Group AI and Global AI
Moderators: hoijui, Moderators
Agree with you here: Group AI should be deactivatable when the players agree to play without. Also, a cheating Global AI is what we need to play those games where you dont know where to start if you want to survive. Or to play coop games vs a not _totally_ dumb UberEnemy (spawn scripts are totally dumb).
Re: Group AI and Global AI
Oh really? Then I must have misunderstood. As far as I could see most of your post was about how you are strongly against my suggested groupAIs (or any new ones for that matter?) and losing respect for any AI developer who releases them. You even start your post in such a negative manner and calling the suggestions as "very stupid crap" and calling the people who suggest them as having "simple minds". Though indirectly, it does get the point through. I'm no Sherlock, but your post does seem "VERY" negative to the suggested groupAI's.krogothe wrote:A pretty post with lots of smileys and emotion, but please take your time to actually read my posts in full instead of skimming it so you wont waste your time writing big posts.
I dont mind group AIs (that includes all existing ones) as long as they are controllable server-side, and i stated that several times, and i think pretty much any veteran will agree with me on that point.
One more thing I don't get is how you can say that you don't mind new groupAI's but you'll lose respect for anybody that codes them?
*edit*
I read your last post nowkrogothe wrote:Guys, no one will have a problem with group AIs (id even like to try them, for a minute or two to take a break from real gaming), but just wait until we work out a way to keep it from being used when the other players in the server dont want it (even the MM AI, in my opinion, and i use it all the time)
So I will stop this "discussion" of ours now 
i dont think the group ais are currently exploitalbe, and restricting the use of them is just plain trollish.
You are are here working in a way that cuts the original game design.
i ask for proof of concept by krogothe prior to locking the groups ais.
that would require the writting of "spoiler" ais, and there is none to my knowledge yet.
Even that scout ai that was released some time ago, seems unharmfull imho.
i repeat my request for a proof of concept of the harmfullness of any such group ai known to the date.
You are are here working in a way that cuts the original game design.
i ask for proof of concept by krogothe prior to locking the groups ais.
that would require the writting of "spoiler" ais, and there is none to my knowledge yet.
Even that scout ai that was released some time ago, seems unharmfull imho.
i repeat my request for a proof of concept of the harmfullness of any such group ai known to the date.
Sorry if i sounded too negative crow, i am just really badly against having those AIs if they are out of control. Id love to try and even write some myself if they are controlled server-side. The respect issue goes for anyone releasing them now when they cannot be controlled.
Mongus, youll have your proof of concept by mid january to early february if thats okay?
Mongus, youll have your proof of concept by mid january to early february if thats okay?
Any AI released would only be a helper. There is no way an AI can do the job better than a good human player, it would just allow the player to use his mind for something else. I personally would like a mex-placing ai, a scouting ai, and a llt-building ai. those three would allow me to focus on actually fighting the battle, rather than having to remove my attention for negligible things like self-ding a little mex and placing a moho while I'm in the middle of raiding my enemy.
But decimator, managing all that together, developing and raiding is the definition of a good player, right ?
Its the ability or not to make all things at the same time that also makes u a good or not player.
What means you have to be atentious to the right things at the right times to be a good player.
Its the ability or not to make all things at the same time that also makes u a good or not player.
What means you have to be atentious to the right things at the right times to be a good player.
If I manage an AI that does that, it's going to be fighting against you, not working for you.Decimator wrote:Any AI released would only be a helper. There is no way an AI can do the job better than a good human player, it would just allow the player to use his mind for something else. I personally would like a mex-placing ai, a scouting ai, and a llt-building ai. those three would allow me to focus on actually fighting the battle, rather than having to remove my attention for negligible things like self-ding a little mex and placing a moho while I'm in the middle of raiding my enemy.
I would hope that it would be doing both, Triaxx, so we can focus on blowing eachother up.
And the skill would become the delegation of tasks to be carried out by dumb constructors while you focus on more important things. At the moment though, the most useful groupai would be a proper formation AI, hopefully customisable so I can build my own formations. It is rather annoying to have my anti-air always running off ahead of my artillery simply because they are faster.
The player who wins should be the person who can out-think his enemy, not the person who can self-d that little mex so he can place a moho and then quick click back on his units so they don't get obliterated by a commander while he wasn't looking.
The player who wins should not be decided by micro-management. This is one of the core elements of the game we are all looking forward to. It has a built-in AI so that you do not have to constantly manage your base. Generals don't manage bases! Why should we have to?
We should be able to order a con to build anti-raider defenses around a certain area and have them run off and do it. I should be able to tell a group of cons, "I need heavy anti-air over here" and they'll run and build a bunch of flakkers and anti-bomber turrets without me having to lay the foundations. I should always have the option of microing them, but I shouldn't have to. Think of it like putting a former in Alpha Centauri on automatic so you don't have to deal with it anymore (and that game has a great deal of automation even though it's turn based!)
As for military units, simple behaviors would be nice. For example, I want a weasel to run away from pretty much anything that can hurt it (it will probably still die). However, I don't want it to run away from things that can't, like constructors. Again, I should have the option of modifying behaviors as it suits me, perhaps I want to run a swarm of twenty weasels to ignore all defenses and bum-rush a nearly finished annihilator. I should have that option.
I want to command my units, not pilot them.
I think the best way to implement groupais is to let the mod-maker decide whether to allow them in their mod, though even that perhaps isn't the best soultion. What I do know is that having the host decide would be a bad idea as it would cause a rift between people that like them, and people that don't. Having the modmaker decide would keep that rift between mods, where there is a rift anyway.
Well, now that a large chunk of the playerbase hates me, I should get some sleep as it is almost 3:00 AM
And the skill would become the delegation of tasks to be carried out by dumb constructors while you focus on more important things. At the moment though, the most useful groupai would be a proper formation AI, hopefully customisable so I can build my own formations. It is rather annoying to have my anti-air always running off ahead of my artillery simply because they are faster.
The player who wins should be the person who can out-think his enemy, not the person who can self-d that little mex so he can place a moho and then quick click back on his units so they don't get obliterated by a commander while he wasn't looking.
The player who wins should not be decided by micro-management. This is one of the core elements of the game we are all looking forward to. It has a built-in AI so that you do not have to constantly manage your base. Generals don't manage bases! Why should we have to?
We should be able to order a con to build anti-raider defenses around a certain area and have them run off and do it. I should be able to tell a group of cons, "I need heavy anti-air over here" and they'll run and build a bunch of flakkers and anti-bomber turrets without me having to lay the foundations. I should always have the option of microing them, but I shouldn't have to. Think of it like putting a former in Alpha Centauri on automatic so you don't have to deal with it anymore (and that game has a great deal of automation even though it's turn based!)
As for military units, simple behaviors would be nice. For example, I want a weasel to run away from pretty much anything that can hurt it (it will probably still die
). However, I don't want it to run away from things that can't, like constructors. Again, I should have the option of modifying behaviors as it suits me, perhaps I want to run a swarm of twenty weasels to ignore all defenses and bum-rush a nearly finished annihilator. I should have that option.I want to command my units, not pilot them.
I think the best way to implement groupais is to let the mod-maker decide whether to allow them in their mod, though even that perhaps isn't the best soultion. What I do know is that having the host decide would be a bad idea as it would cause a rift between people that like them, and people that don't. Having the modmaker decide would keep that rift between mods, where there is a rift anyway.
Well, now that a large chunk of the playerbase hates me, I should get some sleep as it is almost 3:00 AM
Specifying server side wetehr groupAI's shouldnt be used is a bad idea implementation wise I believe.
My diea has the same result for a much much smaller amount of work, and can eb changed ingame should players wish, and can be implemented NOW, rather than simply being a request that is added to an evergrowing queue of feature requests for a limited Spring dev team.
We're AI develoeprs, why should we solve an AI problem by asking for a solution from non-AI team members, even tho an AI-orientated approach is available that is much much simpler.
example:
Super cosntruction GroupAI:
For it to work all players must type "+superconAI", at whcih point the SueprconAI will start working and will spam a laod of messages saying it's started working so everybody knows it's turned on.
If you dont like it you simply dont type in "+superconAI" and nobody can use it as a result.
If you decide against it at any point you would type another phrase, say "+superconAI off", and the whole process has to start again for it to be turned back on again.
If someone doesnt type in the phrase then the GroupAI does absolutely nothing and the player has to do what the groupAI was meant todo by hand likew e all do now.
Simple enough? It makes sure that groupAI's only run where every single player wants them, and it's off by default. It's very simple to implement, and can be done right now, rather than in a few months when the spring team has done all the other stuff needed.
And if you're worried that someone will implement the same GroupAI without this measure incorporated, then it'd be very easy to incorporate it directly into the spring engine so GroupAI's are forced to use it else they simply wont work.
My diea has the same result for a much much smaller amount of work, and can eb changed ingame should players wish, and can be implemented NOW, rather than simply being a request that is added to an evergrowing queue of feature requests for a limited Spring dev team.
We're AI develoeprs, why should we solve an AI problem by asking for a solution from non-AI team members, even tho an AI-orientated approach is available that is much much simpler.
example:
Super cosntruction GroupAI:
For it to work all players must type "+superconAI", at whcih point the SueprconAI will start working and will spam a laod of messages saying it's started working so everybody knows it's turned on.
If you dont like it you simply dont type in "+superconAI" and nobody can use it as a result.
If you decide against it at any point you would type another phrase, say "+superconAI off", and the whole process has to start again for it to be turned back on again.
If someone doesnt type in the phrase then the GroupAI does absolutely nothing and the player has to do what the groupAI was meant todo by hand likew e all do now.
Simple enough? It makes sure that groupAI's only run where every single player wants them, and it's off by default. It's very simple to implement, and can be done right now, rather than in a few months when the spring team has done all the other stuff needed.
And if you're worried that someone will implement the same GroupAI without this measure incorporated, then it'd be very easy to incorporate it directly into the spring engine so GroupAI's are forced to use it else they simply wont work.
just to post some random shit.
have you ever played shared commander?
if you have, im sure you recall the ping pong of different orders to construction units.
I mean, there is a guy building ... a lab, but at that very same time you realize that you need to do any other stuff more inmediat or important.
(build some mt, or solar, or reclaim some free metal, or assist another lab... etc etc etc etc of thousand things can go wrong in any game).
The thing is, if you dont have a basic organization and CLEAR PERCEPTION OF THE ACTUAL BATTLE/BATTLEFIELD, and build/project/plan accordingly to THAT, you will lose.
I know it can be done, you can share, and if the there is good comunication, and the player is at least average, you can get some COORDINATE and ACERTIVE, gameplay toguether ouf of shared commader.
The point is, what makes you any of you think the "ai controlled" units will be doing what is correct for the actual situation and not fuxoring all up as some shared comm games end into.
for instance, what says a construction unit to stop building if its being attcked, and refuge somewhere near an llt?
or to start spawning some mt as soon as it sees a l1 gunship rush??
in short, those "automatic" "base building" ais wont ever (near now) know what is correct to do for the current time and game.
appart that ball of nosense, a challenge for you guys:
Point an Actual Group AI that does what you are saying, that is "play the game for the player".
have you ever played shared commander?
if you have, im sure you recall the ping pong of different orders to construction units.
I mean, there is a guy building ... a lab, but at that very same time you realize that you need to do any other stuff more inmediat or important.
(build some mt, or solar, or reclaim some free metal, or assist another lab... etc etc etc etc of thousand things can go wrong in any game).
The thing is, if you dont have a basic organization and CLEAR PERCEPTION OF THE ACTUAL BATTLE/BATTLEFIELD, and build/project/plan accordingly to THAT, you will lose.
I know it can be done, you can share, and if the there is good comunication, and the player is at least average, you can get some COORDINATE and ACERTIVE, gameplay toguether ouf of shared commader.
The point is, what makes you any of you think the "ai controlled" units will be doing what is correct for the actual situation and not fuxoring all up as some shared comm games end into.
for instance, what says a construction unit to stop building if its being attcked, and refuge somewhere near an llt?
or to start spawning some mt as soon as it sees a l1 gunship rush??
in short, those "automatic" "base building" ais wont ever (near now) know what is correct to do for the current time and game.
appart that ball of nosense, a challenge for you guys:
Point an Actual Group AI that does what you are saying, that is "play the game for the player".
Making all players agree takes too much time. and effort, most people just wont do it.
Making it mod dependent defeats the purpose of it, i want to be able to play XTA and AA with or without groupAIs, depending on my mood. Why restrict the choice?
Server-side control before the start of the game is extremely easy (one checkbox) and all players can see it and choose to play or not in that server. Everyone is happy then!
Crow, I think its great to develop group AIs, but please get the devs to implement controls on them first! Then instead of limiting ourselves to "harmless"AIs, we can then put some uber stuff up like advanced construction and attack...
Making it mod dependent defeats the purpose of it, i want to be able to play XTA and AA with or without groupAIs, depending on my mood. Why restrict the choice?
Server-side control before the start of the game is extremely easy (one checkbox) and all players can see it and choose to play or not in that server. Everyone is happy then!
Crow, I think its great to develop group AIs, but please get the devs to implement controls on them first! Then instead of limiting ourselves to "harmless"AIs, we can then put some uber stuff up like advanced construction and attack...