Common AI mistakes and tech speed
from 
GalCiv2 Forums
GalCiv2 Forums
Purge asked for my opinion on game difficulty as it related to tech speed on smaller maps in another thread. I've been (rightly) accused of thread-jacking lately, and as my reply started to get longer I realized I should just make this a new topic.
First off, and as a slight response to an excellent point by Mumblefratz, suicidal is suicidal. If you haven't been playing this game for a very long time and come to one of these threads it is easy to be misled. When I talk about a map size being easier or a tech speed being easier, I'm discussing incremental differences that only become apparent after playing a lot. Changing your map size and tech speed won't let you change difficulty levels. It will require that you adapt to a new strategic situation. There are an insurmountable number of variables that go into any game state - so when talking about "difficulty" here (I'm assuming suicidal level) I'm referring to broad over-all trends that I see over time, and the differences are probably (no offense to anyone) going to be almost imperceptible to anyone but an expert. In this thread I'm splitting hairs. At a point you get deep enough into the game that splitting hairs is still an interesting discussion.
That aside, to really answer Purge's question, I need to discuss game versions. In the old days of DL the slowest tech speed was IMO the hardest - because of the lack of flexibility if you made a mistake. If you found yourself outclassed and with few planets (on smaller maps) there wasn't the time typically to make up the deficiency in tech and adapt. On very fast tech the game was easier because if you realized you were being outclassed in an area (a weapon type vs. defense, for instance) you could focus your sliders on research and catch up.
However, I now think very slow research is the easiest tech settings on smaller maps ever since DA. The reason is the AI now consistently makes a mistake.
The thing with very slow research is it IS still the most unforgiving if you make a mistake. But that applies to every race, not just the player. Ever since the release of DA, even on suicidal, the AI always makes the same mistake when playing with very slow tech, and so the game is now ridiculously easy (splitting hairs people, this is relatively speaking). The mistake? Extreme colonization tech.
As long as there is at least one extreme environment planet on the board the AIs can be counted on to waste gobs of resources and precious turns researching environmental tech. Why? So that they can get an extra planet or two (or three!) that operates at half capacity. In the time it takes the AI to greedily research and claim a few extra worlds, I am able to leap ahead in actually useful tech. Because of the very slow speed the AI then never has a chance to recover from its mistake. And its a fundamental mistake the AI makes about 98% of the time.
Why is it a mistake? I'm sure Purge and Mumble (and many of you) already know this, but I feel like pontificating. The reason is because of Planetary Invasion. Unless you are playing the Altarian Tech Victory Rush... Planetary Invasion is THE technology of GalCivII. Planetary Invasion and Hyperspace technology are the only two you will always need, and every race but the Custom race HAS to start with Hyperspace technology.
Planetary invasion allows you to invade worlds that you can't colonize, is the key to any blitz strategy, and is the most necessary tech once you want to actually start winning the game. While the AI wastes time researching extreme colonization techs I research planetary invasion and when I start declaring war the AI can't even launch troop transports. It has no means of fighting me at all. And it will take it (typically) months to research the tech on very slow. Plus, all of those planets the AI colonized that I couldn't grab in the colony rush I grab anyway in the troop transport rush that I start preparing for almost immediately.
If the Iconians, for instance, had "Extreme Aggression" in place of "Extreme Colonization" as their superability - and instead of getting a few extreme colonization techs just got to start with planetary invasion, they would instantly go from being the worst race to the best race. Planetary Invasion is that good, and extreme colonization is (at least on medium and smaller maps) that bad.
The thing is, I have learned, and the AI has not. I know to check what the tech speed is on the first turn (I play with random settings, but I can look at techs I've memorized the cost for, like Planetary Improvements, without changing my sliders the estimated time for completion right after game start tells me the tech rate). Once I see the speed is slow I know the extreme planets are off limits because of how prohibitive the research cost is for the teeny tiny benefit. The AI has never learned this fundamental and repeatedly makes the same mistake.
If I'm playing against a few opponents I can often conquer all meaningful resistance before any enemy has planetary invasion. If I'm playing against many opponents I can usually knock 3-4 races before someone researches PI and the AIs all start trading to get it from each other. But with 3-4 races worth of planets I usually have an insurmountable lead for the rest of the galaxy at that point. On small maps you don't want to build colonizers - 60-70% of the time you want to start off building troop transports.
See for instance my Altarian Rebellion AAR. Purge, I know you at least read it because you commented on it. There I completely ignored the colony rush - and there was one aquatic world out there and 7 of the 9 AIs wasted time researching aquatic colonization. By the time the AIs were ready to fight (The Terrans declared war against me about 60? turns in) I had already conquered half the galaxy and was untouchable.
On larger maps extreme colonization, at least in theory, can sometimes be useful. I have played games on larger maps though - and have never had a problem with going straight for planetary invasion in place of any extreme colonization tech. Personally (feel free to disagree) I think until late game extreme colonization is always a mistake no matter the map setting.
But on a small map? Never. Sometimes I'll get a small map with no extreme worlds and a very slow research rate, and then I go back to my original analysis of those being the hardest tech rates. But that is rare. More times than not there will be a hazardous world, the AI will predictably make a mistake, and the game is won before I've even hit the "End Turn" button for the very first turn.
Hope my analysis helps,
- Wyndstar
First off, and as a slight response to an excellent point by Mumblefratz, suicidal is suicidal. If you haven't been playing this game for a very long time and come to one of these threads it is easy to be misled. When I talk about a map size being easier or a tech speed being easier, I'm discussing incremental differences that only become apparent after playing a lot. Changing your map size and tech speed won't let you change difficulty levels. It will require that you adapt to a new strategic situation. There are an insurmountable number of variables that go into any game state - so when talking about "difficulty" here (I'm assuming suicidal level) I'm referring to broad over-all trends that I see over time, and the differences are probably (no offense to anyone) going to be almost imperceptible to anyone but an expert. In this thread I'm splitting hairs. At a point you get deep enough into the game that splitting hairs is still an interesting discussion.
That aside, to really answer Purge's question, I need to discuss game versions. In the old days of DL the slowest tech speed was IMO the hardest - because of the lack of flexibility if you made a mistake. If you found yourself outclassed and with few planets (on smaller maps) there wasn't the time typically to make up the deficiency in tech and adapt. On very fast tech the game was easier because if you realized you were being outclassed in an area (a weapon type vs. defense, for instance) you could focus your sliders on research and catch up.
However, I now think very slow research is the easiest tech settings on smaller maps ever since DA. The reason is the AI now consistently makes a mistake.
The thing with very slow research is it IS still the most unforgiving if you make a mistake. But that applies to every race, not just the player. Ever since the release of DA, even on suicidal, the AI always makes the same mistake when playing with very slow tech, and so the game is now ridiculously easy (splitting hairs people, this is relatively speaking). The mistake? Extreme colonization tech.
As long as there is at least one extreme environment planet on the board the AIs can be counted on to waste gobs of resources and precious turns researching environmental tech. Why? So that they can get an extra planet or two (or three!) that operates at half capacity. In the time it takes the AI to greedily research and claim a few extra worlds, I am able to leap ahead in actually useful tech. Because of the very slow speed the AI then never has a chance to recover from its mistake. And its a fundamental mistake the AI makes about 98% of the time.
Why is it a mistake? I'm sure Purge and Mumble (and many of you) already know this, but I feel like pontificating. The reason is because of Planetary Invasion. Unless you are playing the Altarian Tech Victory Rush... Planetary Invasion is THE technology of GalCivII. Planetary Invasion and Hyperspace technology are the only two you will always need, and every race but the Custom race HAS to start with Hyperspace technology.
Planetary invasion allows you to invade worlds that you can't colonize, is the key to any blitz strategy, and is the most necessary tech once you want to actually start winning the game. While the AI wastes time researching extreme colonization techs I research planetary invasion and when I start declaring war the AI can't even launch troop transports. It has no means of fighting me at all. And it will take it (typically) months to research the tech on very slow. Plus, all of those planets the AI colonized that I couldn't grab in the colony rush I grab anyway in the troop transport rush that I start preparing for almost immediately.
If the Iconians, for instance, had "Extreme Aggression" in place of "Extreme Colonization" as their superability - and instead of getting a few extreme colonization techs just got to start with planetary invasion, they would instantly go from being the worst race to the best race. Planetary Invasion is that good, and extreme colonization is (at least on medium and smaller maps) that bad.
The thing is, I have learned, and the AI has not. I know to check what the tech speed is on the first turn (I play with random settings, but I can look at techs I've memorized the cost for, like Planetary Improvements, without changing my sliders the estimated time for completion right after game start tells me the tech rate). Once I see the speed is slow I know the extreme planets are off limits because of how prohibitive the research cost is for the teeny tiny benefit. The AI has never learned this fundamental and repeatedly makes the same mistake.
If I'm playing against a few opponents I can often conquer all meaningful resistance before any enemy has planetary invasion. If I'm playing against many opponents I can usually knock 3-4 races before someone researches PI and the AIs all start trading to get it from each other. But with 3-4 races worth of planets I usually have an insurmountable lead for the rest of the galaxy at that point. On small maps you don't want to build colonizers - 60-70% of the time you want to start off building troop transports.
See for instance my Altarian Rebellion AAR. Purge, I know you at least read it because you commented on it. There I completely ignored the colony rush - and there was one aquatic world out there and 7 of the 9 AIs wasted time researching aquatic colonization. By the time the AIs were ready to fight (The Terrans declared war against me about 60? turns in) I had already conquered half the galaxy and was untouchable.
On larger maps extreme colonization, at least in theory, can sometimes be useful. I have played games on larger maps though - and have never had a problem with going straight for planetary invasion in place of any extreme colonization tech. Personally (feel free to disagree) I think until late game extreme colonization is always a mistake no matter the map setting.
But on a small map? Never. Sometimes I'll get a small map with no extreme worlds and a very slow research rate, and then I go back to my original analysis of those being the hardest tech rates. But that is rare. More times than not there will be a hazardous world, the AI will predictably make a mistake, and the game is won before I've even hit the "End Turn" button for the very first turn.
Hope my analysis helps,
- Wyndstar
Wishful thinking and all, I know, but still, it'd be fun to be able to see these kinds of things and attempt to fix them. But that's the programmer in me talking, heh.
Seriously I'm glad if I can be of help. I've always really appreciated your response in fixing the combat imbalances I preached about from the very first version of DA, so I'm happy to do anything I can.
I'm pretty confident I could hold my own against most of the players here if there was multiplayer, although I by no means claim to be the best (and Purge officially holds that honor, credit where credit is due). This whole thread is really just a couple of longtooths sitting down and talking shop. I HAVE played this game enough to the point where the AI researching half a dozen techs in one order or another makes the difference between a walkover game and a tough challenge. I already know much of what I will face, which is what allows me to predict a win very early on in very specific situations. That may mean I'm a good player, it may just mean I'm a freak (or both). There comes a point where you become familiar enough with the game where you can exploit just about ANY weakness your opponent shows, even something as simple as research order. I tried to warn at the beginning of the thread though - I'm really splitting hairs here. Most people won't notice these differences.