Starting question: Is this fair????

[Cort Haus]

Originally posted by Olavi
A good reason to wait for Civ V!

In fact, it's not wise to look at the next game for a better AI, because invariably it's not the game designers but hardcore players who best know how to beat the game. Civ 4 is the first to have a user-programmable AI, and I expect that by the time Civ 5 comes out, the Civ 4 AI might be pretty handy.

Of course, if they hire Blake to write the AI for Civ 5 ...

[binTravkin]

To say even more, it's often obvious that AI gets programmed by people who are not so keen on playing the game or games in general, so their understanding of an AI is mostly theoretical and their initiative in building that AI often leaves a space to wish for better.

Some things Blake has already fixed or spotted to be wrong in code makes one think 'what were they thinking?'.

[Thedrin]

Anyone writing the AI for a new game only has theoretical knowledge and severely limited first hand experience of the game mechanics. I would say that the reason we are seeing such improvements now is not because the original code was written by people without sufficient interest in the game (far from it) but because there do exist, now, players who can be described as experts at the game.

to Firaxis for giving players the chance to make their own changes to the AI.

to Blake for giving it a go.

to the writer of the original AI code for making it challenging enough that people continue to play the game.

[jbp26]

Originally posted by Blake

For example, for me, a typical terrible (economy) start goes something like this:
Build a second city very near my capital for logistical support (ie training units).
Build the Oracle in my capital and grab Code of Laws.
Run Caste System and turn nearly all my population (except those on grassland cottage or better) into Scientists, once the banked food runs out, run as many as possible without starving.
Lightbulb Philosophy and switch to Taoism & Pacifism (possibly slavery too, since Caste has served it's purpose).
Use Philosophy to help get the Oracle Prophet and build a shrine.
Pump out Great Scientists and settle in capital.

This kind of power-GPP opening provides an indomitable tech-lead over the AI's and the Great People provide income regardless of terrain. And the AI's can't even imagine pulling off such a complex strategy - that's how big the expert humans advantage is over the AI's.

One thing I don't understand about this strategy- running all of those great people (presumably just in your capitol?) means that your best city is effectively not producing any units, buildings, or wonders- just GP. So does this mean you just leave the other cities to churn out units (so you dont get attacked) and ignore/capture wonders/cities once your capitol has a good number of people?

[Blake]

Great People provide free production once settled - like that +1h from each great scientist adds up. I usually go with burst production to build wonders and if there's nothing critical to build it runs more specialists. A specialist burst doesn't take long - for example the first GP costs 100GPP, if I use Caste System to run 5 scientists it only takes 7 turns - only 4 turns for a Philo leader - think about that, with like 11 turns of max specialists, a Philo leader can have 2 settled Great Scientists, generating +2hpt and 12bpt, that most likely will be a larger effect than most nearly any wonder.

As I said I very quickly found a second city capable of popping out units. What I try to do is try to found a second wonder pump, and a second specialist pump, then my capital is freed up to build wonders and work cottages. The great people still get settled there of course - but they don't have to be generated there.

Usually after building half a dozen wonders in the capital you don't even need to run specialists anymore, or maybe just 1 engineer. The important thing is doign that early burst of GPP, to get the gains from the cheapest GP's asap. It's a better investment than nearly anything else.

[Diadem]

If you settle a lot of GP the production from them can indeed become quite significant. Especially if they are engineers or great prophets. You want to avoid generating artists if you use this strategy. And of course you beeline for civil service and run bureaucracy.

Since specialists and settled super specialists produce quite a lot of commerce (in the form of gold or beakers) you can focus on food and hammers in the tiles you are working, allowing a decent to good production rate in your capital. Add hammers from super specialists and you can reach a very good hammer output.

Enough to build wonders and buildings. Units you want to produce with your satellite cities.

As Blake said, later on you can use another city to generate great people. That way you can build cottages instead of farms around your capital and really optimize the Oxford University you will build there.

I usually distribute my national wonders like this:
- My capital gets both Oxford University and Wall Street. It's fat cross will be filled almost entirely with cottages, only enough farms to reach size 20.
- A city with very good food resources or floodplains gets both the Globe Theatre and the National Epic. This city gets only farms, and perhaps a few mines to add some production so I can build wonders there. This city will be my main GPP pump. While early wonders go into my capital, I try to build late wonders here, for the GPP points.
- I try to found a city somewhere where it has very good production, and optimize it for that city. This city gets West Point and Red Cross. This city will build elite units.
- In some other good production city I build both the Ironworks and the Heroic Epic. This city mass produces cannon fodder, as well as air units in the late game (which don't have experience).

As you can see each city is highly specialized. And of my 4 best cities only has has any science production to speak of.

[Beatie]

Very good advise I've read is this tread here! I already learn to specialize my city's off course, but the advise to specialize the way Diadem told us, seems very intresting indeed. I usual just use my capital for very high production and sience as well. Building cotages instead of farms, is something I'm going to try. So main city builds sience/money!


And Blake: thanx for insight into tactics to create GP fast. I've never used Castle system to many. Perhaps I need to give it a go.

After reading the post here, Can you say that the AI is more like that they behave differently, and the bonusses that recieve in the higher levels, makes the AI more difficult to win upon?

And about the specific AI's like napoleon and Alex.. who can make a list with lets say 8 AI's to have a somewhat quiet first part of the game? Does it matter on wich leader you pick to play with? Example does Isabela like to wipe out the USA rater that anybody else ?

Thanx again for those usefull posts here!

[Diadem]

Well your main science / money city doesn't have to be your capital of course. Sometimes your capital is ideally geared towards production or GPP generation and then you should put it in that role.

But your capital has a natural advantage over other cities. It's usually at the best site, it has bonus commerce and production if you run bureaucracy, etc. And since science output is the single most important quantity in your civilization, it is logical to specialize your best city in it.

Another role that makes sense for your capital is that of GPP pump. If you have good food resources or floodplains near, that's a good role for your capital as well. You usually build your early wonders there anyway, so that's some bonus GPPs. But then you need another city to recieve super specialists.

[Diadem]

Oh, by the way, Blake, I'm a little surprised to see you arguiing for caste system here. I've always enjoyed caste system myself, of course. But in your Improved AI thread you wrote somewhere that you programmed it in so that the AI would never run caste system because it's useless. Appearantly that's only for AIs?

[fed1943]

The only point I do not agree with Diadem:

I see no need of a lot of medics: so, I prefer Heroic Epic paired with

West Point and Iron Works paired with Red Cross.

Best regards,

[binTravkin]

Appearantly that's only for AIs?

They can allocate specialists as they wish without CS.

[Bobby Chicken]

If I took a game into the middle ages, and my dad picked it up from there, I'm pretty sure we could own ass on deity. He always starts out like ****, but then gets rolling about that time. I just can't seem to bridge the gap between the early and mid game. In the early game, no AI, even on Deity, stands a chance against me. And whenever my dad has a chance to get his army rolling, it starts rolling and keeps on rolling. But right now, we're working on winning on noble with no tech trading and science and culture bars set at 0% the entire game. Deity is not really in a league of it's own. As I see it, the game levels are broken down like this. Settler thru Noble: anybody can win
Prince thru Occosional Win on Emperor: take a good player; most players probably fall in this category
Emperor thru Deity: On these levels, only orthodox players will own. By orthodox, I mean town emphasis, tile working layout that won't leave regrets, smooth start, clever diplomacy to guard your back and get tech, no early religions, etc.

Keep in mind that the only win I consider a win is a pangea time victory. Pangea is for real; continents terra, etc. are fun but they don't showcase orthodoxy and include a lot more luck. I'm tempted to put in that orthodox players never pop rush and always play financial civs. But, although this is the general rule, there are a lot of exeptions.

Diadem, I don't know about you. I beleive in having a medium sized or larger empire, using essentially no great people, and always playing financial. I never specialize cities. If you can consistently when on deity, feel free to correct me.

I sound kind of harsh (I always sound so mean in writing), but it wasn't until I began winning every time on emperor that I finally broke away from the kind of GPP and city specialization stratagies that you guys suggested. It works great on lower difficulty levels, but your ARE NOT EVER GOING TO WIN ON DEITY with great people, religions, etc. You win by working your tiles in the manner most conducive to dominating the tech tree, progressing up it in the manner best for progressing further up it, attacking only to maintain your edge, and trading only for a technological edge. Not only does this style allow for a win on very high levels; it is far more consistent. As I have said, I am capable in winning on emperor with such consistency that it's boring. Do what ever you want to do to have fun, but if you want to win, you're going to have to place all extras aside and make it up the tech tree.

I'm sorry, but it really pisses me off when someone has one on a duel sized terra map on emperor and thinks they know everything about the game. I'm open to trying new ideas, but usually I find that your ideas are the same ones I had before I had ever actually played the game. Focus on the fundamentals; keep the end goal in sight. Win.

[Blake]

Originally posted by Diadem
Oh, by the way, Blake, I'm a little surprised to see you arguiing for caste system here. I've always enjoyed caste system myself, of course. But in your Improved AI thread you wrote somewhere that you programmed it in so that the AI would never run caste system because it's useless. Appearantly that's only for AIs?

It's like this... Caste System might be useful for 20 turns of the game - it might be VERY useful for those 20 turns, but it needs EXTREME focus. The AI fundamentally... meanders too much, they can't focus on one goal. What's more, when they do generate a great person, they don't use it well. So it doesn't matter *which* GP they get.

Exampels of uses for Caste System:
Generate a Great Scientist for an Academy or Philosophy lightbulb, without first building a Library.
Run LOTS of scientists to help overpower the prophet GPP from oracle - like running 2 scientists gives a 33% prophet chance, running 4 drops that to 16%.
Run a lot of Artists to force a city to flip, or to get a Great Artist to bomb.
Run Merchants to get a GM to lightbulb Civil Service (under limited cases this is better than Prophet CS, and you can go with mixed Priest+Merchant points and use either)

Caste System is a useful precision instrument - very specialized. But the AI does much better work with blunt objects.

[Blake]

I sound kind of harsh (I always sound so mean in writing), but it wasn't until I began winning every time on emperor that I finally broke away from the kind of GPP and city specialization stratagies that you guys suggested. It works great on lower difficulty levels, but your ARE NOT EVER GOING TO WIN ON DEITY with great people, religions, etc. You win by working your tiles in the manner most conducive to dominating the tech tree,

There's only one cause for Wins on deity.
The AI is retarded.

Specializing cities, building wonders, founding religions, these are not nessecarily poorer play than "Orthodox" what you may or may not get, is that on Deity these things are IMPOSSIBLE and that's why players don't do them. The player is forced to make do without them. That you CAN win without them doesn't mean it's EASIER to win without them. It's just that the Deity AI - even empowered by the bonuses these things provide, is still so retarded that they can lose.
Financial is a popular trait at Deity because Financial is LEAST reliant on leveraging bonuses from Wonders and Religion. In fact it seems many players find they need to use Financial to win on Deity, and that is primarly because they're being denied all the things which other traits leverage.

Not everyone cares to play on a difficulty where AI bonuses eliminate 80% of early options. Some people prefer to play a more balanced game, even if the AI is easier to trounce.

Deity isn't Civ.
Deity is a difficult variant where you aren't allowed to found religions, build wonders or generate great people. Play that variant if you want, but don't pretend you're playing Civ.

PS. My favorite variant is playing with no terrain improvement except roads and pillaging any captured terrain improvemen. Monarch is entirely winnable with that variant, altough I havn't tried Deity. (Oh yeah, all land map, no lighthouses allowed).

[Diadem]

Orthodox? No, I have never considered myself orthodox. Even in the old civs (1 and 2) I always built only a dozen or so cities, where everybody else used ICS strategies. Moreover I still consider my strategy superior ICS just can not compete with democracy + 100% luxury slider

I don't play much on deity precisely for the reasons given above. It's much more fun to be able to fool around a bit, use strategies that are maybe sub-optimal but a lot more fullfilling, etc. But I disagree that deity is so bad as you guys make it sound.

It's possible to get the pyramids on deity, for example. Not without a very good starting position, granted, but it's possible. And great people are most definitely still useful on deity. More useful than on other difficulties almost, since you have a lot less cities (because they AI fills all the available land in a blink of the eye) you need to optimize the few you have. You do this by specialization and optimal usage of great people.