Crushing the cheating AI

Introduction[]

This document is a tutorial outlining some of my key tactics, with which I can consistently beat the cheating AI in a 5-player free-for-all, v2.5.10 standard settings. I assume the reader understands game mechanics, and I do not assume this will be used against humans: certain ideas are exploitative of AI weaknesses but could be in turn exploited by a clever human.

The first turn[]

Set the current research to pottery and the target to map making. In general, you want your early cities to be built ASAP on tiles such that they produce at least 4 food (2 surplus), 2 shields, and 1 trade, for a total of 5 surplus resources; this usually means something like built on a plains with a whale within its environs. Anything better (such as pheasants) is welcome. Anything less is bad, and you should prioritise other sites first. As a hard rule, every city should get at least 2 surplus food and 2 shields, possibly after some terrain improvements; and every tile gathered from should get at least 3 resources. Irrigated roaded plains are a good standby if you can't get suitable special tiles.

Don't take more than a turn or two to found your first city. As a rule of thumb, your civilisation should grow by about 5% per turn, unless you have to devote serious resources to defence; so any sort of investment should pay a 5% or better RoI. If your initial cities both produce 2 food, 2 shields, and 1 science, that's 10 resources per turn; so delaying this by more than 2 turns to put the capital on wheat rather than bare plains isn't worth it. Also bear in mind that your third city could take that wheat anyway, and with the despotism capital city production bonus, your capital will actually produce 3 shields; so even delaying by 2 turns isn't worth it.

In the early game, build your cities close to each other, viz 2-3 tiles apart. This looks ugly as sin, and there'll be some contention for resources, but it's worth it. 1) If a city could produce 5 resources per turn, each turn spent walking has an opportunity cost of 5 resources, 6 if a city has to pay food for the settler's upkeep. 2) Under despotism, corruption increases rapidly with distance from the capital, so further cities will give you less science or gold. 3) It's convenient to have cities around your capital. An enemy attacks? You can get an entire army on short notice to protect your capital. You want to build a wonder? With 5 2-pop cities, each collecting from a mined hill, you can get 25 or so shields worth of caravans per turn. 4) You begin with two workers. This way, they can both work full time, transforming forests to plains and irrigating/roading, and the improvements will all be used right away. They should build roads even on hills; it's slower and doesn't give trade, but the mobility this affords is worth it.

Point 1) is very important, even later into the game. For this reason, once a city is surrounded by other cities, it shouldn't make any more settlers, and should instead focus on units: workers, triremes, or soldiers, depending.

I like to site my cities so that some are in good long-term positions (viz non-overlapping or minimally overlapping environs, lots of special tiles, ideally coastal) and the rest are spaced between, which I call towns. Don't invest in towns: no improvements, no trade routes, certainly no wonders. They won't ever get very large, they're strictly to gather resources fast. Later on, you can get rid of them. I like to mark a town with a single letter after its name, eg "Roma a", so that I don't forget and waste resources on city improvements. They usually make straight economy units, settlers-workers-triremes-caravans; better to leave military for cities with barracks.

The reason you want pottery and map making is as follows. In general, you have 3 strategic priorities: attack, defence, and economy. (Science is part of economy.) You don't see any enemies yet, so there's no-one to attack or defend against, so you should focus on economy. Even if you do see enemies, you don't have the tools to do much about them yet, so ignore them for now, other than to defend against any attacks.

You want workers ASAP because road and irrigation are invaluable. In particular, often a city site is perfect except that it's on a forest; so build it anyway, then make workers to transform to plains. Roads are also useful for attack and defence, by helping units move around faster, in addition to the very useful trade bonus.

Set your capital to make one militia, then settlers, getting workers as available and as required. All other cities should get settlers. You'll grow very quickly. Your explorer should do as its name suggests, taking any villages possible, as should the militia, who's obviously much worse at it.

Map making is a critically important economic tech, albeit a little less urgent than pottery. The opportunity cost of a settler-turn is 6ish resources, so to walk one 4 tiles away (the minimum to not have overlapping environs) costs 24 res. If you get a trireme, you can move the settlers 8 tiles in 2 turns, and lose only 12 res. Even if you stick to your starting island, it's worth it just so you can spread your cities out faster. In general, it's better to take new islands if possible, because for your trouble this typically gets you more access to special tiles, and there's a trade bonus once you get caravans. Even settling the poles is worth it if it's fast and you can at least get some whales: the terrain is bad, but harbours will mitigate this, and a polar city is a fine site to send caravans.

Enemy contact[]

Around now, you'll probably meet another civ. Agree to the truce, find its cities, and march your militia toward it, along with any legions you may have found who can reach it in a timely manner. Don't make any extra troops yet.

There's a few ideas here. 1) Sometimes, the AI leaves cities empty, and you can just walk in, which is a crippling blow to them and a massive boost to you. 2) If they aren't empty, you can still poke around and force the AI to invest in defence, which means it can't expand as fast. 3) If they send troops to you too, you'll meet in the middle, and you might be able to block theirs. Even if your militia dies, you'll at least be forewarned. 4) The AI will almost certainly declare war on you eventually. Try to ensure that the inevitable war occurs on terms favourable to you, ie stealing a city or two. 5) You probably won't be attacked yet, and committing extra troops probably won't achieve much, so no need for more units yet.

After map making, writing is an excellent tech. Send diplomats to establish embassies, and once you do, trade techs. In particular, the AI likes to get early masonry instead of pottery and maps (this is a terrible build, do not emulate it), so you should be able to get something. If you get embassies with multiple civs, you can trade your tech for different techs from each, or get tech 1 from civ 1 and then trade that to civ 2 for tech 2. Don't worry about helping the enemy: if you trade each enemy 1 tech, you'll get 4 techs and have a massive advantage. Don't worry too much about what you're getting; everything is useful eventually.

You may or may not want bronze working or warrior code, for the option of stronger fighting units. They're not usually necessary yet, but it's only 28 bulbs. Either way, set tech goal for either monarchy or republic, and switch as soon as possible. I prefer republic -- the trade bonus is massive, to the extent that it's still a good wartime government because you can buy extra units -- but you can get monarchy earlier and it's almost as good, so it's not bad to get at least temporarily.

When I start making fighting units, now or later, I like to build a barracks in one city with high (>= 5) shields per turn, and designate it a factory city that exclusively pumps them out. The units then spread out across your civ and re-home at other cities. This way, you can have veteran units everywhere for the price of a single barracks. If you have no high-shield city, don't bother with this, it isn't worth the price of the barracks. I should mention I'm a fan of early archers: they're only a little more expensive than phalanxes, but they can chase off enemy catapults, which are otherwise a pain point.

When I began playing, I followed Sid Meier's advice and defended each city with 2 veteran units. This is a terrible idea. Get one or two veteran units for each cluster of cities with adjacent environs, and move them around to defend any threat as it arises; any more is a waste of resources that could be extra economy. This is another reason to make sure you have lots of workers, so that you can build roads and redeploy military units as needed. Possibly keep a defender in your capital and any wonder cities, but otherwise, only leave units fortified in a city if it's to ward off a specific threat.

Mid game[]

I said earlier that there are three strategic options: attack, defence, economy. There's a loose paper-scissors-rock between them. Archers kill workers, phalanxes block archers, and if you go phalanx and I go workers, then in 30 turns I'll have an overwhelming tech and economic advantage. (It's more complicated than this, obviously.)

In a free-for-all, attacking is always a questionable idea, because even if you kill one opponent, that leaves 3 more who've spent the entire time building up their economies; and killing anyone is hard, because defence can normally stymie offence. In particular, the AI goes very heavy on defence, to the extent that they overinvest and lack resources for either offence or economy.

(Side note: the AI likes to build city walls in all its cities. Never, ever, do this. Yes, they're nigh-impenetrable, but it's a massive investment of resources that your opponents can obviate by simply not headbutting into a walled city. Both walls and a city itself costs 60 resources to build -- 40 shields for the settler, 20 food to replace the pop it consumes if you build at pop 2 -- so for the same price, you could have twice as many cities, and "defend" by counterattacking with twice as many units. One walled city at a choke point might be acceptable, but fortresses are a fraction of the price and almost as good, and even they are rarely worthwhile.)

So you should focus on your economy. Keep getting more cities; get caravans, and establish routes between your bigger and more distant cities; build libraries in cities with enough science (viz >= 5, prioritising the highest-science cities first). The limiting factor is tech.

I recommend getting all the tier 1 techs now, since they're only 28 bulbs. Next, target invention. It's nigh-useless itself, but it's a prerequisite for 3 critical techs: democracy (the best government by far), gunpowder (a massive improvement to ground forces), and steam engine (a massive improvement to naval forces). Which of these to prioritise is highly situational: democracy if you're under no pressure, either other if you're under attack. There are certain key techs that you might choose to target before invention -- I suggest monotheism, bridge building, and magnetism -- but I think invention is normally worth prioritising.

Once you have gunpowder, you can upgrade your defensive units, and the enemies shouldn't give you any trouble for a long time. With steam engine, you can head off any enemy ironclads, or go raid their coasts. (If they get gunpowder and coastal fortresses, stop attacking their cities, you'll lose.) Once you have democracy and revolt over, congratulations: you've reached the endgame.

The endgame[]

This is the part where you crush the AI.

Set luxuries to 40%. Your objective is to keep as many of your cities as possible celebrating, because this way, they'll grow by 1 pop every turn, a game-breaking effect. You'll have to keep your military units at home (unfortunately, this includes triremes). Cope; you'll get to attack in good time.

I like to set governors for this. Make a custom setting with positive food, celebration, arbitrarily high gold loss, and everything else weighted however you please. You might want to set it to need at least one surplus shield too. Set every city you want to grow to the governor; if it's impossible, you'll get a notification, and you can go to the city and intervene manually.

In every city, worklist markets, harbours, and temples if you don't have either Michelangelo's Chapel or Bach's Cathedral, then aqueducts. Building will take too long; you're going to buy all of these outright. This will force you to raise tax, possibly to 60%; this is fine, you can delay science for a few turns while your economy explodes, then switch back. A 12-pop city with a library and 2 trade routes will make up the difference very quickly.

Get caravans with excess production. (If one of those throwaway towns makes one, make sure to re-home it to a bigger city and send it to another big city: this will give much better yields.) Get seafaring (harbours make celebration very easy), sanitation (sewers), and university. Explosives are very valuable, as you can finally make use of all those desert and hill tiles, as is railroad. Eventually, buy universities, supermarkets, highways, offshore platforms, factories, and anti-pollution buildings as they become available and useful.

(This is where you finally lose the towns. A city uses a given tile better because it has resource multiplier buildings like markets, so when the city becomes large enough to use every tile in its environs, you are well advised to deallocate gatherers in the towns, which causes them to starve. Eventually, they hit 1 pop, and you normally have to build a settlers to get rid of that. It's not my fault the game rewards starving my own townsfolk until they migrate to the big cities.)

The AI doesn't know how to do any of this properly. Presently, you should have 20-pop cities and a massive tech lead. At that point, winning is a formality. Get much higher tech and total production, make an army of high-attack units, and run them over. I'm a fan of airports, fighters, bombers to knock out hilltop cities, and paratroopers to take the cities.

Sucker traps[]

The following is a list of buildings that are almost always terrible value that IMHO the balance team really overestimated. Some have some situational utility, but never make any of these a key part of your strategy.

City walls: cost as much as a city. Static defence, so you need one in every city, or the enemy will just bypass. Can be defeated by not attacking it. On a related note, coastal fortresses are even worse, costing more and being less generally useful, and don't even get me started on SAM batteries and SDI.
Courthouse: 60 shields, plus $1/turn, to halve corruption. If we want a 5% return, we need at least 8 corruption. The issue is there's a very narrow window between getting a city this large and getting democracy, whereupon courthouses essentially become worthless; so it's only worth it for more like 12 corruption, which has an even narrower window. And of course, if you do have such large cities, you'd probably like to spend all your money on libraries instead, which remain useful into the endgame.
Granary: this costs 40 shields plus $1/turn, so we want at least 3 food per turn value for a 5% RoI. This is almost never possible and never sustainable for long enough to be worth it.
Police station: with 40% luxury, you're not revolting anyway. Just home all your aggressive units in one city that's done growing and has enough surplus luxury.
Port facility: this is only worthwhile for a high-shield city (viz >= 15), but if it's that large, it's probably miles from the front, and your ships will spend ages going back and forth, and will be obsolete by the time they hit. Compare barracks and airports, which can make their units and get them to and from the front very quickly, especially with railroad or transports.
Power plants: just get the Hoover Dam. Solar plants have the distinction of being necessary to eliminate pollution from extremely high production cities, but it's cheaper to just have a spare engineer cleaning up.
Research Lab: by the time you get this, you've just about got every tech. Put the resources into getting more units.
Space anything: just go kill them.

And here are some units:

Catapult: good units, but they get stopped by the AI's ubiquitous walls, and they look really silly when musketeers come out.
Chariot: essentially the same stopping power as horsemen (can kill workers and militia and nothing else, at least, not reliably), for a higher price and more tech. Knights, dragoons, and cavalry are the same but more. These are partly so bad because the AI cheats and pulls workers back even without seeing these; without that, their speed can be useful. Also, you don't cheat, so don't go underestimating these.
Legion: same complaint, compared for archers.
Pikemen: obviously strictly better than phalanx, but it's a significant tech investment for very little gain because veteran phalanx fortified in cities can see off chariots and knights anyway. It also hurts to lose access to militia: sometimes it's nice to have a cheap blocker when you know you can't kill the enemy anyway.
Caravel: like pikes, these clearly dominate triremes, but not enough to justify the tech investment. You should only get navigation as a prereq for better ships, and if they'll be available in a few turns anyway, there's not much point building ships that'll be obsolete so soon. It's fine to get a few galleons rather than wait for transports; the difference is that galleons are non-military ships, so you can finally ferry settlers and caravans around without your cities rioting, and galleons do that practically as well as transports.
Frigate: maybe useful if enemies rush them out, so you can chase theirs away from your shoreline faster than if you held out for ironclads, but these are very much master of none generalists. Galleons are vastly better transports, and ironclads come out in a few turns and are vastly better warships.
Battleship: unless you have Magellan's Expedition, these are just so slow. Yeah they're cool, and they hit hard, but then they're dead in the water. The UN helps with this a bit.
Armor: just about the most expensive way of not being able to punch through city walls.
Mechanized infantry: these are excellent, but you need guerrilla warfare, and then every single blasted time you conquer a city, you'll have partisans springing up like mushrooms.
Nuclear: I guess this could be worth it if they have like 10 advanced units sitting in a city, but if they can do that, they should be attacking you.

Wonders[]

I don't recommend many wonders. Most of them simply cost too much; they're strong, but a pile of military units or new cities would be stronger, especially considering how much more quickly they'd hit. Also consider the cost of getting out-of-the-way techs, and delaying other important techs. However, there are some exceptions.

Adam Smith's Trading Co: for 400 shields, we expect 20 resources per turn for that 5% RoI. This is certainly possible during the endgame, with 20 cities with harbours or libraries. Of course, once you make it to the endgame, you've already won.
The Colossus: if you have a 4-pop city and every tile gives trade, that's +5 trade, which makes the 5% RoI simple enough. It's a fine investment if you can get a library or market to multiply this, if not very flashy.
Darwin's Voyage: 300 shields to buy two techs, both probably worth quite a bit more than 300 bulbs at that point.
Hoover Dam: as an endgame wonder, it's usually not necessary, but it's certainly excellent value to get 50ish hydro plants for roughly the price of 3.
Isaac Newton's College: another endgame wonder, but 1 free library per university for the price of 5 libraries is good value.
JS Bach's Cathedral, Michelangelo's Chapel: both great value. Not the end of the world if you miss them, though, and you shouldn't need both. It may be worth detouring to get monotheism specifically for this, especially if you think you can get it for free from philosophy; but then again, that entails delaying the endgame.
Magellan's Expedition: great if you expect to use lots of warships. This is pretty much necessary if you want to field battleships: you don't want them to kill one enemy unit, get badly damaged, then spend 10 turns limping back to port to heal, or 15 turns sentried in enemy territory, doing nothing.
Statue of Liberty: definitely worth it if you want to try communism, probably if you want democracy and you time it so it's built the turn after you finish the research. Saving ~3 turns of civil war is a pretty big deal at this point in the game.

There are some that look great but are actually mediocre. These aren't the worst wonders of all, but they're still not worth the price.

Great Library: you'll pick up a couple of techs, but you can get the same by either trading or making more diplomats to just steal the tech. 300 shields for the wonder could buy 10 diplomats, and I don't think you're getting 10 techs out of this.
Leonardo's Workshop: remember the 5% RoI rubric? This would need to give you 20 shields of upgrades per turn. Most of the time, it just ... doesn't. It's extra disappointing when its chosen unit is a caravan into a freight, but even something like upgrading a phalanx to a pikeman, in a city that clearly isn't going to be attacked any time soon, is useless.
Pyramids: by the time you get this, there's very little value in having cities grow by collecting enough food.

Then there are others that are just obviously bad.

The Lighthouse: just get another 5 triremes. It'll go obsolete before you can make any units that are worth having as veterans, and getting units out fast is much more valuable than having veterans, especially for mostly-economy units such as triremes and caravels.
Marco Polo's Embassy: just get 4 diplomats and make normal embassies. It costs 60% as much and doesn't go obsolete.
SETI program: research labs come so late that they're practically useless, so something that comes even later and requires them to have any effect is even worse.