City proliferation, why settlers being cheap is bad

[Kaiku]

In every game I've played so far, the AI has been expanding at an incredible pace, literally littering the map with settlements. A swamp surrounded by tundra? Well, doesn't that look like a lovely place for a town! That spot surrounded by 4 capital cities of your worst enemy? Why not!

In nearly every strategy game, (early) expansion is a very risky strategy with a high reward, if it pays off. You choose to invest in the future, at the cost of having military units now. So anyone who doesn't choose to expand will be, for a time, stronger than you. But if left alone for long enough, your strength doubles - perhaps more, since you're cherry picking the optimal locations, which are often better than your starting city.

But expanding in MoM doesn't carry risk. The cost of a settler is negligible (pretty much just 1 turn), yet the reward stays the same. It's no wonder the AI covers the map in towns, as to them, there's only advantage to gain. Who cares if they get razed? It'll just build a dozen more instead.

And yet it is also flawed, as it causes the AI to spread itself to thin. It needs to defend all those settlements, but it has not enough decent units to do so, so their 'winning' strategy puts them at a disadvantage. Any proper doomstack can just hop from one town to the next without being slowed down (if you even want to bother with that).

But it's not really about AI having a dis(advantage), as we can raise cities just as fast, so it's still 'balanced.' It just doesn't make much sense in a strategy game, and in my very personal opinion, also isn't fun. Rather annoying, actually. I'm not interested in making that granary in my 20th city (even if it's worth it), or razing the AI's 20th outpost.

You want expansion to matter, and the enemy's cities to matter and be well defended. You want making that settler for that prime location to be an important decision - which can get you killed. An expansionist AI would also be noteworthy. It'll be vulnerable at the start, but if left alone for too long, it might be the most dangerous trait there is.

Settlers are simply too cheap. Don't know by how much - I'm not going to throw out a number that'll get this thread instantly dismissed. But I'd love to have a discussion about it and hear counter arguments, as I do believe this to be an unpopular opinion.

[Nelphine]

One of the big things is amplifying towers. They increase your casting skill by 7 (NOT your skil l production per turn, your actual skill).

Number of cities directly translate to power for the wizard. 

Not saying I'm against your thoughts (personally, I think cities are the least important part of your city, after wizard power, nodes, towers, doomstacks, overall military) but its something to be aware of.

[Seravy]

I'd argue settlers are not cheap at all. 150 production and a whole unit of population lost is a lot at the beginning of the game, as much as one or two buildings, or 3-5 units. But more importantly, you need to wait at least 50 turns before it turns profitable, usually more, and on top of that, as you say, you even have to defend it. If anything, building units and attacking people is still a way better strategy than building settlers, and that is bad, the two should be equally valid on average (obviously different maps and player strategies should favor one or the other more). In my previous test game I was able to fill about 50-55% of the world myself, leaving the other two Myrran wizards with 25% each - and I still lost because the cities haven't turned profitable fast enough.

However, before anything else, I'm awaiting your suggestion on what rule the AI should use on deciding when to produce a settler under whichever new settler cost you want. Without that, any sort of change to how the game should be played is not possible - building cities is too much of a core game mechanic to allow the AI to be bad at it.

The current rule is (as part of the "AI select production" procedure) :
1. Population 2 or less? Don't build a settler, otherwise continue
2. City race is best available for settler production on the plane? If not, don't build one, otherwise continue
3. Calculate wanted settler amounts.
Settler needed = 2, but 1 if land size = Tiny, but 4 if Myrran AI being alone on plane on Large or Huge land and it's before turn 100.
4. If settlers owned or being produced on continent<Settler needed, and
  settlers owned or being produced on plane<Settler needed+2, produce settler.
5. Proceed with picking nonsettler production, such as units or buildings.

Also note the AI will not have better garrisons because they have fewer cities, quite the contrary. Fewer cities allow less research and power, so they can have fewer and worse fantastic creature, while even the best normal units are not a threat to a good doomstack. So an AI that does not have lots of cities loses more easily, as they have less time to live between the human starting a war against them and destroying their last city.

[Kaiku]

I'm very wary of suggesting concrete numbers or changes to AI logic, as I just don't (didn't) know what's going on under the hood of this game. All I can base it on is what I've experienced in the games I've played. I've noticed what I view as a... problem, in that the AI always tries to cover the entire map in settlements, disregarding the normal decision making process that goes into such a thing (location, etc). Since we can do this too, it's not so much a balance problem but something else. I'm firstly interested in a discussion and to hear the opinions and experiences of others, or to maybe have my mind changed about it. If it's indeed considered a "problem" by more players, I'm obviously not the best person to suggest the actual changes that would fix it, while keeping the AI formidable. That would be you, Seravy. I do of course have ideas, but I don't know what (side) effects they'll have in practice.

I do realize that simply making settlers more difficult to make without any changes to the AI logic would break them, as they would still try to pump out as many settles as now and become effectively locked up.

[Seravy]

Okay, I wasn't expecting specific numbers, more like logic with the numbers left blank.
"If something then do this otherwise that" sort of stuff.
But we don't even need to go that far, let's stay at the basics.

The core design concept behind the AI is "numbers". The AI stands a chance because it has lots of everything. They can't stop an invading force that consists of a decent doomstack or two, so the plan is to let the AI attack the player so many times in retaliation that they fail to defend and lose cities at a faster rate than how the AI is getting beaten by the doomstack. Either because the player runs out of mana, or because they run out of defending troops, or production capacity (through losing population and buildings during battles at their cities).
If the AI has 10 cities and the human can kill one per turn, the AI has 10 turns to break the player's empire, otherwise the player wins. If the AI has 25 cities, they not only have 25 turns to do it, but have 2.5 times more troops they can throw at the human's cities, as more AI cities produced more troops. If the former case the AI's troops probably won't even make it to the human's continent or plane in time. In the latter, they can.

Obviously, this is only important for the late game, the last few AI the player faces. However, no one likes to see "rush AI" that eliminates the human player consistently on turn 50 intentionally. That's not fun so the AI must be designed with the long term game in mind (which does not mean they aren't allowed to have early units or attack people sometimes, but it shouldn't be the only thing they can do well, and in fact it is the thing they shouldn't be able to do TOO well.)

You are suggesting a gameplay where this core AI design is not possible, so you need to suggest a new core AI design philosophy to replace it otherwise it's meaningless to discuss whether we like one or the other better as human players. Even if everyone agrees we want expensive settlers, if the AI can't play with that, we can't do it and just wasted time. I believe there is no other possible AI we can do, which is why I'm prioritizing the AI in the discussion.

disregarding the normal decision making process

No, this is wrong. The AI has no brain. It's not "normal" for them. In fact they cannot comprehend it at all. They need to decide on building or not building a settler without knowing in advance where that settler will be used.
Sure, there is a "move settler" procedure that finds where they need to go to build, but first of all it needs all the settlers involved to already exist, and second, the settler will still be sent blindly towards a ship or tower if the current continent is full with cities (which might happen to the very first settler on island starts). And third, the procedure only cares about buffs already on the settler so they AI can't say "yeah I plan to put waterwalking on that so also consider that". Finally "there is no way I'm going to be the first to reach that spot" and "I'm planning to raze that so I need to prepare a settler for that spot" are also things the AI doesn't know but humans do.

[Kaiku]

Okay. So aside from, what I assume can be done, blatant cheats to AI economy (which are unfair and you obviously want to avoid as much as possible), the only way for an AI to stay competitive to a human is by doing that mass settler thing and racking up numbers. Makes sense, I guess, since they're a lot better at it than us. Or at least won't suffer from a lack of patience. That's why I didn't want to vomit out numbers like "5x as expensive!" as I already feared unintentional consequences, and now realize such a change would obviously wreck the AI.

That is unfortunate. But if there are no ways around it, I guess the discussion ends here, because I do very much want the AI to stay challenging. I trust you on what the possibilities are for an AI.

[Nelphine]

As a note, its still a good idea to have discussions and thoughts like this. Seravy has done an amazing job, but occassionally he does do changes - he's awesome in being open to these discussions.

So if you do find it a problem, don't hesitate, don't hold back - bit do cone prepared to here about why it might not work.

In this case, I'm still on your side - but for me the biggest issue is that the best tactic for the human is to wait for the AI to make hamlets, and then take them away - because you get to use the AI resource advantages.

However, we've made enough changes that risking those wars is actually a big risk so there are many situations you need to make your own cities, but that doesn't mean its the best strategy.

[Kaiku]

I think that when it comes down to it, other than cheap expansions feeling a bit wrong as a strategy veteran (I can't help but compare it to expansions in games like Starcraft and such), I'm mostly annoyed by the AI interfering with your own expansion plans - hogging resources, building too close and giving you 1/2 of tiles, and planting them at locations that don't make sense - so taking them over doesn't makes much sense either. It forces you to alter your plans (gotta get a settler there before he does!) just 'cause the AI's dumb. It's... I believe it's called 'artificial difficulty.'

But sadly that can't be helped.

Anyway, yeah. I do try coming up with feedback, but I also have this feeling that everything I'm thinking has been thought of a hundred times before already. Many other things that I consider problems aren't really that, I just haven't figured out the solution yet.

[Arnuz]

Ideally, we should have a way to spot areas further away from the others and try to send colonists there. If there are no such areas then the number of colonists should be 0. Doable? Probably not ;( but it would really make a lot of sense.
At least though, it's always sad to see AI cities drop in size when they're already at war... Would it be possible to make desired colonists=0 while at war? Would it be desirable? I think so. All resources should be kept for the war effort, it can't be without consequences.

As a result, given that AIs would risk less in colonists, we could remove the outpost bonus which doesn't make much sense and only makes it even more worthwhile to get AIs to colonise for you.

However, we've made enough changes that risking those wars is actually a big risk so there are many situations you need to make your own cities, but that doesn't mean its the best strategy.

Is that true? Currently AIs (at a high enough difficulty level to have the initial advantage, which is not much at all considering the free swordsmen) make a check and see if they're acting opportunistically aggressive without wardec. This, outside of the lucky combination of "peaceful" or "lawful" neighbours.

What happens when they do so is that if you're on the best spots on the map (close to the good places to colonise, close to the low strength lairs/nodes that they don't need to scout) you end up close to AI stacks. Given that they have the strategic advantage to begin with, you're a target for opportunistic behaviour.

That happens IIRC after turn 40, year 3.5, which gives you just barely enough to prepare 2-3 cities?

So, what we've actually done is to make it so that early aggression by the human pays off more because you know that you're going to be attacked anyway, then at least take the initiative and decide when and where rather than waiting for a doomstack to wander close to a city. If you're aggressing you can even intercept a doomstack, then the first AI you encounter is defanged and easier to deal with. Furthermore, you get to spot all the AI outposts and move there some troops for conquering the turn they become hamlets, and notice if the AIs are summoning stuff there...

[Nelphine]

It is true - lawful, peaceful ais are at least 1/3 of your opponents. Chaotic ais who have just allied with you exist. You could have a military strength high (or low) enough to avoid wars. 
The AI might be an awful race, and you need more cities of your own race to face the 2nd or 3rd or 4th AI.
The AI might gave just declared peace with you, giving you a few turns without hostilit0

I'm not saying these are the common state of affairs - but these do happen often enough that there are times you don't want to declare war. The best is still to conquer. Its just not the only thing.