TeamPlay AI

[AwarE]

I feel compelled to go into bat for the AI's honor.

I have made over 300 AI maps, I have watched countless replays, I have never seen any so-called experts (say +1500) play any of my maps properly. They are not fools and they know how to gain an advantage over the the AI before the game even starts. Generally they do one or more of the following:
1./ set 2x resources ....this halves the AI strength.
2./ set AI faction as UEF ...this ensures the AI can't get an Experimental into the Human base.
3./ set AI type to Adaptive ...this kills the AI economy as the map has been set for Turtle or Water.
4./ set more than two AI ...this ensures the AI has nowhere to expand to and will not build a proper defense.
This is all very clever and mainly done by 'experts'
Noobees make a different set of fails apart from all of the above, they tend to:
5./ set the AI at the wrong end of the map ....FA markers are one direction only.
6./ set pre-built bases ...this prevents the AI from having any build power advantage.
7./ set the AIx multiplier too high ...1.2 max for noobees and 1.5 max for experts.
8./ set non Sorian AI.
9./ set a 'No Rush' timer.

If the multipliers are set too high you will not enjoy the game. Depending on how high you set them the AIx will hit you with Experimentals before 20mins. Then the AIx will nuke the shit out of you at 40mins. NO FUN.
Its funny that the only people that I have seen play my maps correctly are noobees (1000 or less). Often they are told by 'experts' that have no clue ... "Playing the AI is not good practice."

Noobees will learn to play 'End Game' against the AI whereas in most custom games they are dead in 30mins. I agree that the AI does not do nasty things like transport its commy to the back of your base, attack with early gunships or attack your mex with drones ... Dealing with these finer points can be learnt later.
Playing the AI when it is set correctly will teach a new player many important facets of the game like:
- upgrade mex
- early AA
- naval strengths and weakness
- how to attack with Air
- defend against Experimentals
- ACU upgrades
- SCU upgrades
- nuke defense

Generally I try to leave the Human base the same as the Ladder map so that you are at least practicing a good build order.
AI games start very quickly compared to Custom or Ladder ...no waiting.
My TeamPlay maps have FA markers set for TWO Human verses TWO AIx.
To play single player you should set two AIx against you and one AI partner. Thats a non-AIx with you. Set the AIx multipliers @1.0 to begin then increase as you improve.

At present on any given day 70% of games are played against the AI ...if you don't believe me check the live replays by right clicking each game and look at the teams. If there is only one team then its an AI match.

I would just like to see my maps played properly , I put a lot of effort into each one to ensure the AI builds and attacks well.

roj

[tatsu]

"Playing the AI is not good practice."

they're right. quite simply. It will teach you too be good against ai and nothing else.

I say this as a guy who defeated AIX:adaptive several times before ever DARING to go online. Oh the regrets. if I had gone straight to online I would have gotten out of the 1000 league long ago and wouldn't still be stuck here 8 years after.

[Mr-Smith]

Ais are to stupid to counter single units, they build a counter for everything. they always get those nice 1 pd 1 aa 1 tmd bases everywhere. that helps with stopping small raids pretty well and secures map controll, but is totally inefficent agaisnt bigger raid.

Ai's fail to stop snipes. they feed me reclaim as they dont adapt they just try it with brute force and have no plan. if i build a few t3 artys myself, or satelites. i can easy break any base of them as they have no plan how to overlap shields properly.
they dont get their acu underwater when they see me gettign telemazer. or out of water when i spam t2 subs

they dont know how to block an acu with units so you can catch up with your own. or that you have to focus walled pds with arty first

but the worst is their feeding. always sendign 5-10 units vs acus or pds , that just gives such an insane early game mass boost.
its nice that the ai has a multiplier and gets t3 mexen in 10 min. but they fail to get tmds at that time and jsut lose them again . also they lack on that classic 50+ engis to spam 1 eexperimental fast. they always take forever to build them. and mostlikely they build 3 differnt experimentals the same time as they got t3 artys and nukes

[Vee]

Ais are to stupid to counter single units, they build a counter for everything. they always get those nice 1 pd 1 aa 1 tmd bases everywhere.

OMG! Yama-Dharma is an AI.

[AwarE]

Thanks for the replies ...you raise some valid points.

I particularly agree with Mr Smith's comment...."if i build a few t3 artys myself, or satellites i can easy break any base of them as they have no plan how to overlap shields properly."
Since shields were nerfed by FAF the original Sorian AI shield build is inappropriate. This is why I often restrict all forms of LRC and Satellites. Another issue is that Black Ops and other mods can use Sorian AI although these mods weaken the AI further because the enhancements that the mods provide were written after Sorian released his AI. Things like upgrading Hydro Power, all faction T3 Point defense, T3 mobile AA, SCU and ACU upgrades were not present when Sorian's wrote his latest version.We really need to edit the Sorian AI code to attempt to improve these finer points. Do we have the 'rights' to edit Sorian AI code?

Please keep in mind that In a TeamPlay_AI game the AI's ACU is expendable because victory conditions should be set to Annihilation. It should not effect the AI build very much if its ACU is destroyed. If either Human ACU is destroyed the game is a LOSS even if you manage to continue and kill all the AI units.

I contend that only at most 60% of FAF player base are interested in Ladder, Rank and gaining avatars. The other 40% just want and enjoyable game that, presents a mild challenge, has no rage quitters, has no extreme moves like a MonkeyLord in 12mins and they are not sniped by their team mate after completing the Co-Op task at their own pace.
The Co-Op missions are excellent and the majority of the players that like to spar with the AI are satisfied with those. You may find it hard to believe that there are other people who find the Co-Op missions too demanding to call it entertainment. lol They brake out in an anxious sweat during the missions and will not continue. The TeamPlay AI maps can provide a more pleasurable experience because they at least get to build an initial base and stabilize their economy before being heavily attacked. These people will restrict Air and set non-AIx Sorian AI's as the enemy, I think because they want to play a non-cheating AI at a slower pace.

You can't expect a pair of scripted AI to be better than a Pro Human team. Although things can be done to improve the AI ...greatly if we can edit Sorian code, but at the same time the maps landform, terrain, access to recourses and restrictions can balance the equation. TeamPlay maps that have "AIx Challenge Series' in the description are using all of these factors to gain an advantage. Just two AIx Sorian AI's are needed ...any more will weaken the AI. Good players will find these at similar difficulty as the Co-Op missions.

"It will teach you too be good against ai and nothing else." ...tatsu
That is probably correct. Umm ...that's OK.
roj

[pza]

I agree with you 100% concerning the playerbase topic. There are a LOT of people that just join FAF to be able to play Co-op missions but then stay at games vs AI, due to different reasons, for example they dont WANT to learn how to play against an adaptive human brain. In addition to that, those players minds CAN change with time, but it's their own decision when they feel prepared or bored enough to face a real opponent. As a conclusion there's no need to force anyone to play against humans "because AI is stupid". I BET, if there was no Sorian Ai, the game would have way less players and less community. SO just let them play whatever they want, sooner or later they'll join the PVP modes if they like to.

THAT also means, improving Sorian AI and the way it works within a map is a very important work for improving the game and the newbie experience. The more people like to play against the AI, the more people will stay with the game, and eventually try those coop missions and join the PVP stuff. That means it's a direct way to increase the FAF community, which should be everyones main goal.

Downlord is doing a new lobby, which COULD add better visibility for map descriptions for people playing against AI. I bet he will definetly be happy to discuss wiuth you the possibilities to implement that. So i recommend you to talk to him directly (here's the link to his thread).

[Resin_Smoker]

As much as anything else the AI should be totally revamped. It does us no good as a community, if the AI isn't at the very least compedative vs an average player or less. Reason being is that new players look at AI skirmishes as a measure of what online compedative play will be like. This assumption is incorrect and leads to allot of new players giving up with online play in frustration.

Resin

[speed2]

Side note: I asked Sheeo to give me numbers to Coop, and it's +- half as popular as Ladder, so if you check Ladder statistics on faforever.com devide them by 2, you'll get number of Coop games played, it's quite popular. Dunno how it is with playing against sorian, but Id be intrested to know that as well.

[Col_Walter_Kurtz]

THAT also means, improving Sorian AI and the way it works within a map is a very important work for improving the game and the newbie experience. The more people like to play against the AI, the more people will stay with the game, and eventually try those coop missions and join the PVP stuff. That means it's a direct way to increase the FAF community, which should be everyones main goal.

Agreed. The AI is an important recruitment tool, basically.

As much as anything else the AI should be totally revamped. It does us no good as a community, if the AI isn't at the very least compedative vs an average player or less. Reason being is that new players look at AI skirmishes as a measure of what online compedative play will be like. This assumption is incorrect and leads to allot of new players giving up with online play in frustration.

I don't understand why? If newbs play other newbs coming from AI skirmish, that works fine. It's only when they move up to higher rated games they need to adapt. Some will, some won't like that challenge but it has nothing to do with playing AI at first.

[speed2]

And there are people that just don't want to play against other people and they are happy with AI, most of them won't leave AI no matter how good and atractive you make playing against other players, yes Im talking about the one who looks down.