Spring RTS Engine

I find most of those AIs absolutely wrong, it is plain obvious cheating, a type of which we cannot restrict.
I could get my metal class to tell me where to put mexes in a way that it will give me the 100% best spots, giving me a definite metal advantage on maps where the metal map is "cloudy". Then i could move on and let it do all my micro as i say, giving me a huge advantage in any game where the speed is over 0.1...
Are you guys serious? It is blatant, obvious cheating, not to consider the pointlessness of letting the AI play the game for you, especially when the AI can do some things A LOT better than humans.
I would be fine with it if it could be disabled server-side, but coding those group AIs (provided you are a decent programmer) will simply screw it for puritan players. I am already unhappy enough with having players not knowing about the metal maker AI and wasting millions of energy units over a game while other players do know about it and use it. Now imagine you can have all those fancy AIs that lets you concentrate on specific things while doing the background work, which will probably be updated often so whoever knows about them and has the latest version gets an advantage, wow that├óÔé¼Ôäós fun isn├óÔé¼Ôäót it? It├óÔé¼Ôäós like playing chess with a chessmaster 10k advisor helping you: Its nothing a player couldn├óÔé¼Ôäót do himself but somehow whoever uses it wins a lot more often.
If i see those AIs implemented in large scale/numbers without server side control ill stop playing spring, they simply defeat the point of the game, which is managing your time well between economy and fighting.

The problem is not with the pros not liking it on, the problem lies with all the noobs who cant play turning them all on and beating the pros because all their economy was sorted for them and they knew a tiny bit about microing attack forces... You guys might think the AI is crap compared to a human but in some aspects (especially economy), AIs can take "perfect" decisions, such as picking a mex spot in cloudy metal maps like small divide or telling you the right time to build a certain building based on your income....

AF, please don├óÔé¼Ôäót code that, it really spoils it for those who play a lot and actually have enough brain cells to manage their forces properly...

lol, aint the ethical use of ai a mtr

e: woot there is mtr guy!
just for the record, i recall the "ethical use" topic in every major thread related to the use of ais and how they will interfere with the game/gameplay.
That is like 3 threads, including this one. i may miss some.

Nope

I think work AI of any type should be possible to disable server side. At the same time however I don't think we should just rip the function out of the game, it could be really interesting to see what kind of playing AI people could come up with and make work with a little guidence. I'm expecially interested in micromanagement AIs, as they could do some really cool things. If anyone is interested in writing a micromanagement AI I'm more then willing to help you with the tactical side of it.

krogothe wrote:I could get my metal class to tell me where to put mexes in a way that it will give me the 100% best spots, giving me a definite metal advantage on maps where the metal map is "cloudy". Then i could move on and let it do all my micro as i say, giving me a huge advantage in any game where the speed is over 0.1...

Good! If it's so "perfect" then we'll use yours

What is really unfair is the noobs who don't know the ctrl + z playing against people who do know the ctrl + z function! No wait. It just means the player is a noob and hasn't learned all there is to know about the game. One day that noob will learn about ctrl + z and about the MMAI

We could call the new AI's "The noob tools!"

the NoObIe WhEeLs groupAIs can announce themselves whenever people use them in-game!

Now that we got a name, we just need the AI

Krogoth, I have alreadyw ritten the code,a dn so ahev you and zaphod and sub and all the other AI devs, it's all lieing there all done and dusted in our skirmish AI sources for anyone ti simpyl copy paste itno a GroupAI and recompile.

Why these people who want ti so badly havent done that already is beyond me, you dont need to dunerstand the lot, you just need to know which files to copy paste form then do the smallest of modifications and hey presto you've got what you want without any major work.

I am not going to do that, if they want it they can do the impossibly simple task of copy, paste, compile themselves. I will release GrouPAI's that i feel help the human player make decisions, I will not make groupAI's that amke the decisions for them, and I will most certainly not do them because I want kudos, but rather because I can and to leave it sitting on my hd after doing ti would be a waste.

I also think server side GrouPAI control isnt the best idea either.

My suggestion (whichc an eb implemented in the here and now without the necessary changes that serverside cotnrol would need, which are actualyl much greater than what you anticipate) is that a grouPAI have an activation keyword phrase.

For example if you want to use the metal maker AI (this si a new version I've made up that does resourcing for you! as an example of course). I would set ti up so that every player has to type in "+metalmakerAI" before it'll work. And if one person doesnt type it in then Whenever the players startup their metal maker AI's, they wont work, nto untill that last player agrees and types the phrase.

GroupAI's should also make a big show of themselves in the console that they've been started so other players know when you use them.

It wouldnt be that hard to make a GroupAI respond when .groupAIs xx is typed where xx is the player number of that player running the groupAI, thus creating a way of getting a list of all the groupAI's another player is running.

Those 3 above checks are good enought o solve all these problems and can easily be implemented in groupAI's in the here and now. Whatsmore they mean GroupAI's will only work ingame if EVERY player agrees to usign them. It also means groupAI's cant be used in skirmish games forcing rookies to learn the fine art of micromanagement and decision making themselves, thus preventing the predictions of a n00b player culture where players only play half the game.

Anything that only allows groupAIs when all players agree with it (server side achieves that if displayed on server options in the lobby). However anyone can still remove the check for chat, so please dont release the source for it...

We don't need an AI to teach noobs, we need several simple AI's to take care of the lame bits of TA, like spreading out your mex's evenly. Its simple, it's mind-numbing, and I could go without it. What would be nicer though than assigning a unit to an AI would be to have a mex button. When you press it, an idle Con KBot or whatever decides where to put it, and builds it, and when its finished it's removed from the que. The button should have nothing to do with the units you've selected, it should just be a useful button for managing resources while you are focused on a battle or whatever. Same for mohomines - a left click means build one on unused area, a right click means reclaim a mex and build a moho.

That's probably all impossible but anyway...

krogothe wrote:I find most of those AIs absolutely wrong, it is plain obvious cheating, a type of which we cannot restrict.

What about a group AI for construction units that would cause them to automatically seek out and repair damaged units within a defined area? Or a similar one that causes them to automatically reclaim or resurrect dead units?

Or, instead of placing an order for each dragon tooth when fortifying a position, how about an AI that you can give a "starting" DT position and an "ending" DT position to and it automatically has the construction unit fill in an unbroken line of DTs between them? You could have similar AIs for having construction units to build a square box of DTs, or to surround an existing structure with DTs.

What you're calling "cheating" is what others call "relief from pointless micromanaging," and it's not obvious which is which, it's an opinion. You're free to not use such features if implemented, or to argue that they're not a good idea to have in the game in general, but calling me a cheater is IMO a bit beyond the pale.

krogothe wrote: Are you guys serious? It is blatant, obvious cheating, not to consider the pointlessness of letting the AI play the game for you, especially when the AI can do some things A LOT better than humans.

The point of something like this is to free up the human from having to bother with the things an AI can do "better" (at least in terms of being able to pay attention to it at all times, if not the details of execution) so that the human can focus on those parts of the game that the human can do better. And that this particular human happens to find the most fun to focus on, to boot.

When you've got an AI that can, completely on its own, defeat a decent human player then perhaps this aspect of the issue will become significant. I don't expect that any time soon, I've never won a multiplayer game against humans and yet I've never lost a game against AIs even when I was just intent on messing around with new units to see what they can do.

krogothe wrote: If i see those AIs implemented in large scale/numbers without server side control ill stop playing spring, they simply defeat the point of the game, which is managing your time well between economy and fighting.

Every multiplayer game I've been in so far has been a team vs team match. What if one of those teammates chooses to be the "economy" specialist, spending more of his time planting resourcers, with another player as the "fighting" specialist? Does that also defeat the point of the game? If so, would you support removing the ability to share resources among allies, or to have several players starting with the same Commander?

I agree entirely with everysingle thing BeeDee said.

BeeDee wrote:What about a group AI for construction units that would cause them to automatically seek out and repair damaged units within a defined area?

No need to use an ai for that.
Select some const. units, set repeat, and apply the area to repair, reclaim.
One tasty addition to that is a "rest" rally point at some point of the queue, that way the unit wont stay in the line of fire, but return to a "safe" area when there is nothing else to R/E.
One MAJOR flaw, yes major, for the serious use of "repeat-area-repair" are construction units running behind planes... this is in the "bug" category, seriously.. i had submitted it, in old system.

BeeDee wrote:Or, instead of placing an order for each dragon tooth when fortifying a position, how about an AI that you can give a "starting" DT position and an "ending" DT position to and it automatically has the construction unit fill in an unbroken line of DTs between them? You could have similar AIs for having construction units to build a square box of DTs, or to surround an existing structure with DTs.

Hm.. er.. have you ever used "shift" and "dragging" a line of buildings?
It gets you a line of buildings of the same type!!! one aside of the other!! woot!

As for "squares", use shift+control+click to sorround a building with whatever you like.

There is even a modifier key for the "space between" rows, only for 5 button mousers. (ah.. the minefields...).

I think too, that the line between "aids" and "cheating" is thin.

And its really hard to tell how to decide that.

As a note on this, there was a guy that made a scout ai some time ago (never used, as im superb scout :) ), and he told if we wanted some more features like:
-a list of "spotted" units and buildings (shit, i know i suggested that in some forum b4 of that.. hehe).

Well.. is that cheating? having a list displayed on hud about what the scouts saw?...

All i can say is, wtf when that is possible, i want a list too!!, imagine what a usefull information does that give you in the battle field!.
Its certainly an innovation that is worth having.

BUT... this can affect balance of the game, giving scouts a value they didnt have previously.
So, this is matter of the mod makers too.

Switching a bit off this, and focusing more into reality:

What "features" of an existing Player AI, (to any game to the date, read WC3, CCGenerals, etc etc..), would you want included into this group ais??

e: or..

What things the AI could perform better than a good player?

Being a good player, is something any ai aims to, but hardly fullfills.

And thinking that relying in a very good ai, is like "sharing commander" with a good player, then what you think that player could be better at?

For example, a better player than me would find better attack paths to strike from.

mongus wrote:

BeeDee wrote:What about a group AI for construction units that would cause them to automatically seek out and repair damaged units within a defined area?

No need to use an ai for that.

I know, I was deploying a devastating barrage of sarcasm. I seem to have missed. :)

mongus wrote: I think too, that the line between "aids" and "cheating" is thin.

And its really hard to tell how to decide that.

This, however, was exactly my point so I guess it's okay. :) There are already lots of helpful micromanagement-reducing features in Spring that OTA didn't have, and OTA itself had some micromanagement-reducing features such as order queues that other RTSes of the day didn't have. Being able to tell a construction unit to go forth and plop down mexes wherever its tiny brain thinks is remotely useful isn't a huge leap that would dramatically change the game, IMO.

I'd be interested in more formation options. Not sure if it counts as a group ai or what. For example what about the ability to set a group of 15 fighters to fly in an A shape ahead of a group of bombers. Then have the bombers automatically pick tarrgets in a selected target area. And then have all the planes return to base for repair after the bombing run.

Or the ability to set up a group with tanks in the center, artillery behind and faster vehicles in two groups on either side. The group attempts to perform a coordinated attack ona specified area.

In a real battle the commander can give general commands to generals etc, and expect them to show a fair bit of intelligence and initiative in carrying them out. Yet the job of the commander still involves skill in setting the overall strategy, and reacting to special events instead of micromanaging the whole army. I am sure this job would be interesting to imitate in a computer game, particularly if you could capture the psychological issues of dealing with independant humans (obviously a far fetched idea for spring). Although it would require different skills from a highly micromanged click fest based on reflexes, tactics and keeping track of lots of stuff at once.

And can you imagine the other extreme? You could say that allowing the units to think for themselves about what target to shoot at is cheating. Instead of using an ai to help manage battles in that way every fire command must be given specifically.

Oh and I also like the automatic metal extractor build options. And maybe another option to set a construction unit to automatically build base defenses. I think I could make use of a level one con vehicle going around and automatically spamming out missile towers and laser towers around the edge of the base.