Version 0.82 discussion thread

[PieceOfMind]

it makes warring much more efficient and constructive

lol I love your choice of words.

 

[TheMeInTeam]

Worker first? Really? I always go worker as soon as I hit Pop 2. Work boat is a different matter though, as the city keeps growing. What difficulty level are you talking?

Several things are different for the AI which mean that human strategies don't always work as well. On Immortal and Deity, the AI starts with at least one worker so it's kind of a moot point. The AI also doesn't have the same lower happiness/health caps that the player does on Monarch+, so growing its capital early is more useful.

This change was made from watching AI starts on Prince and Monarch ... often times when the AI went worker first it then had a worker sitting around doing nothing for several turns. The AI doesn't integrate city training choices and tech research like humans do, so it would often go after a religion or hunting while its worker waited until the AI decided to get Ag or Mining. Or, it will get Mining but then require Bronze Working to clear trees off of hills, so its worker remains idle.

It's certainly easy to change back if we get to the point where it makes sense for the AI.

Worker 1st is far and away the best opening for the vast majority of starts (assuming the difficulty doesn't hand it one at the start). Rather than anything else, the AI should be encouraged to properly tech so it can improve its best tiles. You will see virtually every immortal/deity player out there open worker first with most starts/leaders other than possible scenarios like inland tokugawa, fishing resources w/o a better land resource (6F corn beats seafood), or those mysticism/hunting civs with forest heavy starts. For most openings, worker 1st owns everything else soundly and I'd imagine if the AI picks the proper tech the worker would handle the rest.

 

[scu98rkr]

The problem is these immortal/deity tatics has been created with the current AI in mind.

For instance on immoratal/deity there is very little point trying to get a religion as the AI will beat you. Obviously this doesnt apply to the AI and if you can get a religion the Shrine can be a huge boost.

Therefore if an AI is researching religious techs it may not have time to research mining/hunting/wheel etc and therefore initially building a worker is a bad idea as the AI might not have the techs needed to utilise him.

In fact I think human immortal/deity level tatics should actually be ignored. They are tatics for a very select situation ie where every one else is researching quicker than you.

In fact it might be better going back and looking at the general tatics that allowed people to get past noble/monarch or get the best scores on noble monarch.

I still play on monarch and if I have the right CIV will still beeline for a religion often in these situations it is best to wait for you city to grow before building a worker.

It might be an idea to have a selection of openings prebuilt into the AI. A bit like a Chess AI. Each AI can then choose an opening depending on its starting techs / capital.

In fact its pretty stupid to say build a worker for each AI as the AI will then have to research worker techs (ie mining/hunting/wheel) this will leave the path open to researching religions for the human which to be honest I think is actually more powerful.

If things are kept simple personally I agree with Jdog preposal to wait to build a worker as actually this allows the AI more choices and gives them an extra unit to stop human/barbarian rushes.

 

[MartinHarper]

I think it very much depends on the AI handicaps. The starting units and techs are:

Prince and below: nothing
Monarch : 1 archer, archery
Emperor : 2 archers, 1 scout, hunting, archery
Immortal : 3 archers. 1 worker, 1 scout, hunting, archery, agriculture
Deity : 4 archers, 1 worker, 1 scout, hunting, archery, agriculture, wheel

Emperor AIs, for example, should mostly be starting with a worker or a work boat. 2 archers and a scout is plenty of land units for the first 20 turns. There are a couple of minor exceptions. Some civs could part-build an archer while they research Fishing, and then switch to building a work boat. Other civs might find a neighbour in the first two turns of scouting, and could build an archer to steal a worker.

By the way, what happens when a civ's free techs are the same ones it gets as a handicap?

 

[jdog5000]

In games...when I receive a great general, I generally use my first one to become a "super medic" (and do not use it for an attack for risking losing it) giving it all the healing promotions and then move on to withdraw from combat options. This unit is so important in conducting blitzkrieg type wars in that, along with spies leveling defenses, I can attack and take a city, and then move on with minimal "heal time" to the next city. When my units are healing in 2-3 IN enemy land instead of 6+ turns, it makes warring much more efficient and constructive. From what I've seen, the AI NEVER does this. And, honestly, I'm not sure the AI CAN do that. But...that is why you are here and hence my question. Is it possible to have the AI plan for this...or at least implement it when possible? This would help the AI not stall so easily on attacks once it has entered enemy territory. Is this in the capabilities of coding BBAI? Thanks for any response. If what I said doesn't make sense, let me know...

The AI can do this in theory, but I also have never seen it either. Solver (IIRC) is the one who actually added the ability for the AI to choose Medic III in one of the Unofficial Patches, before then it would never happen. The AI will choose Medic 1 every once in a while, and it really does help their attack stacks keep on the move.

I'll see what I can do. The first step, of course, it helping the AI remember not to kill its medics during the attack phase ...

 

[Yakk]

So, the religion race:

The more civilisations out there, the less likely you will win the early religion race.
The more excess commerce you can muster, the more likely you will win the early religion race.
If you are spiritual, you are far more likely to win the religion race. (ie, you have the prereq tech)
There are two early religions; which you go for should vary and be non-predictable.
The better your position is (commerce + free tech + fewer rivals), the more likely you should go for the 'cheap' early religion.
Possibly role-play wise, AIs shouldn't change strategy based on meta-game issues like how many other players there are known to be.

 

[r_rolo1]

By the way, what happens when a civ's free techs are the same ones it gets as a handicap?

Nothing Said in other words, the AI ( or the human in levels where the human receives free techs ) receives the X tech for free .... if he already has the free tech, too bad for him That alone makes a lot of the diference between some civs in high levels : for example a civ that starts with Agriculture and Hunting ( Persia for a example ) in reality will only get archery as a free tech in Immortal while a civ that does not start with neither of those techs ( Portugal, for a example ) will start with 5 techs....

In fact I think human immortal/deity level tatics should actually be ignored. They are tatics for a very select situation ie where every one else is researching quicker than you.

I've been beating on this point for quite a while: human tactics in high levels are not necessarily good for the AI ... not only because of this point , but also because normally there is only a human in the game If all the AI start acting like a human in high levels does today everyone would go to the lib path only to discover that no one would have any techs to trade with them , just for a quick example. And to add, a lot of the tecniques used by high level players ( a thing that I consider myself to sort of be in spite of there being a lot of players that are a lot better than me in terms of beating the AI out ) are specifically based in the fact that the AI is somewhat uncapable to see certain dangers ( like garrisoning heavily a recently conquered city with a heavy enemy stack filled with CR siege in a hill right next to it .... ) or the fact that the AI pattern of actions is clear as water to a experienced player ( a little of self advertisement here as a example ... nothing compared with HOF games OFC Anyway this kind of analysis is only possible because of the knowledge of the AI patterns of action ....)

Getting back to the topic of the worker rule : regardless of how you look at high level strats, the fact is that the AI would win a lot by learning micromanagement of the high level players that do it right ( a group that I don't belong to ). In fact If I had to resume the principles behind the human MM in this early days it would be:

-Do what it will give you a faster settler and a stronger position ( with stronger I mean a aglomerate of techs, units, improvements ,population ,gold and some other stuff I really forgot to insert here If would be good if the AI actually had such a number in mind when it makes it's decisions ).

This obviously would decompose in a lot of sub-rules , but in terms of worker start this would probably give in 90% of the cases the advice you hear a lot in Strategy and tips forum : "You should have a worker as first build if there is something for him to do when he gets out ( this obviously implies you don't start with a worker OFC, a thing that applies to the AI up to Emperor ... if you start with a worker, things are quite diferent and would be more like "See if you would be better with a second worker ASAP compared with other options ) ... otherwise better start with something else ( again a caveat for starts with seafood ... a pretty similar rule could be applied to choosing WB as first build ) .

Well, I agree with jdog that it should not be expectable that the AI integrates well tech choice and queue issues ( atleast without some heavy work of the coder ) , but it would probably be sensible to make the queue of the first city a little special Atleast making the AI governor to look at what tech is being researched would probably help in here ( btw , confirm this to me : tech choices are made before the queue ones, right? If it is the other way around things are completely upside down )

 

[scu98rkr]

So, the religion race:

The more civilisations out there, the less likely you will win the early religion race.
The more excess commerce you can muster, the more likely you will win the early religion race.
If you are spiritual, you are far more likely to win the religion race. (ie, you have the prereq tech)
There are two early religions; which you go for should vary and be non-predictable.
The better your position is (commerce + free tech + fewer rivals), the more likely you should go for the 'cheap' early religion.
Possibly role-play wise, AIs shouldn't change strategy based on meta-game issues like how many other players there are known to be.

Being spiritual doesnt necessarily mean you have mysticism. But the rest sounds good.

 

[MartinHarper]

The more civilisations out there, the less likely you will win the early religion race.

Yes, but more civs also makes winning the early religion race generate more gold and more friends.

Possibly role-play wise, AIs shouldn't change strategy based on meta-game issues like how many other players there are known to be.

I'd prefer that the AI had the same information as the human. Players who dislike those meta-game elements can play with more of their game settings randomised.

The AI ( or the human in levels where the human receives free techs ) receives the X tech for free .... if he already has the free tech, too bad for him

I don't play on Emperor, but I wonder whether that should be changed. You could give the civ free beakers to compensate for the free techs it missed out on.

 

[jdog5000]

Well, I agree with jdog that it should not be expectable that the AI integrates well tech choice and queue issues ( atleast without some heavy work of the coder ) , but it would probably be sensible to make the queue of the first city a little special Atleast making the AI governor to look at what tech is being researched would probably help in here ( btw , confirm this to me : tech choices are made before the queue ones, right? If it is the other way around things are completely upside down )

Correct, tech is chosen first.

I don't play on Emperor, but I wonder whether that should be changed. You could give the civ free beakers to compensate for the free techs it missed out on.

That's a reasonable suggestion, give them beaker compensation. However, it means they will probably always win religion races though as they can apply those beakers to any tech.

 

[Mattygerst]

The AI can do this in theory, but I also have never seen it either. Solver (IIRC) is the one who actually added the ability for the AI to choose Medic III in one of the Unofficial Patches, before then it would never happen. The AI will choose Medic 1 every once in a while, and it really does help their attack stacks keep on the move.

I'll see what I can do. The first step, of course, it helping the AI remember not to kill its medics during the attack phase ...

Yeah...I know maybe it doesn't sound major, but having a "medic" unit (that won't be killed off) is a major help. Having units healing quickly and therefore giving the AI the choice to attack instead of squat in enemy territory, I would guess at least, would be a major improvement on sustaining attacks...

 

[MadmanAtW]

On the worker first thing, in the First 100 turns thread jdog mentions that it will build a worker at pop 1 if it can improve the tile it is working.
If we want to improve that, it might be work to change the calculation to being able to improve the tile that it is working once the worker is finished. IOW, the AI chooses Agriculture as its tech, is working a wheat tile. As it stands, it sounds like it won't try to build the worker until it has finished Agriculture, at which point it can improve the wheat tile and so it will try. It would be cool if the AI could look and see that by the time the worker is finished, it will have agriculture, and so be able to improve the tile.

That of course leaves out all considerations of being able to improve tiles that are not the ones currently being worked.

 

[Dom Pedro II]

A number of people seem to like a Medic Scout/Explorer to accompany armies. It might not be a bad idea to have the AI more inclined to give Medic promotions to them.


With regards to the Pop 1 capital thing and AI decision-making in the early game in general, I think that the AI needs to make its best decisions in the early game when the costs of making bad decisions are so much greater. The AI could be much better at calculating what actions to take if we created much more complicated processes. We could save on performance by only doing the most complex of these processes in the early game when the small number of units and cities means a lot less time for the AI to take its turns.

I mean, if Settler production is the main concern for that first city, it seems that when considering whether to build a Worker first, the AI has to consider the surrounding tiles, the length of time to build the worker and for the worker to reach the tile and build the improvement on it for the city to begin reaping the extra yields and compare that with its alternative options. But you don't want the AI doing that all the time.

 

[Poppis]

A number of people seem to like a Medic Scout/Explorer to accompany armies. It might not be a bad idea to have the AI more inclined to give Medic promotions to them.

At least it would solve the suicide super-medics issue, since scouts/explorers can't attack. I don't personally like these units as super-medics, since its hard to get them more exp, but this might work better with the AI?

That is, until we can teach the AI to properly protect it's GG units.

 

[Dom Pedro II]

At least it would solve the suicide super-medics issue, since scouts/explorers can't attack. I don't personally like these units as super-medics, since its hard to get them more exp, but this might work better with the AI?

That is, until we can teach the AI to properly protect it's GG units.

It would probably be quite difficult to turn a Scout or Explorer into a super medic. But they can certainly be a regular medic... particularly once they're no longer useful for exploring.

 

[MartinHarper]

Medic 3 is available to Recon units, but Woodsman 3 isn't. It looks like you can attach a great general to Recon units, but I've never done it myself. Protective or Aggressive Civs are better off with a gunpowder unit, to get the prereq for Medic 1.

 

[phungus420]

The medic 3 is available to _UNIT is irrelevant, these are hardcoding decisions that would not be implemented in BBAI; you don't hardcode this sort of thing (specific unit types, especially military unit types) in the dll. You guys need to take a look at why the scout/explorer is a preffered medic by players, and use that principle to describe a softcodable strategy. Remember, BBAI's logic, like BtS's, must allow the AI to handle major changes to the XML.

 

[PieceOfMind]

The medic 3 is available to _UNIT is irrelevant, these are hardcoding decisions that would not be implemented in BBAI; you don't hardcode this sort of thing (specific unit types, especially military unit types) in the dll. You guys need to take a look at why the scout/explorer is a preffered medic by players, and use that principle to describe a softcodable strategy. Remember, BBAI's logic, like BtS's, must allow the AI to handle major changes to the XML.

I second that.

Regarding the first-build choice, I agree that often it's not necessarily the best move for the AI to go worker first. However, I really don't think it's the best move to go for settler first and IMO the AI is doing this too often at the moment. It's not usually doing it as the first build but often the second build immediately after building a first warrior. Remember that first warrior comes out super fast because of the free starting hammers, so the settler is still built with a size 1 totally unimproved city. Settler at size 1 should be only really feasible at the levels where the AI gets a starting worker or possibly for Imp leaders who can get a decent number of hammers to start.

Again, none of the above needs be hardcoded.

Just reiterating, a non-IMP Prince level AI spending 30+ turns on Normal speed to build a settler while its size-1 capital is defended by a single warrior is really poor IMO.

My 0.02

 

[Dom Pedro II]

The medic 3 is available to _UNIT is irrelevant, these are hardcoding decisions that would not be implemented in BBAI; you don't hardcode this sort of thing (specific unit types, especially military unit types) in the dll. You guys need to take a look at why the scout/explorer is a preffered medic by players, and use that principle to describe a softcodable strategy. Remember, BBAI's logic, like BtS's, must allow the AI to handle major changes to the XML.

The game already does this. Virtually every promotion decision is made based on the unit's UNITAI type.

 

[MartinHarper]

It sounds like we need a UNITAI_MEDIC. In BTS, I'd give it as an option for recon units, cavalry, and gunships. Possibly Paratroopers.

You guys need to take a look at why the scout/explorer is a preferred medic by players, and use that principle to describe a softcodable strategy.

Scout/Explorer Super-Medic pros:
* Cannot be upgraded to a unit with a decent strength/power (if already built).
* Has higher than average movement for a land unit
* Can receive three of the four +healing promotions

Scout/Explorer Super-Medic cons:
* Cannot receive one of the +healing promotions
* Does not get a free prereq promotion for a +healing promotion
* Cannot gain xp.

In BTS, I think the cons are more important.