Group AI and Global AI

[mongus]

1) there is no good enough player ai atm to be a real trouble for a human player.*

2) There is no player ai ported into group ai YET.

* Im sure there will be, great ais, this is no pun to ai programmers, im just a darned player that has tested the ais, not a programmer, thus my ignorance and short sight. I really love all the ais there, ntai, kai, jcai.. forgive my insolence.

Oh and to adress some questions and statements:
Mongus, if you´ve tried KAI you will know that AIs can give you a tough time.

While playing kai can be interesting in metal maps, it depends on 100% handycap to be viable.

I cant explain you what hard times i can give to anyone if i play with 100% handycap. (and 2 commanders +2k m/e).

Group ais are one of the best (and less developed) feat. of this game, dont nerf it bc of nothing.


read 2)., then read 1). then read *.

pd: and for the record, giving ais the ability to cheat is, imho, much more exploitable than this.

pd2: oh, and choca is right, just refer to the ghosting building setting.

[Chocapic]

Just make the AI that will beat me and not crash the game, that's all I want. -.-

Well give me some time, and i guarantee you that my first release will not let you walk happily through the map

Edit:: KAI for now doesnt either, try it in a metal map or so its not that strong i now, but its quite ok now.
Give the ai dev's time, and it will be built

[AF]

KAI isnt all its cracked upto be when you set it at 0 handicap and put it all alone on comet catcher with rival AAI OTAI and NTAI 0.29 in a freefor all.

For one in such a situation, KAI is incapable of defeating AAI which sticks in its own corner yet ahs a roving defence of goliaths that cosntantly crush nearby KAI outposts but rarely go on an all out assault.

OTAI can eb a real pain for KAI when ti wants to be, it destroys tis mexes early on, and expands rapidly into KAI territory sometimes, building defences in awkward positions. KAI eventually crushes these though but not ebfore AAI and NTAI deal damage.

NTAI causes a headache for KAI because both KAI and NTAI expand agressively usign their commanders, and in most of the rplays I had, if KAI and NTAI where both on the east side or the west side together, then their commanders met up in the middle and NTAI d-gunned. Either that or NTAI built adv kbot labs which sent out pyros and missile kbots to destroy the large nubmers of undefended KAI resource units.

If an AI position can defend itself against two goliaths then the chances of KAI over-running it are very slim indeed.

In the games where JCAI was included, it was over-ran within a minute or two by OTAI, which sent a commander in and blew up it's factories, after which the JCAI commander was picked up by KAI and OTAI scouts and blown up.

Of all the AI's, NTAI & AAI where best at keeping hold of won territory, OTAI was best at gaining it and loosing it quickly, KAI dealt with losses quickly.

In all games NTAI started off as soon as the game started, KAI waited 10 seconds, OTAI & AAI waited 1 second. KAI always built a veh plant first then suffered minor stall at 2 mins lasting till 5, while all the other AI's had plenty.

As kai has a build tree based system, it could benefit from the commander helping the initial factory more, aswell as building more con vehicles. When NTAI builds a con kbot it tells it to help build the unit that comes after it before it goes building other things, KAI would benefit from this greatly too.

AAI needs to be more agressive, it has all the tools but it isnt using them.

OTAI needs a coherent attack force, it needs to juggle the current system it has with an NTAI/AAI style system otherwise it cant dominate without a buildup..

AAI also stalls for a while after building a mex and solar on some maps, sometimes more than the 10 sec stall of KAI.

KAI cannot direct an attack either, it roams the map , and if there are multiple targets spread across a wide area of similair threat it becomes too difuse and any well defended position will handle KAI much better. In future KAI battles, spread out, on anglo saxon redux, spread to the water too.

OTAI was the onyl AI ton swampy islands that took advantage of UW mex spots, and as a result it gained a big lead early game.

[SeanHeron]

Group ais are one of the best (and less developed) feat. of this game, dont nerf it bc of nothing.

What do you mean by nerfing it ? I´m not trying to say that I dont like helper Group AI (The opposite is true, I am a great fan! ). I just think there has to be some sort of control. And I still cant see the problem. If everybody chooses to use them: fine ! If everybody chooses not to use them: shame, true, but you cant go about making people play the way you want. If they dont want them then thats their choice.

You´ve got to admit that sooner or later there are going to be group AIs that give quite an advantage. And at that stage, people are going to be "forced" to use it to stay competitive. And quite a few people will probably be pissed of about "having to play" with AI helpers, because they think its not playing properly. So giving people the choice is the only sensible option.

Oh by the way Mongus, whats your name in the lobby ? I wouldnt mind a game against you :D .

[mongus]

Group ais are one of the best (and less developed) feat. of this game, dont nerf it bc of nothing.

What do you mean by nerfing it ? I´m not trying to say that I dont like helper Group AI (The opposite is true, I am a great fan! ). I just think there has to be some sort of control. And I still cant see the problem. If everybody chooses to use them: fine ! If everybody chooses not to use them: shame, true, but you cant go about making people play the way you want. If they dont want them then thats their choice.

You´ve got to admit that sooner or later there are going to be group AIs that give quite an advantage. And at that stage, people are going to be "forced" to use it to stay competitive. And quite a few people will probably be pissed of about "having to play" with AI helpers, because they think its not playing properly. So giving people the choice is the only sensible option.

Oh by the way Mongus, whats your name in the lobby ? I wouldnt mind a game against you :D .

ive kicked ya but b4 "frog" (and you mine.. but your cpu sux man.. upgrade ffs! :) ) , and aint it obvious? its MOnG.

Another stupid question, have you ever used (yah curse me for saying this again) Central Build AI?. (or Metal maker AI?).

Mind testing "Central Build" you vs me (using this group) and see any real improvement, difference?

And i see the potential enable/disable-ness of the current group ais (very nice helpers) as a potential nerfing of this group ais. but its picky, i know.
That option could be enabled, and it wont affect Group AI's usage, adoption and dev. much, i recognize.
(if you ever do that dont turn the ais off by default please!)

What disturbs me is ppl acusing the actual Group AI's of being evil, some sort of "cheating".
They are not, they are teh pwn0rz!

Its clear many of you have never even used them, and are ignorant of its function, YET, you are saying they are evil, they will turn the game into script kids land, and stuff like that. (see my previous post on why i think that is not the case yet.).


And alantai, i have not managed to get 2 ais working for the same host, this can be done, buy you need 2 clients runing each ai, this ive only tested Ntai vs Ntai (latest build, 29, doesnt build const bots!? (for arm)).

My puter quickly "copes" and i start getting.. some error.. overflow sending to local host.. so no real good tests yet.

Back to what i was asking for, REPLAYS! yeps, as i cant generate the games, can you post replays of them? :)

im interested in kai vs ntai for a start.
(1v1, no handy, core faf, sd, ashpen, duelng forts, dueling bases, rolling hills.. are good maps for 1v1 good players. but it could be Comet too, or Caustic if you dare :p, wtfomglolzokrlytybb). (and anglo saxon sux, (if you dont use water))

and great your are paying more attention to gameplay, altougth you still a n00b :D

[AF]

That sort of attitude will change if I managed to write and release all ym planned groupAI's in the next week.

They're not gameplay affecting AI's, in that they are information providers, not information users.

3D minimap
White lines for visible LOS
Border lines
Custom flares
Kills/losses
and a few others

[krogothe]

Yeah those groupAIs would be great, especially the kill/death one!

[Chocapic]

Im with krogothe too, those group ai's are totaly harmless and are very ok to be released as they will only ensure more pleasure and dont affect game play what so ever

[AF]

How about the GroupAI NTAI 0.3 port?

[Min3mat]

3D minimap? O,o

[mongus]

How about the GroupAI NTAI 0.3 port?

hm.. wonder how will that work exaclty.

assign it to a const unit and all it spams will be under ntai control?

or manual assign to every unit?..

The other day, we played with a kai player to level teams.

Must say that its better than a noob, granted.

Same thing can be done with ntai.

now, asigning things to an ai... ... .. .. . hm.

i dont trust real players... sure i dont an ai

3d minimap, oh, nice.like another spring window, but smaller and with less datail, stuff. cool.

What is border lines for?

and custom flares ai?? what does taht do?

White lines for visible los is nice too, but wonder theres a better solution..

i do use "L" key a lot, and its as usefull as it should.

(small feat req here, when placing structure radars... auto switch to "L" (los+radar) mode fits good :) ).

[AF]

Yah, the everything it builds it owns would work for groupAI NTAI.

More like a smaller version fo the map made of lines, whose colour depicts information, maybe with extra blobs ontop for units.

A white line is simple, and anybody wanting anything more cna easily press L or simply not use it.

Custom flares is a gimmick. If you draw a big circle using a groupAI ti shwos on the minimap too. So lets give users unusual flares, rtaher than just a middle click. How about an animated symbol or a picture, or a krogothe jogging, or snow even.....

[Gabba]

Personally I'm very much on the "go ahead with group AIs" side. As several people were saying, I doubt we'll see an AI that can do a good evaluation of what's going on on the battlefield any time soon. So, for example, the "perfect mexx placer" of Krogothe may be scary for some of you, but if not properly directed, it will build stuff in unsafe places, make you lose constructors or even allow your enemy to capture them, and similar stuff.

Actually I think that group AIs that really enhance the player experience should be integrated in the game interface. So, in this case, giving an area command when placing a mexx would automatically fill the area with mexxes places at the optimal positions. Why not? Personally, I don't get the most fun out of placing metal extractors manually: drawing squares to define the areas where I want them to be is way enough for me.

Lastly, if some people really start abusing group AIs, it will be still time to take action against that. All of us are aware that the problem may come up. Just don't slow down development out of fear.

[krogothe]

You guys think the dangerous AIs are the ones that place defenses/mexes for you. The dangerous ones are the ones that tell you the best places to, so you can do them yourself, preventing it from doing stupid things...

I already have an uber-ownage helper AI, its called KAIBlobs v2, sharing comm. Absolutely no human being has never ever even come close to beating me and KAI on 1v1, 2v1,3v1, or 4v1s, in fact, it took 4 star ranked players, all on 100% handicap, all allied against me, and all rushing to kill me. Well, i think its damn funny, but the people who got owned because the AI helped me probably didnt...

You lot are way too stubborn though . I already got the new mex class, and some other surpisingly powerful concepts for the GroupAIs that i will gladly use and not ever share if i see people abusing groupAIs.

If you guys need proof, have a game of nanoblobs against me with my AI. Weve never been beaten and im confident it wont ever happen so long there are humans on the other side.

And please dont come with the predictable "ooh but NB is different from XTA", i know that already, but its just a few steps away.

[Kelson]

I'll only reply to the flamebait with two comments.

Humans pick near optimal mex spots themselves by looking at the metal map. The biggest gain for a human in an AI placing mexes for them is that they don't have to pay attention to it after tasking it to do its mission. The gain isn't in getting an extra 1.2 metal from 13 metal extractors - if that were the case, 5 handicap would completely lead to players dominating everything.

[second comment removed: NanoBlobs isn't a game for which humans stand any chance against the ai - it greatly favors microsecond control over macrosecond strategy - a field the computer (by design) has huge superiority...thus the flooding of network traffic by ais playing it recently]

[krogothe]

Kelson, please take your time to look at screenshots and replays of KAI on anglossaxon i posted on this forum, check if the difference really is 1.2 metal in 13 mexes or more like 10 times that...
I kid you not, most players actually suck at placing mexes on spring-style metal maps...

[Gabba]

Kelson, please take your time to look at screenshots and replays of KAI on anglossaxon i posted on this forum, check if the difference really is 1.2 metal in 13 mexes or more like 10 times that...
I kid you not, most players actually suck at placing mexes on spring-style metal maps...

I'll try a game vs your AI sometime, or even you if I meet you online. But:

- Isn't an honor system better than anything else, when people are honest? Few have the capabilities to create powerful helper AIs, so if you can do one, don't piss off people by using it online against them without telling them.

- If your mexx placer AI works really great, why not propose it's inclusion in the next version of Spring, so it's integrated in the UI and everybody can benefit from it? Then we can see if people enjoy using it, or if they feel it plays the game for them. I feel it's the kind of thing SJ would have loved to integrate.

[krogothe]

I will not release, approve of or acknowledge the release of player-enhancing groupAIs until i see some sort of working control for them. Quit thinking im against groupAIs, im against the current lack of controlling measures.

Of course, if theres anyone whos played the game against humans for more than a just a couple dozen hours AND has done some actual Spring AI work instead of just theory-mongering in the forums, please lets have an educated, informed discussion about it, my mind is open to those in a position to discuss the topic.

Simply using exaggeration, sarcasm and utopic views on the moral ethics of players will surely not change my opinion that publishing AIs that improve its user's play when no one can restrict or even detect their use is utterly wrong.

As i stated many times before, once there is an effective way to control and restrict the use of groupAIs in games, i will not only joyfully support and accept the creation of ANY kind of GroupAIs, but i will most probably also create and release them to the community.

[Chocapic]

I will not release, approve of or acknowledge the release of player-enhancing groupAIs until i see some sort of working control for them. Quit thinking im against groupAIs, im against the current lack of controlling measures.

Of course, if theres anyone whos played the game against humans for more than a just a couple dozen hours AND has done some actual Spring AI work instead of just theory-mongering in the forums, please lets have an educated, informed discussion about it, my mind is open to those in a position to discuss the topic.

Simply using exaggeration, sarcasm and utopic views on the moral ethics of players will surely not change my opinion that publishing AIs that improve its user's play when no one can restrict or even detect their use is utterly wrong.

As i stated many times before, once there is an effective way to control and restrict the use of groupAIs in games, i will not only joyfully support and accept the creation of ANY kind of GroupAIs, but i will most probably also create and release them to the community.

Amen!

[AF]

mwha, gabbas suggestion to itnegrate krogs mex placement into the GUI would be exactly what people like storm and so on would have hated.

And now we have a neverending thread on the topic, storm would be proud.