounts of money maintaining an obsolete fleet.

The AI also doesn't appear to assess its chances of success before attacking a fleet, or if it does then it's not working very well.

Not only does the AI need to redevelop it's planets if there is a good reason to do so, it needs to do a good job of developing them in the first place. This is the only reason I can win on Masochistic and above. With the bonusses that it gets on the higher levels, if it actually developed planets as well as I do then it would be almost impossible to beat. And that's really what we all want isn't it?

I agree mismatched, but could you give a bit more detail, please? What exactly do you think it's doing wrong? What should it be doing instead?

Not only does the AI need to redevelop it's planets if there is a good reason to do so, it needs to do a good job of developing them in the first place. This is the only reason I can win on Masochistic and above. With the bonusses that it gets on the higher levels, if it actually developed planets as well as I do then it would be almost impossible to beat. And that's really what we all want isn't it?

for example in my current game (which is in genius AI), the altarians have a class 20 planet. sure it has great and advanced infrastructure but not all tiles have been filled up. the home planets and the ones colinized earlier seem to be "ok" but the highher class ones that are colonized later into the rush are not completely filled up. i noticed this on my games on the higher level and also with the drengin. i think the ai should prioritize building up high class planets since it is after all high quality   

by the way great job random 50! this post should be stickied imo


ill post some more things that i noticed playing on higher difficulty levels a lil later

Planet development and planetary improvements:

The more advanced a research lab or factory is, the more expensive it is both to build and maintain relative to its output. This difference is often substantial. Consequently, the best way to develop planets is this:

1) Fill every tile with low tech buildings first. The very first planetary improvements are probably just about worth researching, the industrial one for basic starbase modules, the research one because the *early* buildings are not substantially more expensive than the others.
2) When almost every planet is fully developed *and only then* start moving down the research trees. Now all the space is filled, the only way to improve production is better buildings.

The AI's main problem here IMO is moving too quickly down the research tree. It ends up with a (somewhat diluted) version of the Dread Lords problem (and only thing that makes them beatable!) whereby it doesn't have sufficient infrastructure down to build new infrastructure in a reasonable timeframe.

Rush Buying

Rush buying is almost never a good idea. If you have a surplus, it is far more efficient to build things until the surplus is used up than it is to pay approximately 10 times the amount to rush buy.

There only times rush buying is good IMO are

1) The colony rush, but *only* if there's a very small amount of available planets, and even then it is often better to rush buy factories instead. On large maps with lots of habitable planets, you shouldn't buy anything at all. The goal here is to stretch your starting 5000 as much as possible, not quickly grab half a dozen planets then sit there in recession for ages.
2) Times of war where inadequate planning has left you militarily weak. Even there, it's not an automatic call that you should rush buy ships and try and fight rather than surrender a few planets whilst building up your military then taking them back later (personally, I always fight to the last for every planet, but I don't actually think that's great strategy)
3) You are racing somebody to complete a Wonder or Trade Good. Even there, think very carefully whether it's worth the cost.
4) If you're stuck in the situation of being a long way down the tech tree with planets lacking infrastructure (maybe through scorched earth tactics), rush buying the first factory to kickstart things can be worthwhile.

I get the impression, maybe wrong, that the AI just rush buys as soon as it has the money and doesn't make very intelligent decisions on exactly *what* to rush buy.

Upgrading and Decomissioning Ships

This is similar to rush buying. Upgrading ships can be very expensive. If your military is at full strength numbers wise, it may be a better idea to slowly decommission and rebuild rather than upgrade. The things you lose by doing this

1) Position. Potentially anyway. You'll need to get those ships back out from the shipyards to wherever the ships they're replacing may be.
2) Time. Whilst your shipyard is building these new ships, it could have been turning out something else. Or you could have used that production on social or research.
3) Experience. The most important one IMO. Decomissioning a highly experience unit with twice the starting hitpoints is rarely a good idea!

The point is, it's a balancing act. Personally I only upgrade the experienced ships. The others I either decomission or change their duties. The AI seems too obsessed with the numbers game. I often see ships that are 3 or 4 generations out of date being used for main attack duties. Great news for me - loads of experience without a scratch - not so great for the AI which has been forking out loads for the maintenance of essentially useless ships. Worse, it doesn't even get the cash back from decommissioning.

I play at Suicidal and win. I'll admit I haven't looked at the above posts so I apologize if I rehash. I avoid cheese tactics like conquering planets only to return to reconquer again. Only thing I do is tech trade to stay in the game economically, but I usually halt that mid-game as I start to pull ahead. But here are some things the AI can do, which if done correctly, will whoop our asses:

1. Improve planetary improvement buildup! In version 1.2, the AI has gotten better about what to build, but not about the pace. I'm still surprised to find mid- to end-game AI planets that aren't fully developed! I realize AIs already have bonuses and may see additional buildings as a waste, but imagine how much more deadly they would be if they actually had more factories/labs/econs. They're good about developing bonus tiles, but too slow with the rest.

2. Have the AI actually SPEND their excess cash. I don't see the point of keeping excess cash, especially if you're starting to lose a war. Spend the goddamn money and buy the ships/planetary improvements and crank them out. It amazes me to no end that the AI just sits on piles of cash.

3. What wins the wars for me is constant updating of my vessels to match the AI I'm attacking/defending against. That is the only reason I may pile cash for. Have the AI do the same. And have them update before they start attacking. Or, more importantly, have the AI move their ships outside of my range before updating. A battleship with one hitpoint is no match to anything.

4. Have the AI prepare for war...not just declare it without having been prepared with appropriate fleets/transports. Again, that leads to updating the attacking fleets to defend against my strenghts and attack my weaknesses. Move the fleets towards the planets/bases then surprise attack just like every Human player does. This would go a long way to making a game more challenging.

As an ancillary to 4, the AI should also take into consideration my economy/CASH reserves and CURRENT TECHNOLOGY before attacking. On paper, it may show that I have a weak military/outdated ships. But that can change in an instant the second I update my attack ships to improved weapons/defenses. This gets the AI every time and I manage to plow through their fleets with relative ease.

5. As heard before, improve diplomacy so that there is better coordination between allied races in battles. That is usually the only way to overpower a skilled human player.

6. Allow the AI to break alliances and gang-up/attack when it becomes clear that the Human player, even Good ones, are on a conquering war path. Usually to slow to realize and since they don't coordinate, they're toast.

Diplomacy.

I don't give a flying %^&* what their dipolomacy tech level is, while they care about mine enough not to press a military advanatage when they have one.

And I don't get fooled by spin control centers.

I thought I got fooled by a spin control center in a game. There was a planet with about half a dozen defenders, but when I attacked, I only killed three or four before they were all gone.

I only begin invading when I've researched the whole troop tree, and Tri-Quan is on it's way. The result is that my attack bonus:9 faces defense bonus:0~3 and it's a walk in the park. The AI should consider upping it's troop tech as soon as it sees troop ships with terrifying legions headed for their worlds/nearby. Something clever here would be to be able to set my troop ship's destination to a different race ("Oh, we're en route to Nimbus-3; just passing thru") then pull a bait and switch. They would have to be able to do the same though; like sending a big troop ship nearby but going to someone they are at war with, then pull a fast one and gank my big frontier shipyard.

I also monitor enemy technology levels with the diplomacy screen, and by examining attack/defense of enemy ships in range (scanning); so they should be able to do the same, and plan their ship defenses and troop strengths accordingly.

I also trade tech profusely for technology, upgradable scout ships, and cash. The value I get doesn't seem to be well-connected to the game value of the techs. I can sell the AI Habitat Improvement for big money, then go back and cancel habitat improvements from my own build queues, but the AI gets stuck behind a couple dozen turns of no production. But they have a hard time trading away Basic Logistics, which requires only a few turns to research (150RP).

i agree with random in reply #9 -- i put in basic factories, labs, + markets before going down the tech tree in those areas. I usually tech trade up as soon as i have the structures in which usually means tech trading for factory upgrades as soon as i get my factories in then tech trading for lab upgrades after i get them in and the tech trading for market upgrades when i got a lot of markets -- hard to fill up all the squares on all the planets as some planets run out of tiles -- so sort of have to balance it out.

The AI is weak on using speed to its advantage -- i almost always have the speed advantage with my ships and fleets when i go to war. I've seen a few escort ships that have some speed but have never seen a ship with just 1 weapon and loads of engines. Their survey ships are usually slow and outdated. If they updated them with more and better engines and put a weapon on em, it would be a tactical challenge to have to deal with em. Personally, i upgrade my survey ships like that when i'm done exploring.

i use speed to my advantage from the start -- I tech research straight to impulse drive( 100% spending, 100% research ) -- I buy 3 survey ships ( basic survey no engines, no add ons ) right away -- I upgrade them with 3 impulse engines and 3 basic life support when i get impulse drive tech -- then out to hog all the anomalies ( i probably get 60-70% of them ).

I take +50% military production bonus and taking +50% population last time on race bonus's ( custom race ); And take Tecnologists party ( +20% research ).
I take Galactic Warfare, Xeno Engineering, Stellar Cartography, and Sensors as my starting techs.

I Buy the equivilant of at least 2 labs as fast as possible then I buy or build the equivilant of 4 factories as fast as possible ( no building till after getting Planetary improvements tech )-- this means I buy a lab on my best tile on 1st turn and buy a lab on another tile if I didn't have any research bonus or if I have a second tile with a research bonus. Then I buy factories for the equivilant of 4 basic factories ( if I have the money ) by the time I get PLanetary Improvements.
If I don't have the money -- I build em ( the labs cost less and save as much time as it takes to build factories ).

Then research space militarization ( +10 military production )
Then research Planetary improvements ( + 10 military production, +10 social production , + 10 research )

In here I build 1 entertainment center so i can set my tax rate to get 100% morale.

Now is when i turn down research % and add some military production % to get a colony ship every 2 turns.
Then I design a fast colony ship with 3 impulse engines and crank out a colony ship every 2 turns -- get morale to 100% and put 300m to 500m colonists on each colony ship -- I keep homeworlds population even more or less

Then usually research Xeno Communications, Universal Translator, and Xeno Entertainment ( +10 morale ) so i can set morale to 100%. This is necessary to get 100% morale on colonized worlds i believe.

Then research impulse drive2, impulse drive3, and warp drive --
Upgrade survey ships anytime it is convient.

I use the starter colony ship to extend the range of my ships and try to use the colony ships for that too -- of course it is a balancing act -- don't want to miss out on good worlds or let my neighbors colonize inside of my area.

I buy a starport right off on all my worlds ( if i got the money ). I'll also buy a factory or lab if i have a +300% or +700% tile bonus.


This strategy works for me on suicidal, gigantic map, middle of the line map options, 9 opponents, tech trading on, normal research rate. I like getting to all of the anomalies -- fun to sort of tail their flagships and not let em have any anomalies. I'm a little slow starting the colony ships but the speed of the ships makes up for it and the survey ships will have given me a good idea of how and where i want to expand. The AI has a lot of trouble adjusting to what i'm doing -- I've seen em ending up with 10 or so extra colony ships at the end of the colonizing.

I also agree that the AI seems to give a lot of warning when it goes to war and The human player doesn't even have a way to give warning as far as i know.

I'm really looking forward to update 1.3 -- hope they got some surprises for us.

Sneaky Planet/Resource Stealing

Now I don't do this a whole lot, but I've noticed it's very easy to ride on the back of a friendly or neutral's war efforts. You can let the AI do all the work taking out enemy starbases and planet defences then nip in with your transports and contructors and steal the prize from under their nose. Two suggestions here -

1. Instead of the current system in which the AI appears to have distinct
"wipe out the military" and "invade planet" phases, have them do what human players more often do and combine the two together. Personally I usually invade the same turn I wipe out the last planet defences.

2. If the AI ever did this to me I would be royally p*ssed off. It should be too! Instead, it doesn't seem to care at all!

I've had the AI come in and steal a resource after i wiped out a starbase on it. Now i maake sure i have a constructor along to setup my own starbase.