Different playstyles and how game mechanics & exploits affect your preferred playstyle and enjoyment

[Buttercup]

there is this thing about controlling resources and whatnot , if not within the zero culture city limits , then certainly within the first cultural expansion . You look at a place , me myself would consider rivers to increase trade , having two or more resources or bonuses to used directly by that town , so start up moving that way and there will also be Al settlers on the way , because there is a gap there and it has to be filled , mostly with a disgraceful location that will waste tons of options and capturing the city will still be of no use , because ı wouldn't be able to replace it . Yesterday ı got the Al park into a spot where a previous city had existed and razed in war between 2 civs , and ı could two settlers of a different Al stop and turn back because the spot they marching to was already taken . Two settlers , let me tell you .

Yes indeed, settling is a very big deal for the AI. And, yes, it's settling pattern does seem strange to a human being, but for what the AI wants to achieve it is ideal. The AI's priority isn't the Cow or the River, the AI's priority is land-grabbing and making sure it occupies as much land as possible when its borders expand. It settles as if it was settling a 100 culture Metropolis, not a town hoping to immediately take advantage of its immediate surroundings.

And even when all the land has been settled, the AI will usually always have a few settlers in reserve, for just such moments.

 

[Chicken Pizza]

Would chess or tetris be fun if the main goal was to share exploits with each other? No, so why should one of the best strategy games, Civilization, be balanced and watered down, when there could probably be lot more interesting discussions, experiences and stories if the difficulty setting meant changing the number of turns that the AI think in advance.

 

[Buttercup]

Would chess or tetris be fun if the main goal was to share exploits with each other? No, so why should one of the best strategy games, Civilization, be balanced and watered down, when there could probably be lot more interesting discussions, experiences and stories if the difficulty setting meant changing the number of turns that the AI think in advance.

I'm very sorry, but Chess doesn't regularly make the top 10 computer games of the year and Tetris is an easy game that is made hard by increasing speed, not complexity. As for the rest of your post, interesting for whom? You and 10 other people?

I find it fascinating that when you care to mention 'competitor' examples to Civ you fail to mention any and instead mention two completely different kinds of games. If it was so easy and so obvious to make a strategy game that was both popular and as complex as playing Sir Pleb when you are just on 'normal' difficulty, don't you think you'd have said something like "Well, [this similar well known strategy game] does it!" Alas, you don't

 

[need my speed]

I do agree with the point that the AI exists for fun, not for competitiveness. Compare the irrational 'playing to win' AI of Civilization V with the immersive personalities of Civilization IV; the latter is clearly superior and preferable, to me.

 

[Theov]

Would chess or tetris be fun if the main goal was to share exploits with each other? No, so why should one of the best strategy games, Civilization, be balanced and watered down, when there could probably be lot more interesting discussions, experiences and stories if the difficulty setting meant changing the number of turns that the AI think in advance.

Some like the one, some like the other.

 

[vorlon_mi]

There's quite a lot to unpack here, while at the same time not really enough specific information to know whether each point you make is indeed relevant to the conversation.

For example - you state "I don't have any relevant experience w.r.t. the game consciously trying to hinder the human player" and then use the rest of your post to detail how you always play the game exactly as the developers want you to play the game in its most text-book way possible - which, ironically, doesn't actually disagree with anything I've said, and, in fact, reinforces what I've said, in that it is wandering away from the assigned path that gets all the weird things happening.

And you play the game in that way because that's the way the mechanics have 'encouraged' you to play, not because you've started the game thinking "I want to play the game in a different way today".
.

While I was away, the thread moved into a discussion of "coincidences" that you (and others) attributed to the AI predicting what the human player will do, and consciously opposing that. An example was given of a "perfectly timed settler clash." I do not have corroborating experience for that phenomenon. I don't find it very surprising that the Hittite settler was on its way to settle a city about 5 tiles from its last city. That's pretty typical loose city spacing, where the AI assumes every city will grow above size 12.
I don't want to weigh into that discussion. I have not observed as many "weird things happening" as you have.

Part of the reason I don't subscribe to the "AI is changing its behavior just to thwart me" theory is that my day job deals with computers and computer programming. The complexity of AI programming, especially given the state of development tools in the early 2000s, means that the AI in Civ3 is not very complex. The target PC where it would run, back in the day, was pretty underpowered. Its algorithm for deciding what to build, what to move, and when to plot war is simple and predictable. Indeed, on this site a number of contributors have explored the tendencies and flavors of the table-driven code that the AI uses. Part of the reason I know what I know is because I rely on knowledge shared here, not confirming my own particular biases.
The developers knew this; they only way that they could make the computer opponents tougher at higher difficulty levels is to change the values of production bonuses, research costs, and starting units. It has been known for decades that the AI does not have a fog of war and that the AI gets happiness bonuses similar to the human at "Settler" difficulty level. This is not an ideal game. It is a game of its time. It's not even the best Civ game in the franchise. But on to my other points ...

Regarding the details of how you always play the game, the kind of details you miss are all the settings you usually use. For example, a tiny archipelago map on Warlord difficulty is going to be a very different experience to a huge sid Pangea game, not to mention all the other options the game gives you. You also don't illustrate what kind of starting positions you 'always' play at when describing the above list 'what always happens with me'. And the screenshots would be useful for seeing your starting position, where your Iron is, exactly what turn you get the Philosophy bump, and, more importantly, What turn you are learning Monotheism in. Because running pure tech is a lot more than just getting a quick Republic bump anyway, especially if you're just going to slow down any future progress from that point on with continuous warfare.

You make an excellent point at the end of your post about the AI knowing what strength of army you have and behaving accordingly. Why does the AI magically know this information? What is the purpose of the AI always making its diplomatic and strategic decisions based entirely on what your military value is? Surely, in an ideal game, shouldn't one be able to perform such manoeuvres as "bluff" and "illusion". What you are saying here is a reinforcement once again of my point, the game is inherently forcing you into its preferred style of play, if you don't expend heavily on those more expensive units then, well, the AI will know about it... and... presumably... do something about it - which you wont notice because you're always happy to go along with the one exact way the game wants you to play.

Most often, I play at Regent difficulty, a continents map (about 2/3 of the time) or Pangea map (the other 1/3). Standard size, I vary the age and amount of water. Lately those values I set to Random, and I usually do not pick my opponents. I can provide screenshots if that helps the discussion.

After re-reading your reply, I feel that your characterization of my playstyle is influenced by your view of your own. I believe that I have learned from the folks here, and that I am playing the game the way expert humans recommend that I play it. You describe that as "the way the mechanics have encouraged you to play it" and those are not the words I would use. True, I am not coming to the game and saying, "I'm going to play Always War today" or "I'm going to try a one-city challenge today" or "I'm going to raze and replace every AI city today."
If I were to try to play "Always Peace" or a "Five City Challenge", I would expect to fail the first dozen times. I would expect the AI logic to be the same; I would expect AI tribes to expand relentlessly as they do now. I would not expect the AI to suddenly realize, "the human is doing something different, engage the other logic to thwart it."

Finally, I will grant that -- using the metric of "attack power per shield invested" -- warriors are economical. Justanick and Lanzelot make excellent points. I will say that in the strategy articles I read here, I found very few examples of players doing warrior rushes on a regular basis. I found more examples of sword rushes, and archer-based attacks to go get iron from an opponent. If I've missed the stories, or if the current recommended playstyle is to conduct an early war with a horde of warriors, then I will try that.

 

[Buttercup]

While I was away, the thread moved into a discussion of "coincidences" that you (and others) attributed to the AI predicting what the human player will do, and consciously opposing that. An example was given of a "perfectly timed settler clash." I do not have corroborating experience for that phenomenon. I don't find it very surprising that the Hittite settler was on its way to settle a city about 5 tiles from its last city. That's pretty typical loose city spacing, where the AI assumes every city will grow above size 12.
I don't want to weigh into that discussion. I have not observed as many "weird things happening" as you have.

Part of the reason I don't subscribe to the "AI is changing its behavior just to thwart me" theory is that my day job deals with computers and computer programming. The complexity of AI programming, especially given the state of development tools in the early 2000s, means that the AI in Civ3 is not very complex. The target PC where it would run, back in the day, was pretty underpowered. Its algorithm for deciding what to build, what to move, and when to plot war is simple and predictable. Indeed, on this site a number of contributors have explored the tendencies and flavors of the table-driven code that the AI uses. Part of the reason I know what I know is because I rely on knowledge shared here, not confirming my own particular biases.
The developers knew this; they only way that they could make the computer opponents tougher at higher difficulty levels is to change the values of production bonuses, research costs, and starting units. It has been known for decades that the AI does not have a fog of war and that the AI gets happiness bonuses similar to the human at "Settler" difficulty level. This is not an ideal game. It is a game of its time. It's not even the best Civ game in the franchise. But on to my other points ...

Lol, the AI forward settling is a commonly explored criticism that predates Civ3 and was a huge topic in Civ2. Lmao @ this reply.

Anyway, you seem to be writing a lot just to say "if buttercup says it, I will be contrarian, but if someone else says it I'll take note", which, aside from being hilariously insulting, suggests that I'm wasting my time replying to you. I honestly can't believe you've never competed with an AI for a settlement spot, even if there was no algorithm or other kind of programming for it, you'd still have games where you compete for land in similarly timed ways by genuine coincidence, and the statement by you that you've NEVER experienced a settler race is just more such obvious shade that I have no option but to assume you're being deliberately absurd.

And, yes, try a one city challenge, please do, you'll be amazed how the AI reacts differently to you...

 

[justanick]

the AI gets happiness bonuses similar to the human at "Settler" difficulty level.

AI plays at regent. It gets 2 citizens starting content just like a human player at regent (or monarch). Also AI gets the same bonus vs. barbarians as a human player at regent. AI gets the same corruption as as a human player on regent. AI gets the same hut results as a human player on regent.

 

[tjs282]

I honestly can't believe you've never competed with an AI for a settlement spot, even if there was no algorithm or other kind of programming for it, you'd still have games where you compete for land in similarly timed ways by genuine coincidence

I have (also) seen AI-Settlers (or Barb-camps!) show up at one of my planned town-sites in more than one of my games, so I have an honest question here, based on my habits.

Do you (also) generally send Settlers (or the boats carrying them) to prospective town-sites using the "GoTo" order?

Because we all know that the Civ3 AI-routines have access to all the information in the autosave file, and therefore knows the current locations and any preset destinations of all the units on the map. So (I would imagine) it should be relatively easy to program the game-AI to calculate the shortest path-length and hence likely turn-of-arrival of a human Settler at any given tile — and possibly then use that information to pip the player to the post, provided it has a Settler of its own available.

(And AFAIK, the AI continues to build Settlers as long as there is any 'useful' open land accessible to it, certainly while at peace — and frequently even while at war, if the Fall Of Rome scenario is anything to go by)

That could also go some way to explaining why habitual DG+ players — and aspiring players who base their gameplay on expert advice (specifically, "Never automate anything!") — might genuinely see different AI-responses to their Settler-expeditions, compared to more casual Regent–Emp level players like myself (and also yourself?).

(Despite whatever opinion you might have formed about me, I do still consider myself a fairly casual player — and will continue to do so until I can reliably beat the game at Standard all-Random DG at the very least, i.e. probably not before I retire, and can play all day every day!)

It has been known for decades that the AI does not have a fog of war and that the AI gets happiness bonuses similar to the human at "Settler" difficulty level.

AI plays at regent.

I'm pretty sure @vorlon_mi knows that, since he refers to "Regent" level in his very next paragraph.

He just mis-typed in the quoted sentence ("Settler" is the Regent-equivalent in Civ IV, IIRC?), so his point stands.

 

[vorlon_mi]

Anyway, you seem to be writing a lot just to say "if buttercup says it, I will be contrarian, but if someone else says it I'll take note", which, aside from being hilariously insulting, suggests that I'm wasting my time replying to you. I honestly can't believe you've never competed with an AI for a settlement spot, even if there was no algorithm or other kind of programming for it, you'd still have games where you compete for land in similarly timed ways by genuine coincidence, and the statement by you that you've NEVER experienced a settler race is just more such obvious shade that I have no option but to assume you're being deliberately absurd.

You're putting words in my mouth (or keyboard).

I've been reading these boards for more than 10 years, specifically the Civ3 boards. I followed Vanadorn's "Pax Romana" novella, as well as studied Cracker's Opening Moves, multiple articles in the "Strategy Articles" subforum, and vmxa's frequent postings in the late 2000's. I'm grateful for the continued, consistent presence of the people who respond in the "Civ3 Strategy" forums. If five people write something, if ten people write something, and others corroborate it, I pay attention. Buttercup, you're one of the few people who consistently advances the notion that the computer code in Civ3 contains logic specifically designed to anticipate the human player's move, and counter it. If there are others who consistently observe this algorithmic behavior, I haven't seen them.

I'm not trying to be insulting and I'm sorry if I've given offense. Of course I have had conditions where my settler-spearman pair is headed for a spot and I meet an AI settler-spearman pair. Not every game. Not consistently enough where I would consider that it is a nefarious pattern. The last time it happened, we were both racing to get a spot near some horses, IIRC. That makes sense from a desire to acquire resources. Occam's razor.

My experience (and the articles/advice given here) indicate that the AI tries to grab land and is willing to walk a settler/escort pair through my land to fill in a gap I may have left. My experience shows that some AI have a greater tendency to warmonger (Aztec, Zulu, Celts, Rome, others) and that they will plan to invade when they perceive weakness. They're using the same criteria to evaluate weakness as the F3 advisor does. The AI's threat algorithm seems to factor in distance / close borders; note that we're not having this discussion about random declarations from across the land mass. Many AI have made demands of me when my armies are weak. The "farmer's gambit" is a risk whose size depends on the aggressiveness of neighbors.

 

[justanick]

"Settler" is the Regent-equivalent in Civ IV, IIRC?

There are no real aquivalents, but Settler is the lowest setting in Civ4, and thus comparable to Civ3-Chieftain, which is 2 settings below Regent and gets 2 more citizens starting content. Hence my attempt of a clarification.

 

[Buttercup]

Just mucking about this afternoon, trying to kill time, I spawned on a random dead island & just fancied "playing like an AI", this info is irrelevant to the post, it just stops questions like "why are you still in a despotism in 500AD?" etc. I was just mucking about.

But then this happens, so I screenshot it for you:



The Galley containing my Settler is to the left of the screen there. the Galley to the bottom right is a Korean Galley which has just this second hurled itself into view.

Hmm... I wonder, is this an exploration Galley, an attack Galley... or is it a Settler Galley making a sharp b-line for the one little spot left open to it for Settling in this entire mini island complex?

Well, my gamble is that it's a Settler Galley.

But, Hah! Crafty ol' Buttercup was, of course, expecting this...

... so crafty ol' Buttercup already has it sorted if it's a Settler Galley:



My two horses and two workers are all ready and prepared to take up all the space. HAHA.

Now, if it was an exploration Galley, it'll just continue moseying around.

If it's an attack Galley, it'll just continue on towards whichever town it's earmarked for invasion.

However, if it's a Settler galley it'll just:



Go away again...

COINCIDENCE!

 

[Buttercup]

I have (also) seen AI-Settlers (or Barb-camps!) show up at one of my planned town-sites in more than one of my games, so I have an honest question here, based on my habits.

Do you (also) generally send Settlers (or the boats carrying them) to prospective town-sites using the "GoTo" order?

Because we all know that the Civ3 AI-routines have access to all the information in the autosave file, and therefore knows the current locations and any preset destinations of all the units on the map. So (I would imagine) it should be relatively easy to program the game-AI to calculate the shortest path-length and hence likely turn-of-arrival of a human Settler at any given tile — and possibly then use that information to pip the player to the post, provided it has a Settler of its own available.

(And AFAIK, the AI continues to build Settlers as long as there is any 'useful' open land accessible to it, certainly while at peace — and frequently even while at war, if the Fall Of Rome scenario is anything to go by)

That could also go some way to explaining why habitual DG+ players — and aspiring players who base their gameplay on expert advice (specifically, "Never automate anything!") — might genuinely see different AI-responses to their Settler-expeditions, compared to more casual Regent–Emp level players like myself (and also yourself?).

(Despite whatever opinion you might have formed about me, I do still consider myself a fairly casual player — and will continue to do so until I can reliably beat the game at Standard all-Random DG at the very least, i.e. probably not before I retire, and can play all day every day!)

Oh yes, the Barb camp spawning at the exact spot that you're sending a Settler to is also a recur'er, though not very often, just often enough for it to tweak the consciousness bar. It certainly doesn't happen often enough to be able to say for certain there's something in it, just enough to make one suspicious though.

You make a good point about the "go to" aspect, and, yes, while I too try not to automate anything, I'll often send Settlers out on long automated journeys if they're travelling through my own territory and then just manually move them once they are near the outside of my territory.

But the example above and my earlier example do not use any Go To instructions.

Regarding my screenshots, I could have posted screenies from every game I've played or started since starting this convo (I'm a serial restarter and big fan of the early game generally), but I've only posted the ones that have tickled my funny bone.

 

[Lanzelot]

We all know (and agree on) that the AI is programmed to go for all unsettled places (especially, when there is a resource there) and to produce settlers as long as there are still unsettled spots. So could the incidents that you have observed many times (namely you sending a settler to some spot and seeing an AI settler arriving there around the same time), simply be a consequence of that algorithm? Or in other words, the AI is programmed to spam settlers and send them to still available places. It does so all the time and it does it whether or not you are also trying to settle that place around the same time. You (or the human player in general) are/is also trying to grab space and therefore it is only natural that on some of these occasions you two are going for the same place and the AI beats you to that place. But that doesn't mean the AI did so, because it can somehow predict your intentions and is trying to thwart them -- it did so, because it does what it is doing all the time: trying to settle open spots within reach!

The above would also be a plausible explanation for the behavior you have observed, and a much simpler one than the explanation of the AI having highly advanced human behavior analysis algorithms built-in so that it can predict our intentions and counter-act them just in time.
And by Occam's Razor, if there are multiple possible explanations for something, we should prefer the simpler one (meaning the one that requires less assumptions). Unless proven otherwise, of course. But the only 100% proof in this case here would be to look at the source code and see, whether these algorithms are really there or not. And as long as that did not happen, you have to accept that others have a different opinion on that point than you.

Anyway, you seem to be writing a lot just to say "if buttercup says it, I will be contrarian, but if someone else says it I'll take note", which, aside from being hilariously insulting, suggests that I'm wasting my time replying to you.

Please, Buttercup, don't feel personally attacked, if someone does not share your opinion or tries to advance counter-arguments against some of your assumptions. As far as I can tell, vorlon-me just stated very reasonable counter-arguments against your theory in a well-mannered and objective way. He was in no way trying to attack you nor being "hilariously insulting". And I'm sure he posted these counter-arguments, simply because he does not believe in that theory, and not because "it was Buttercup who proposed it".

Same for me: as you can tell, I'm not convinced of your theory of these "built-in human behavior prediction algorithms" -- but not because it's your theory, but simply because I believe that other possible explanations are more likely to be closer to the truth. And I would have stated my arguments against this theory even if that theory would have been proposed by let's say SirPleb.

On the other hand, I agree with many of your other points. For example, I have never seen your idea of the warrior rush stated somewhere else, but after I have read your explanation and looked at the numbers, I do think it can be a very good idea in the right circumstances. It may not work at Sid, where the AI can produce swordsmen as fast as the human player can produce warriors, so any such attempt may end in a bloodbath without the human player being able to reinforce the losses, but it may work well on lower difficulty levels and especially during the early Despotism phase, when unit upkeep and war weariness are not a problem. Do some early damage to a close-by rival while he is still weak and at only a low cost to yourself! I may actually try that in one of my next games.

All this is quite natural when discussing points that are not easily to decide one way or the other: some may agree with you and others may not.

 

[CKS]

A lot of the HOF fastest finish conquests are done with jaguar warriors rather than more powerful units, even on relatively high levels or larger maps. Boogaboo has a number of threads about his quests for ever faster conquests, and they are quite instructive. He discusses pros and cons of various units, but it is in the context of reaching a conquest victory soonest. It isn't my kind of game at all, but spewing out hordes of cheap units apparently works well.

 

[justanick]

On the other hand, I agree with many of your other points. For example, I have never seen your idea of the warrior rush stated somewhere else, but after I have read your explanation and looked at the numbers, I do think it can be a very good idea in the right circumstances. It may not work at Sid, where the AI can produce swordsmen as fast as the human player can produce warriors, so any such attempt may end in a bloodbath without the human player being able to reinforce the losses, but it may work well on lower difficulty levels and especially during the early Despotism phase, when unit upkeep and war weariness are not a problem.

I suspect that a properly executed warrior rush works best on Sid. All free land will be taken long before you can leave despotism which may well force you to attempt a warriors rush to increase your amounts of towns beyond what mere ICS allows you within a very limited amount space. But of course you cannot wait till barracks pay off. You may however need embassies very much so that you are not facing a Sid AI alone.

 

[Lanzelot]

I suspect that a properly executed warrior rush works best on Sid.

But wouldn't an early attack on a nearby Sid AI be close to suicide, considering the big amount of extra starting units they have? And considering the speed they can produce new units? It takes a while to catch up with them porduction-wise and to prepare a big enough army to have a chance, and by that time the window of opportunity for a warrior rush may already be closed?!

 

[justanick]

But wouldn't an early attack on a nearby Sid AI be close to suicide, considering the big amount of extra starting units they have? And considering the speed they can produce new units?

That is why you need embassies and thus military alliances.

It takes a while to catch up with them production-wise and to prepare a big enough army to have a chance, and by that time the window of opportunity for a warrior rush may already be closed?!

Keep in mind that the opportunity for the warrior rush closes by becoming unable to pay the unit support for them. This does not happen very early if you are aiming for it. It depends on the playstile, but image it to fit my playstyle rather well.

The warrior rush implies to suffer high losses, which reduces unit support. One needs to find the right balance between too high unit support and too weak a military.

 

[Buttercup]

We all know (and agree on) that the AI is programmed to go for all unsettled places (especially, when there is a resource there) and to produce settlers as long as there are still unsettled spots. So could the incidents that you have observed many times (namely you sending a settler to some spot and seeing an AI settler arriving there around the same time), simply be a consequence of that algorithm? Or in other words, the AI is programmed to spam settlers and send them to still available places. It does so all the time and it does it whether or not you are also trying to settle that place around the same time. You (or the human player in general) are/is also trying to grab space and therefore it is only natural that on some of these occasions you two are going for the same place and the AI beats you to that place. But that doesn't mean the AI did so, because it can somehow predict your intentions and is trying to thwart them -- it did so, because it does what it is doing all the time: trying to settle open spots within reach!

The above would also be a plausible explanation for the behavior you have observed, and a much simpler one than the explanation of the AI having highly advanced human behavior analysis algorithms built-in so that it can predict our intentions and counter-act them just in time.
And by Occam's Razor, if there are multiple possible explanations for something, we should prefer the simpler one (meaning the one that requires less assumptions). Unless proven otherwise, of course. But the only 100% proof in this case here would be to look at the source code and see, whether these algorithms are really there or not. And as long as that did not happen, you have to accept that others have a different opinion on that point than you.

We're talking about a computer here, and, as you yourself point out, a computer doesn't tend to 'act randomly'. It functions by code.

Yes, the easy answer is that it's 'coincidence', however, by playing the game regularly, one notices that there's more to it. Factually, not by opinion. Can this still be 'coincidence'?

How can coincidence explain this:

From a previous screenshot I continued my game. I scuttled all the way to Steam Power before backtracking to Navigation. The first thing I see once I'd done just the most cursory exploration is this group of tasty looking islands just to my east and just to the Inca's north:



As you can see, a whole chain of lovely islands, all of which have resources on them, there's also horses, Iron the whole caboodle stewn around the long island complex.

You'll also notice that the Incas are well within easy reaching distance of them. At this point in time there's no logical reason why the Incas haven't settled them yet.

So I check to see whether the Incas have even learned Map Making yet or not:



As you can clearly see, they are well past Map making, by quite a huge margin. And their towns and land do not appear to preclude the building of Settlers nor ships.

If the AI ALWAYS 'simply fills the space', why hasn't it filled this space?

So I set about pumping Settlers from a couple of Settler farms and prepare to settle these lands, as they seem quite nice.

The exact moment I launch my ships full of Settlers, this happens:



As if by magic, the Incas are now suddenly stirred into activity.

You will also notice that the Incas are on a secluded island, and so they have not been too busy having wars, in fact I'm the only people they know exist (trading & etc-wise), and they have a full island that's not very big and have not been too busy Settling other lands.

Since we last spoke to them they have learned Feudalism, so their tech isn't especially shabby and I think it's safe to assume they got Map Making long before they passed into the Medieval Age.

I honestly can't believe this AI chose, uniquely, to leave Map Making until the very last tech it learned in the Ancient Age, I have never seen that before.



And, next turn, here's my first of two ships full of Settlers:



And you'll see the Incas are now, all of a sudden, furiously settling these lands that they haven't been bothered about for hundreds of years, maybe even a thousand years.

Now, as you say, a computer is just operating by code.

All the other AIs in this game have settled their reachable lands, some occupy many little islands.

It was just these guys who decided not to because...? Because because because...

Now, I think my theory works. Yes, my theory is still an opinion because we don't know why such things as the above happen, but my theory is the Razor theory here. You might well have an opinion that it's all coincidental and explainable by existing knowledge, however, it is your version of events which, the more one looks and takes note, appear to be more in the conspiracy side of any debates about Razors.

What kind of 'evidence' do you require to 'suspect' there's something 'code-based' going on if the above three examples are not, by themselves, completely definitive?

 

[Fergei]

The above isn't an adequate example without knowing if the Incans had populated all land on their own continent by the time you created your own settlers. I agree that while the AI is coded to fill up space sometimes it can be diverted from this and leave easily reachable islands untouched (and not just because they are at war).

Having played dozens of hours of AI only games with the whole map on display (to play test difficulty settings) I can say I see no evidence whatsoever that the AI conducts itself differently depending on whether or not there is a a human player.

I take all the above as a positive. As is the level of conviction some players have that the AI does one thing while another thinks the opposite. This is the sign of a good game.

On the topic of whether balancing the game is good or bad, my contribution is that a Civ game where you can genuinely keep an open mind about your tactics and strategy depending on land, AI opponents, access to resources/luxuries, global diplomatic developments etc is a far better experience than a game where there are clear power spikes and optimal strategies irrespective of the aforementioned variables. I'd score Civ3 extremely highly on how it makes this possible. Every game feels genuinely unique.

I'm currently replaying a game (2x human, 6x AI) on the same map as my last game with the same starting locations. Purely due to the different AI Civs composition the game is much more culture and tech focussed rather than the bloodbath of the previous game. Even something as simple as what islands and continents the game focusses the barbarian spawns (which differs from game to game) has utterly changed the decision making and outcomes of both human players and the AI.