Seeking input on Free Early Settlers for AI civs

[cracker]

Since GOTM8 (Germany) was played back in June, I have been studying the way the AI plays the early game by generating lots of maps and quickly playing the first 50 turns of each game and then collecting statistics as to how different events happen in the early game. For the purpose of this testing I have just been setting my civs to build warriors and then monitoring the AI civs performance.

An immediate revelation that jumps out from the data, has to do with the fact that the AI is hard coded to choose its early build queue based on some ratios that it calculates. At difficulties of Regent an below the AI first build military units because it has no military. At difficulty levels of Monarch and Emperor every AI immediately locks in on building settlers because every AI gets some military units as a head start.

An unintended side effect of this programming is that AI civs on Monarch and Emperor never get free settlers out of goody huts in the early turns of any game (Never!) because they are not programmed to temporarily switch production away from a settler while they pop early huts. Humans do not have this handicap. (there were no goody huts popped for free settlers in a 24 game test sample with 240 total AI civs).

The AI civs at regent level will regularly pop free settlers from goody huts at the rate of about 10%-12% overall (using a 12 games test sample with a total of 120 AI civs).

This thread is to promote discussion and collect input on the potential option to allow the "map generator" for the GOTM games to randomly override this Free Settler block by the AIs in some cases.

Allowing this option in some games would mean that it would be possible that one or two AI civs in a game could have a different early build behavior from all the other AI civs in the game. Human players would still be able to equalize this advantage by using queue manipulation to maximize their chances of free settlers.

 

[DaveMcW]

I think free settlers are just too powerful.

In human vs. AI deity games, a settler in the first 10 turns cancels the AI's production advantage if you know what you are doing.

In human vs. human games, a settler in the first 10 turns guarantees a QSC win or earliest conquest if you know what you are doing.

The only real way to balance this is disable barbarian and huts.

 

[Yndy]

Originally posted by DaveMcW

In human vs. AI deity games, a settler in the first 10 turns cancels the AI's production advantage if you know what you are doing.

Thank you for telling me that I don't know what i am doing. (related to GOTM 14 spoiler thread).

This is about AI settlers not human ones. I sense bitterness for not having settlers. I, for one agree with free setlers for the human. So if i win over Aeson one game in 10 due to free settlers (In my dreams of course). Aeson would still outscore me the other 9 games and the ranking would be clear.

 

[DaveMcW]

Sorry, I am a bit bitter about never getting free settlers.

I am also bitter about the fact that someone could reload until they know when the free settler will appear, and that it is impossible to detect this type of cheating.

 

[Yndy]

No problem. Let the cheaters be exposed and torched on the main road.

OK this is spam. So to avoid being banned i have to say that AI not getting free settlers because it is already building them is acceptable IMO and changing that would only add some spice to the game, but we most probably won't feel it. On higher levels the AI settles very fast anyway.

 

[civ_steve]

Free Settlers are rare (I think the discussions I've seen would suggest about 10-15% of the time under ideal circumstances.) For the human player this is a chance to accelerate their catch-up to the AI.

I'm not in favor of 'giving' the AI free settlers, but if the AI did the same things that the human player does to improve their odds of getting a settler from a Hut (wait until no settlers are in play, and change their build queues to something other than settler), I see no problem with that. I also see no problem with changing the AI's building priorities.

 

[Skyfish]

It is a good idea.
Let's try it out in some games, something different for once.

 

[zagnut]

I generally feel that changes to the basic game would not be a good idea. There is a certain amount of balancing that went into the initial game design and many of the tactics we use are designed to thwart those programming decisions. As a result we do things in developing our civs that we wouldn't do in real life if we were Genghis Kahn.

Changing our production queues to try to pop a settler is one of them. In doing so we are trying to overcome the deficiency we have at the higher levels versus the AI. If you give the AI the ability to pop a settler what you are really doing is giving it an even greater advantage than it already has. I am always trying to catch up to the AI in Emperor and in Deity I never do. I'm not sure I want the AI to be even more challenging, so I would vote against this change.

 

[Kemal]

I'm also against these type of changes that effect the gameplay decisions of the AI. I think there's really no need to change this because as it stands now most participants of the GotM are already challenged enough by Emperor, and certainly Deity level.

Besides, even if you let the AI change its production, that probably won't give them a chance at finding a settler at higher difficulty levels as it is easy for the AI to outsettle the human in the early phases and thus not fulfill the prerequisite of having less\equal to the average number of cities of all players.

 

[Bamspeedy]

Simple question.

Is Cracker saying he can actually change the AI build orders, or is he saying he will just automatically give (1 or 2 civs) a settler right at the start (thus changing their build orders indirectly).

I interpreted it as he will just give 1 (or 2) civs a settler right at the start of the game. Since they already have a settler, they may change their build orders by producing military, infrastructure, and wonders a little sooner than they normally would. All the other AI would still play at the level that is chosen and is normal for that level, but 1 or 2 civs will have a little 'jump start'. I doubt he will give any AI a free settler at Deity, since they already start with an extra one.

 

[Yndy]

The way i understand cracker's suggestion is:

randomly the AI would ignore the prerequisites for popping a settler from a goody hut and hence will have a chance of popping a settler even while he produces one or even if he has more than average number of cities.

I might be wrong as you seem to understand it differently.
Cracker could you elaborate please.

 

[cracker]

Bamspeedy, you are wise beyond your years.

Essentially this poll is one of many you will see to give people a chance to both express their solid opinions (I really want feedback and all the GOTM staff volunteers really need a better feel for who our players are and what they think) plus it is an opportunity for the community to discuss some of these fundamental issues.

There are some big picture issues that we may want to creatively address but first we need to get a pulse as to where we are with understanding how some of these things might effect people's opinions of the game.

The general idea on testing the waters with the idea of slightly altering the starting unit allocations would lead to two major options:
1) exploring eliminating goody huts in some games will retaining the effects through randomly generating a list of the goody huts results a civ would get.
2) leave all the default settings for the difficulty levels unchanged but explore the option to have a random external process that would potentially choose to give a slight boost or a slight handicap to some civs so that there would be more diversity in the levels of AI civ development.

We do not want to change the way the basic civs in the game behave in most games because it may be important to some people to keep things identical to the out-of-the-box game settings.

Giving some more though to the "boost" options would just allow us to eliminate some of the random variable outcomes that create early differences between the games of two different players when one of our objectives is to try and play the same game to foster learning opportunities and promote some sort of valid skill comparisons.

I am really noy trying to lobby in favor of one approach or the other at this time but really just wanted to fish for inputs and try to get a sense of general feelings.

 

[zagnut]

Posted by cracker:
"Giving some more though to the "boost" options would just allow us to eliminate some of the random variable outcomes that create early differences between the games of two different players when one of our objectives is to try and play the same game to foster learning opportunities and promote some sort of valid skill comparisons."

With this explanation I would change my vote against the mod to a change in favor of the mod.

 

[ControlFreak]

After watching the enemy pop all the good goody huts on my land in GOTM14 and watching his/her/their (I'm trying not to spoil here.) tech jump by leaps and bounds over mine I voted to "Surprize us and see if we notice the difference."

I think anything that will lead to the elimination of goody huts in comparison games like GOTM is a good thing. I've read through too many threads that included the phrase "if you hadn't gotten that free settler I would have..." Since my true interest in these comparison games is to see what strategies work and work better having the randomness of goody huts is a drawback. I understand that including them teaches that exploring is benefitial and strategies with a ton of warriors scouting all over is a good thing not just for the hut but for the contacts. But the range of a single RNG roll and the fact that goody huts are mostly in the very, very early game have a huge impact on a civ from the beginning. The difference between a free settler and 3 angry barbs that kill a warrior + 1 population point in your only city is a gigantic swing in power that can not be overcome because of the steamroll effect of Civ3. Even the difference between you getting a tech you can trade and the enemy getting a tech to prevent you from trading with it is a big difference.

Well enough :rant:

Give a few settlers out, don't tell us how many or to whom, get rid of goody huts altogether (if you can do this). Then let us know after the games are in where the settlers went so we can compare relative strengths at the end of the game.