Effect of Game Speed on AI Difficulty

[Blake]

There is no "gameplay" effect which functions differently for humans and AI when the game speed changes. Any effects are due to the AI or human play being more or less "optimized" for that particular game speed.

An example of this is that Quick Speed requires very careful consideration of when to go to go on an offensive war - units obsolete very quickly.
On Marathon the cheaper units (workers, settlers and military) relative to the tech pace, combined with the slow to obsolete military (both player and AI) means that you need to consider very carefully what to research. On quick speed you might neglect a military tech because it only takes 4 turns to grab, so if things go sour, no problem. But on Marathon that would be 18 turns - there's a world of difference between being behind militarily for 4 turns of a war, vs 18 turns of a war. 4 turns is only long enough for the enemy to cross your culture and bombard down your defenses, 18 turns is enough to conquer half an empire.

As a human you're fairly good at adapting to certain settings (it's what brains do), if you play marathon a lot you'll be naturally good at making carefully considered tech choices, if you play quick a lot you'll be used to neglecting that aspect of the game and will thus be stronger at other aspects.

The AI plays all speeds equally badly, to some degree or another. The way it tends to play - and the bonuses it gets - means it doesn't rely on making good decisions.

That is the difference - as a human, you gain skill by learning optimal play for the settings you play (the more varied the settings you play, the wider your skill set). The AI naturally has no such optimization and gets bonuses in place of optimization.

The AI's bonuses are equally applicable for all settings, while your human optimizations only work for the settings (similar to those) you've played.

[Solver]

As a human you're fairly good at adapting to certain settings (it's what brains do)

Speak for yourself, you flesh-and-bones creature

[couerdelion]

I’m fairly certain that Solver is right. The difference in game speeds is simply that things are slowed down at Epic and Marathon. Taking Marathon as an example, everything costs three times as much but this is also the extra time you have available to get the resources needed to get it.

There is just one thing that changes between speeds which this would be obvious by taking a hypothetical Marathon game and speeding it up to Normal speed. This is unit movement and attacks. When viewed at normal speed, a unit will move three times as fast and attack three times as often as one stuck at normal speed. For this reason the advantage at slower speed swings to the player who has the relative advantage at warfare.

Now if the human player wishes to forego military techs, it is correct that Marathon speed will give the AI an advantage relative to normal speed. Because when the AI realises that there is a weak neighbour, he will move in but the speed of the attack will be greater and leave the human less time to react to it. However, this is saying nothing more than demonstrating the AI here has a relative military advantage and the slower speed increases this. In most cases, the AI does not have a relative military advantage (because it does not know how to fight) so the slower speeds favour humans.

[Solver]

And it's most plainly visible in unit training times. On Normal and especially on Quick, when you get attacked, all of your cities can produce units in time for them to participate in the battle. On Epic and especially Marathon, the battle will be over in the time some of your less productive cities will need to build units.

[ben04]

Well you could always double production easily. Just edit the XML file for the speeds and set the time to 100% but the costs for everything to 50%.

I could see the AI being stronger then as it's in my eyes better at managing many units and cities than the human player. The humaner player just risks to loose track of where all of his units are and therefore play suboptimal.

[couerdelion]

Originally posted by ben04
Well you could always double production easily. Just edit the XML file for the speeds and set the time to 100% but the costs for everything to 50%.

I could see the AI being stronger then as it's in my eyes better at managing many units and cities than the human player. The humaner player just risks to loose track of where all of his units are and therefore play suboptimal.

This is just logistics. The AI might have no problem with 1000 units but it will not necessarily know how to co-ordinate their movements beyond following the “rule of stack” or the odd raiding group. When it comes to serious end of fighting wars, all the AI does is churn out units and launch an assault. As long as the human player has the capacity to offer sufficient resistance, then over time they should easily wear down this sledgehammer approach and turn defensive into offensive warfare (only this time without any major field army to defeat)

Of course, it is all relative. Is my AI neighbour really did have 1000 units then I doubt that I would have the capacity to resist, would probably accuse my computer of cheating and start another game

[ben04]

Suppose you had 1000 units, would you micromanage all that? I have serious doubts. You'd be a lot better coordinating 20 units. The AI however would be just as good or bad with 20 as with 1000 or 10.000 units.

[ZEE]

.

[Blake]

You're building libraries too early.

Note that on Marathon settlers and workers are cheaper relative to buildings/tech, this means on Marathon, the optimal strategy is to expand rapidly and build commerce improvements. In other words, spam out cities, acquire as many commerce bonuses as possible, get Calendar if needed, spam cottages along rivers.

Only once you can expand no more do you need bother with a measly +25% to research from a library - a library does nothing to increase the ratio of Income to Expenses.

HEre is the secret:

In Marathon games you should be trying to flatline your economy. If you have any spare income, you should be using it to found more cities and support more workers, only once you can expand no more, or once your economy has totally redlined, do you stop expanding.

Due to this methodology there is no point in having +25% science - +25% of 0 is still 0. It's only once the expansion phase is over, and once your income starts growing faster than expenses (due to a lack of ability to raise expenses, mostly!), does +25% start to help. And also if your economy is flatlined, then a +25% to gold will do A LOT more, if you have 100 income and 100 expenses, then +25% science does nothing (25% of 0), while +25% gold changes that to 125 income and 100 expenses, which means you have 25 gold freed up to research with. With a flatlined economy, gold multipliers are approximately infinitely more powerful than science.

Note it is very useful to get a single library in a good food city and run scientists near-permanently, settle the first Great Scientist and build an Academy with the second. This acts as life-support for your research, Currency and/or Calendar will end all your economic woes and the scientists will help you get there.

Basically the optimal research path for Marathon goes like this:

Bronze Working (whip, chop, copper)
Pottery (granary, cottage)
Writing (library, open borders)
Iron Working (chop jungle, iron - may not be needed for either)
Alphabet (Build Research(Warlords Only), tech whoring)
Mathematics (essential pre-req, stronger chops, aqueduct)
Calendar (If you need)
Currency
Construction (Simply essential for military, colos)

Priesthood, Monarchy and Code of Laws can sometimes be justified. Got wine and need happy? Monarchy is good. Organized and lack religion? Trying to get CoL could be worth your time.

Usually you'll trade for a bunch of those more expensive techs.

Also lightbulbing Philosophy can let you trade out of a tech hole.

[couerdelion]

Personally, I dislike running high military maintenance cost during the expansion phase. Unless I am building up a military for the express purpose of taking new cities, I will keep my military at a moderate level until the time when I want to knock about some of the AI.

But I have almost no experience at marathon level and here things would be different given the greater strength of units and lower price. Because the relation between units and buildings has changed, unit builds gain greater priority. I would then build more units and it would seem logical that I use these units to earn me something. Offensive war therefore becomes a more logical part of the game.

That said, Blake’s tech path seems to be relying on early game expansion first and then warfare although I guess the balance here might depend a lot on circumstances and difficulty level.

He hasn’t said much about the ancient resource techs (Hunting, Fishing, Agriculture, Animals and Masonry) although Agro or Fishing is a pre-req for Pottery. Without this, you might guess that he places them before pottery although you will notice that he has put granary first in the list of benefits from the pottery tech. This should indicate that this is the primary benefit of the tech for the simple reason that it allows you to combine whipping with a granary to unlock the early game production needed to fuel your expansion.

Ben04, I would not micro-manage 1000 units. But I would be able to divide them into army groups with specific targets and objectives and then send them off on their merry way. Each group would contain the usual suspects (medic, unit-counters, bombard, assault) and they would invade to obtain the maximum hit on the enemy in the shortest possible time. A number of other units might also be sent out to cut links, pillage, scout and generally hamper the defenders.

To my knowledge, the AI has little concept of these types of tactics. I don’t even think it can attack more than one city at a time. So apart from the pillaging units most of its units will simply join the huge stack that advances onto the first city, then the second, etc. Since most of its units are doing nothing more than making up the numbers, they are almost useless.

[Six Thousand Year Old Man]

I have only played about 10 games of Civ4 and most of them have been on Marathon/Noble.

I've played a few Huge Terra maps where obviously the SOD risk is there. I haven't been overwhelmed by the AI militaries; I think this is partially due to playing on Noble, and partially due to my taking the faster build times for military units to heart and building a lot of them. I'm at heart a builder, but I was spooked by the warnings I had heard about the military risks in Marathon mode.

So for me anyway, Marathon has forced me to change my playstyle. Yes, the unit costs are high, and yes, I'm not #1 in techs for most of the early part of the game. But having all those units makes it easy to target a weak/unfriendly/politically isolated AI. If I wasn't playing Marathon, I'd probably turtle a bit and focus on building up my cities. I'm not sure if that would put me any further ahead in the long run than would the early conquest that I usually get on Marathon.

So I guess my advice is... use those units. Burn some stuff down for the cash and run a deficit for a while. Grab a holy city and a capital and hold them.