'Solving' Civ & other Q's about how to get better

[gamemaster3000]

I want to use some lessons I learned from other games to get better at Civ:

1. Find out how good I actually am. i.e. Find out who the best players in the world and what they can do. If you have names, post 'em below.

2. Figure out how they do it. Granted I'm not willing to become the world's best civ player because I'm not willing to do some things I know could improve performance, such as looking at a city's tiles every turn or micromanaging my worker's movements. If the best players have written current guides, post 'em below.

3. Start using math or modeling. Developing some sort of objective standard to compare one result to another is key. There's no way to get past a certain skill until you start looking at ways to get another 5% research/gold/military in.

There are a bunch of problems to modeling Civ though. First, my personal preferences for the game are probably different than everyone else's. There are certain aspects I like about the game (the puzzle-solving nature of trying to maximize production with so many variables) and other parts I hate (AI can't handle 1UPT, the way to win on high levels is to cheat the AI diplomatically, the tech tree goes too fast, particularly as you get better).

I might make a custom map and the idea would be you have to research X techs as fast as you can or on turn 100 you have to add up all the beakers, gold, and military you have and put them in a spreadsheet that gives you a 'score'.

But I also want to touch on how complicated it is to even evaluate what is 'good' in Civ. Just for an example, let's look at factors that could influence the decision of whether to put down a trading post or a mine on a tile.

--What part of the game you're in (hammers are better early I believe)
--Whether or not you have Currency
--Whether or not you have Rationalism
--Whether or not you have Commerce Social Policies
--Whether or not you have a bunch of City States nearby
--Whether or not you're currently running a surplus or deficit of gold
--Whether or not you have Big Ben (gold production cost reduced)
--Whether or not that city has production buildings
--Whether or not that city has gold buildings
--The value of upgrading a unit for XP vs building a new one (i.e. what's the value of military promotions)
--What is the value of the trade route to the city

I guess 'solving' Civ is kind of ridiculous as the number of possible permutations for all your choices is probably big enough for scientific notation as early as turn 35. And you can't even really judge objectively...is a 10 population city with 10 hammers and 10 gold better than an 8 population city with 9 hammers and 9 gold?

Still, I'm inexperienced enough that I'm sure a couple of build orders could probably significantly increase my game play. What I really want is a more scientific way of choosing between what to build, what improvement to do, and what techs and policies to acquire.

 

[Michl2602]

I'm a Warcraft addict that recently unplugged and I want to use some lessons I learned from that game to get better at Civ:

1. Find out how good I actually am. i.e. Find out who the best players in the world and what they can do. If you have names, post 'em below.

2. Figure out how they do it. Granted I'm not willing to become the world's best civ player because I'm not willing to do some things I know could improve performance, such as looking at a city's tiles every turn or micromanaging my worker's movements. If the best players have written current guides, post 'em below.

3. Start using math or modeling. Developing some sort of objective standard to compare one result to another is key. There's no way to get past a certain skill until you start looking at ways to get another 5% research/gold/military in.

There are a bunch of problems to modeling Civ though. First, my personal preferences for the game are probably different than everyone else's. There are certain aspects I like about the game (the puzzle-solving nature of trying to maximize production with so many variables) and other parts I hate (AI can't handle 1UPT, the way to win on high levels is to cheat the AI diplomatically, the tech tree goes too fast, particularly as you get better).

I might make a custom map and the idea would be you have to research X techs as fast as you can or on turn 100 you have to add up all the beakers, gold, and military you have and put them in a spreadsheet that gives you a 'score'.

But I also want to touch on how complicated it is to even evaluate what is 'good' in Civ. Just for an example, let's look at factors that could influence the decision of whether to put down a trading post or a mine on a tile.

--What part of the game you're in (hammers are better early I believe)
--Whether or not you have Currency
--Whether or not you have Rationalism
--Whether or not you have Commerce Social Policies
--Whether or not you have a bunch of City States nearby
--Whether or not you're currently running a surplus or deficit of gold
--Whether or not you have Big Ben (gold production cost reduced)
--Whether or not that city has production buildings
--Whether or not that city has gold buildings
--The value of upgrading a unit for XP vs building a new one (i.e. what's the value of military promotions)
--What is the value of the trade route to the city

I guess 'solving' Civ is kind of ridiculous as the number of possible permutations for all your choices is probably big enough for scientific notation as early as turn 35. And you can't even really judge objectively...is a 10 population city with 10 hammers and 10 gold better than an 8 population city with 9 hammers and 9 gold?

Still, I'm inexperienced enough that I'm sure a couple of build orders could probably significantly increase my game play. What I really want is a more scientific way of choosing between what to build, what improvement to do, and what techs and policies to acquire.

I'm just an average prince-player, but still some advice:
- micromanaging your workers is not too much effort and necessary, forget about automating them
- read everything that comes out new in strategy articles; those that help you are written by good players (names like Martin Avito, Tabarnak and Bibor come immediately to my mind, but there are lots of good players around)
- read CiV-guide (in Strategy Articles section); it might not be updated to after patch 217 conditions, but you can still learn a lot about the game by reading it (especially the short chapter about 'City Specialization' is brilliant)
- all strategy articles are worth the read
- if you ask a specific question here, you'll get a competent answer
- read the 'numbers crunching'-thread, if you are interested in some maths
- read the 'diplomacy by numbers'-thread to get a clue of leader attributes

Want to close the comment with a beautiful quote: "I only want to say 5 words: 'Don't forget to have fun'" --Eric Steven Raymond

 

[Smote]

Here's a mathematical goal for you: Maximize your civ's cumulative Food + Production. (The sum of the food and production output at every turn, from turn 0 to current turn). Treat any unit lost as directly subtracted from your civ's cumulative production.

I find this as the best way to approach Civ. Sometimes military is necessary because outside forces threaten your production. Or military can be an investment to get more production, via capturing cities.

There is no supply cap in Civ, unlike Warcraft.

 

[Martin Alvito]

3. Start using math or modeling. Developing some sort of objective standard to compare one result to another is key. There's no way to get past a certain skill until you start looking at ways to get another 5% research/gold/military in.

I use mathematical models all the time, and this just isn't a game that is well suited to the approach you want. A few techs and social policies are extremely powerful when compared to alternatives. The result is that optimization requires achieving certain conditions as rapidly as possible. The objective function is not to maximize , or . Rather, it's to minimize the amount of time necessary to achieve X, then minimize the time necessary to achieve Y, and so forth. Fill in X, Y, etc. by win condition.

This was also true of past Civs, but the difference here is the number and disparate types of relevant inputs to the production function. You can produce with , or with , or with . Increasing the quantity of one input decreases the quantity of another, so you can't necessarily say that you've improved your game by adding 5% . You also have to account for whether or not that's worth the tradeoffs incurred.

At the end of the day, it's much simpler to go straight to the objective function, identify the key techs/builds for your victory condition and minimize the time needed to achieve them. Heuristics derived from your statistics on the demographics screen don't work very well.

 

[robvollman]

1. Find out how good I actually am. i.e. Find out who the best players in the world and what they can do. If you have names, post 'em below.

That would be me. I can come up with rude names for conquered cities and promoted military units no matter how many I have, even on Deity.

 

[gamemaster3000]

Where can I find some really great player's build orders and an explanation of why?

 

[Rittmeyer]

At the end of the day, it's much simpler to go straight to the objective function, identify the key techs/builds for your victory condition and minimize the time needed to achieve them. Heuristics derived from your statistics on the demographics screen don't work very well

My only problem with this approach is that finishing earlier doesn`t mean winning a higher % of games. If you have to be a little slower in order to avoide being DOWned by someone you really don`t want to face, for example, it would be worth it.

 

[Tabarnak]

I study in mathematics but strangely i don't use them so much. I prefer to ''feel the game'' and play with my instinct in most parts. This approach is essential for multiplayer.

But in singleplayer mode, we have 2 categories. First one implies a general approach for each type of starting locations, neighbors, type of map, etc. but playing the map given and try your best. Another one, more precise and very demanding is HOF type of games where player can reroll a lot of times to get a specific start trying to optimize to the maximum a pre-established logarithm.

The most important is to play the game and HAVE FUN!

 

[gamemaster3000]

Part of the reason I'm here is that my enjoyment of the game is hurt by the fact I'm making tons of uninformed decisions.

Even choices that I probably could model mathematically, like building a granary or a library first, and what turn they'd provide the same number of beakers at, I'd like to evaluate.

Surely someone has done research on this before.

 

[Martin Alvito]

My only problem with this approach is that finishing earlier doesn`t mean winning a higher % of games.

I assumed some understanding of optimization theory (due to the author's comments), so let me clarify a bit. You want to minimize time investment subject to the condition of not getting killed if you're just playing to win - that should be evident.

If you're playing in a Hall of Fame situation where you want to be the fastest, then you're going to assume some risk of getting killed in order to speed things up. How much risk to assume is a strategic calculation. It depends on how hard the start will be to reroll, just how much time pressure you're up against, how much time you have on your hands, and your personal preferences over risk tolerance.

Even choices that I probably could model mathematically, like building a granary or a library first, and what turn they'd provide the same number of beakers at, I'd like to evaluate.

Surely someone has done research on this before.

Granary over Library is never going to yield a solution. The only sane reason to go Library first is if you're immediately following it with the National College. Libraries are a terrible build now for two reasons - they lack specialists and late game techs have to be acquired via Research Agreement. If you're building a Library, it's because it's a prerequisite for a good build - the National College or a University.

That's why no one has bothered to publish the spreadsheet you want. It boils down to the fact that a few things (Chivalry, Steel, Rifles, Education, Constitution/Free Speech, Archaeology, Machinery in an OCC/small civ) are so out of whack relative to alternatives that your game should revolve around getting to the required techs ASAP.

 

[lilnev]

I've wondered about trying to curve-fit, just to get an idea of the overall time constant. Civ is a game of exponential growth, right? So what's the characteristic doubling time? You'd want to come up with some formula combining food/hammers/gold/beakers/culture, as a crude approximation -- try varying the formula, but hopefully it wouldn't be too sensitive -- and average over a bunch of games. That would help you figure out whether e.g. a building that cost 100 hammers to build and returned 4 hammers/turn was a good deal or not.

 

[Udey1]

Part of the reason I'm here is that my enjoyment of the game is hurt by the fact I'm making tons of uninformed decisions.

Even choices that I probably could model mathematically, like building a granary or a library first, and what turn they'd provide the same number of beakers at, I'd like to evaluate.

Surely someone has done research on this before.

If such a thing exists somewhere on these forums it is probably concerning a "world" map type. Without a random map and with set and known conditions(say iron or luxury resources at x,x,x), it works. Random map games on civ (which is what almost all the "strategy" posts here concern) don't have a set correct build order that works with every one.

Symmetrical games can have that. This game isn't.

 

[slobberinbear]

To the OP: I am not one of the players you want to study. However, I find it interesting that the game attracts such a wide variety of player types. Different players are attracted to different aspects of the game.

Particularly in Single Player, the game rewards those whose understanding of the game mechanics is superior. Martin is a great example of this; he can say somethiing that strikes me as arcane ("Machinery in an OCC/small civ," from his post above), and it makes perfect sense to him and those learned enough to understand what he means, though it would take some explaining the meaning of that comment to a big picture thinker like me.

I agree that you need some sort of baseline for comparison. A common scenario or play-along game would probably be the best way for you to get some sort of objective results as to what is "better" in given circumstances.

Also, I would add Dave McW to the list of posters to follow.

 

[Smote]

Machinery is for Iron Works, which adds 8 hammers, which is usually a 10-25% increase to a small civ's overall production.

 

[Smote]

I've wondered about trying to curve-fit, just to get an idea of the overall time constant. Civ is a game of exponential growth, right? So what's the characteristic doubling time? You'd want to come up with some formula combining food/hammers/gold/beakers/culture, as a crude approximation -- try varying the formula, but hopefully it wouldn't be too sensitive -- and average over a bunch of games. That would help you figure out whether e.g. a building that cost 100 hammers to build and returned 4 hammers/turn was a good deal or not.

The way I think about it is:

1. Do I absolutely need another unit, either for defense or an aggressive blitz.
2. If not, am I going to win in the next 25 turns (amount of time for the building to pay itself off).
3. Is there a better alternative? (Ie, a building that returned 10 hammers/turn and cost 120)
4. If the previous answers were no, then build it.

 

[Smote]

Here's some math for how I consider units, since I frequently make things simpler for myself by combining food and hammers. This is valid in the long run, since my cities usually reach some sort of food equilibrium, where surplus food = 0. This is usually because of using hills and engineer or science specialists.

Worker = +1 hammer/turn every 5 turns (if not saturated)
Water mill = +3 hammers
Granary = +2 hammers, +1 hammer/granary resource
Stables = +1 hammer/stables resource, + a lot of hammers while building horsemen
Aqueduct = +66% growth (frequently 4-10 food/turn. Worth it if city is making 6 excess food or more per turn)
Workshop = +3-8 hammers
Forge = +1 hammer/iron, + 1-6 hammers while building any land unit!
Harbor = + a huge amount of hammers while building any sea unit!
Lighthouse = +2 hammers/fish tile, +1 hammer/other sea resource tile

 

[Michl2602]

Where can I find some really great player's build orders and an explanation of why?

This might not be exactly what you are looking for

http://forums.civfanatics.com/showthread.php?t=416112

but it seems to be a good start for your own research, and it's exactly explaining those "milestones" Martin is talking about in this thread also.

 

[gamemaster3000]

Some really good feedback here.

Martin's Tearing up the Tech Tree article that you linked; kinda funny you did. I ran across that post, and I literally loaded up a game to try and follow it. I didn't even get as far as placing the settler when I realized how frustrated I was with the fact that I can't even tell what good terrain is that I went into the worldbuilder and made a little 'puzzle' map for me to play and try to optimize. I'm going to post it after I work out a few kinks.

But I started this post after realizing that there was so much between the lines of that post that I needed to fill in the gaps.

Am I right in saying river hills are best? After that, how much food and hammers do you need? I kind of expect a capital to have 2 luxuries or 1 luxury 1 strategic. And I really don't know how to evaluate the benefit of fish.

One other, totally personal thing. I don't find research agreements/bulbing Great Scientists/Oxford University/Great Library that fun. I need to decide if I want to play a way I don't prefer so that I can get other's advice, or if I want to play my own version of the game, in which case I need to do my own research on what's optimal.

 

[hix]

The best way I have found to improve parts of my game, and therefore the whole, is to eliminate variables in overall strategy.

Play a small map, with Genghis. Grind a Keshik rush out and maximize production.

Then take Siam, and focus on getting as many beakers as possible for a science victory.

...and so on.

By focusing on specific goals (especially rushes) in specific games, with a decided win condition at the start, I have become a much better player.

 

[joncnunn]

Currently I'm playing my first game on King; just hit Industrial era after several blow-out victories on Prince.

Micromanaging in Civ 5 is a lot easier than in previous versions; mostly the having fewer cities. I control all workers myself (in all versions). As to city tiles; in Civ 5 it's more a matter of lock a tile when the city grows or an improvement completes and forget, with occasional checks when a tech is discovered rather than every single turn.

I want to use some lessons I learned from other games to get better at Civ:

1. Find out how good I actually am. i.e. Find out who the best players in the world and what they can do. If you have names, post 'em below.

2. Figure out how they do it. Granted I'm not willing to become the world's best civ player because I'm not willing to do some things I know could improve performance, such as looking at a city's tiles every turn or micromanaging my worker's movements. If the best players have written current guides, post 'em below.