Game mechanics

[Seravy]

They killed one sprite. I got 3 orc units back to replace it.

Orc suck against Death magic, they have no resistance. Try against nomads :D

. If you find settlers and city micromanagement boring, perhaps you should

Turn on the now way superior Grand Vizier.

I've absolutely had the ai attack me in 10-15 different locations in the same turn, it took me about 20 minutes between turns.

I've seen 24 attacks in a turn once after war broke out. This is a new thing though, since 2.51. Previous version AI's failed on recognize their own armies on large maps, let alone use them.

So all that military resource could be spent on more magic, so the AI can get very Rares faster.

The design concept for overland AI is that they act based on personality and random chance, and have resource boosts to make up for this. This is to keep games more varied. If the AI always took the best choice, you'd see much fewer types of units and spells. Though honestly that would be impossible, AI with human level intelligence has not been invented yet.
I've seen plenty of very rares from the AI though even with the current settings, and intentionally kept their research advantage low. If they get them too early, games becomes unwinnable, literally.

[Nelphine]

I definitely don't find city building boring. I've played entire games on strategic combat (even before my investigating here).

My problem is that building cities is not an advantage for you, when you could build more troops/trade goods/housing, and then just conquer an AI city using the extra troops/mana. aside from the early game where you need to get enough cities to keep up with your first opponent (which is only what, 3-6 cities?) until you can waltz over him, settling is always worse than conquering.

[Nelphine]

At the nomad comment: considering I'm using 2 death knights to take out packs of 3-4 great wyrms with full support, inside nature nodes (mostly via throwing spells at them despite their resistance), nomads don't particularly scare me. Earlier in the game, sure.

My point is at this stage, I could use death magic to eat paladins alive (ok probably not, but I would at least fight them to a draw if I didn't have any death knights around), let alone weak things like elven lords or griffins. But my orc opponent thinks she has this massive army even though city units just don't matter anymore (and realistically, haven't mattered since she broke through to my plane.) She went to war based on city units, and stays at war because of them, and keeps building huge amounts of them.

[namad]

Is there really a battle going on between settlers and conquering? What do you do when you win a war against an ai, capture his cities, and then find out there were a half dozen city spots left unsettled? Just ignore them? If you manage to fight less than 3 or 4 wars at a time you'll find opportunities for settlers. If normal units are so worthless why does it matter if cities waste their time building settlers? Might as well cover every last inch of your lands, it improves your wizard's pact, and even generates power in the long run, especially once those little 8 population cities eventually get their amplifying towers.

Also how does the AI convert halberdiers into very rare creatures? They've got a 230-250% production bonus (but only 120-130% to research), they're able to build both the halberdiers AND the research buildings. If you're so sure the orc players is worthless compared to you, why not conquer them? Are you in another war with a stronger opponent and they're a minor nuisance? If so that's probably a smart move on their part. If you're not at war with anyone else, and they're way weaker than you, why not just go banish them?

Stuff like paladins and warlocks and steam cannons can in large enough quantities go toe to toe with powerful fantastic creatures, which the ai in theory knows, and takes into account when choosing which races of settlers to focus on.


IIRC a couple patches ago seravy specifically added code that made it so that ai wizards could comprehend how a dogpile worked and avoid starting wars based on that. So perhaps the orc is really simply inspired to join war against you, despite their weak position, because someone else more powerful is also at war with you? Just guessing, I may have misunderstand the formulas seravy was talking about for that a month or so ago.

Flat out though, a human player gets better usage out of spell skill than the ai. If a human being has equal spell skill to an AI they'll be at a big advantage. That's why the ai cheats and has extra power and skill. They also have double the power so they can afford to use their full skill more often. The human can exploit this though for example by fighting some unimportant battles first until the AI is bled dry, then hitting their capital with the player being the only one with skill left. Around the time I get to first place on the skill power bar is around the time I think to myself the game is now almost assuredly a victory as long as they don't all declare war on me at the same exact time.

[Nelphine]

In this case no, the only AI alive is the one I'm at war with. I was stronger in spells when we first met; despite being expansionist, I had more cities, and even after a few years (I'm no longer settling as 30 cities is about the max I like to keep track of) they still haven't caught up in number of cities.

They declared war shortly after meeting, which presumably was influenced by the fact that (due to prodigious numbers of halberdiers and hordes on their home continent) their army strength was more than 5 times my own. Since the only troops I've gained since then have been from life steal in battles, and 2 death knights (and probably a dozen skeletons), their army strength has stayed vastly ahead of mine, which I again presume is making them think there is no reason to sue for peace.

[Nelphine]

(Why don't I just go kill them? Two reasons: My death knights are out claiming nodes/lairs (there's still tons that are not conquered) and I immensely enjoy seeing how many books/retorts I can get. And most of my demon lords are tied up making sure he doesn't get through towers (although he's not trying crazy hard so I could probably use some of them) so I'm not sure I have enough to beat the heroes and stone giants in his capital (units which can't get instant killed by death magic). If I consolidated all my very Rares, yeah I could do it, but I like nodes/lairs more.

[Nelphine]

How does the AI convert military into magic? Well they don't convert the units into magic, they convert the resources into magIc. Make 40 settlers and colonize the rest of the plane since I'm not. Unless they already have 0 power spent on mana, set most cities to trade goods, convert to mana via alchemy, and put all power towards skill (or research for more very Rares. ) use more trade goods to buy literally ever power building possible, in every city.

However swarm tactics of dozens of stacks, which I haven't seen, would be viable too, so they may not need to stop military in too many cities.

[Seravy]

So perhaps the orc is really simply inspired to join war against you, despite their weak position, because someone else more powerful is also at war with you?

No, that patch prevents wars that would put a wizard into a situation that is too much for them. It's not used to start wars against people already in trouble...alliances between AI does that already though not based on military power.

Make 40 settlers and colonize the rest of the plane since I'm not. Unless they already have 0 power spent on mana, set most cities to trade goods, convert to mana via alchemy, and put all power towards skill (or research for more very Rares. ) use more trade goods to buy literally ever power building possible, in every city.

This is something a human player will do. An AI is a machine that works on set of simple rules that determine what they should do.

Could you list a set of rules that are foolproof (in all situations the performance increases or at least doesn't change) and achieve the tactic you are describing? Like, when to increase trade goods production, when to rush-buy a building, when to push the skill bar higher and by how much, since those three seem to be the objectives here? Simple enough rules that are doable in about 50 bytes of x86 assembly code each. Oh and they have to not contradict any already existing such rules or cause unexpected bad interaction with something else (for example loss of resources if another rule makes them convert mana to gold, or push up the mana bar at the same time, resulting it going over 30k mana stored.)

[Nelphine]

I was thinking about that. four things come to mind -

One:
if the AI has significantly stronger army (2-3, compared to human player), but equal or lower power production, he should be setting more trade goods and changing gold to mana (IF his power maintenance + X times his casting skill is higher than his mana income, where X is some reasonable number of combats like ten, x max cost of spells in combat due to range factor - so 3, or 1 if a channeler, for a total of 30 Or 10)

Example: in my game his army strength is super high, but my power production is ahead; his casting skill is around 150. So he wants mana income equal to his power maintenance plus 4500, so he can cast overland, pay maintenance, and fight 9 full price battles per turn.

Add an extra here to not cap mana - probably to make this rule aim for mana pool, not mana income. So maybe double x, so for non channeler, it would be 60, and that's the mana pool the AI wants, and would use trade goods to achieve, in combination with 2 below.

Two: if trade goods are available for mana income, and spell power is lower than human player, all power that would normally go to mana, should instead go to casting skill. Actually, this should be always, but if spell power us lower than human, trade goods would get a big priority boost.

Three:
Settlers should increase as time goes on, if there are empty city spots. Regardless of available spots, the current number of settlers is OK - 2 per continent, so AI can grab newly freed places. But if there are available spots, AI should increase faster the bigger he gets. Thus my proposal to limit settlers to 1/3 cities. AI would be slightly slower at the start, needing 4 cities before they could have 2 settlers, but once they get big (ex: my game they have 24 cities) they could have 8 settlers all going at once.

4. If you want, you could add power generation, and research, to city units (similar to heroes). For instance magicians could grant 1 research, priests 1 power. That way city units stay meaningful if the AI is churning them out in great numbers, even if the human can destroy those units at will. Not sure how viable this actually is. Would also help combat the economic superiority of races like dwarves (bonuses) and lizard men (high pop). Would really hurt other races like barbarians though. Might be OK.

[namad]

Actually I just realized something. If there is one ai wizard left in the game. They should instantly declare war on the player. There's no reason they shouldn't. In every FFA game with only 1 winner the correct choice when there are only two players left is for them to declare war on each other. A human being can realize this is true, and start building up armies on the border. The ai cannot use this sort of logic, so they should just instantly break their treaties and declare war, since that's the only way to force the ai to try and win. Apologies if this is already the case. By the time I've slain two wizards and am about the slay the third wizard the fourth is usually already at war with me.

Also, having 5 times your military can make them less likely to attack than having equal military, depending on which personality they are, IIRC.

As a kid I loved to conquer every node before I went on to win the game, I certainly wouldn't suggest seravy balance the game around that though.
If they really have left entire continent unconquered, and they've done it without you blockading wizard's towers.... then the new settler pathing ai is not working as intended? That might be the entire problem? Was this game on an older patch? Or were you actively killing/blockading their settlers? I've sometimes seen the ai get stuck in this manner. The thing is settlers aren't that expensive. Sometimes I've even seen 3or4 of their settlers standing around doing nothing (on older patches) seravy is actively trying to fix this. Also what year is it? I'd say by roughly 1415 everywhere on both worlds should be almost entirely covered in cities except maybe ones that share tiles and have maxpop 5-8 ish. All the good resources should probably be settled by 1410-1412 depending on if it's hard or extreme or impossible?