Micro, AI, and the third resource: Initiative

Actually, they’re not the only problems. Game designers not knowing how a game will/can be played is also a huge impediment.

As already stated, they sucked.

You are over complicating the idea and the UI by giving players a way to impart some situational awareness on the units. This is the algorithm, it is only implemented per-unit. Note that the only difference between A-move and this is the addition of the third case.

• If there are no opponents in rage -> Move towards goal specified by location of command.
• If the closest opponent is just within range -> Stop moving.
• If the closest opponent is more than 'just' within range -> Move directly away from closest opponent.

This AI is very simple. There is little situational awareness. It will work most of the time but does not deal with unexpected situations. Players should be the ones which have to look out for these situations because they require complicated quick decisions to deal with.

I don't think we disagree. I think we have independent aims for unit AI which can be summed up as:

• I want good unit AI.
• You don't want bad unit AI.

Yes, this is all technically off topic. We seem to agree about bad unit AI. I just like using the opportunity to advocate good unit AI as well. Anyway, this is more interesting than discussing the unit AI specific to this thread.

IF thats true be more specific then. You claim you want units to do the mundane micro things but you give no examples. It's a lot more complicated an issue than what you seem to think it is. As demonstrated by PKC and googlefrogs discussion. Much of this we've covered in other threads also, there has been no consensus on what level of interference from an automated unit manager is ok.

Eh, most of it's be covered, but I'll summarise my view on matters.

1: No command I give should be overridden by the AI, EVER - if I send my units to a rally point, they should stay there, not run off to fight stupidly.
2: We should be given more commands - or better yet, be able to set up our own commands based on a few rules. eg, move to location, attacking units if in range, but do not chase or stop.
3: Buildings such as mass fabricators should have auto-toggle their on/off state to maintain positive energy.
4: Engineers should perform tasks 'within range', that do not negatively effect economy, without having to be told. eg, reclaiming - if there is room to store (assuming that's a mechanic, and not similar to SC2 where there was no resource cap)
5: I should be able to have units alert me to attacks or units - especially scouts. The game is strategic, not tactical, and I'm a commander, not a soldier, so my army should act like an army.

[

I think the problem is you may be so used to arguing about AI that you did not read my post fully. Just gave the boiler plate dismissal as the way with Forums. No biggy I am used to this. You could have ask for clarification if it was unclear or wanted examples. . .

Also, I have just Joined the Forum and commented on a thread that I had something to contribute to. I did not read all other threads or check the history of this subject (or should be expected to) As this thread was the most resent, live and had a unique bend to it.

As extrodity summarized and I completely endorse. This is what I am game for.

No AI playing the Game for you, no AI running around of when you want it to stay put.
But...

• Enough AI to automate mundane tasks IF wanted.
  Enough AI to automate simple combat if you can't attend - again IF wanted (a so-so AI is fine here as it is better then nothing but not as good as you being there). We have planes following paths and dog fighting already, we have a skirmish AI that can give a normal player a bit of a good tactical fight, why not use and expand this??
  Enough AI that you do not have to fight just to manage your economy.
  Enough AI that you can send a scout off to look for a enemy and play keep away. (again optional)
  Etc Etc

TA and SupCom added allot to most RTS game already, like automated repair, complex Patrols (even FROM the factory), Automated transport waypoints, Some unit response . .etc . . it is some of what makes these games different from the normal faire.

I am at a loss why people are so ageist expanding this a bit? and go out of there way to give crazy/extreme reasons why it "can't" work "the AI will run amuck" "then people will just use the AI insted" "it is too hard" etc when there are simplistic answers to each?

Micro, AI, and cost: Refocusing the discussion

The goal of P.A. being "The ultimate Macro game" I think any move toward allowing AI unit control is a good thing. Whether the AI is good or bad, it is going to be present in some form. What I'm suggesting here is that the player should be given access to the in-game AI as an in-game resource. This would allow un-skilled players to enjoy the macro tasks of commanding massive strikes and building huge bases, as well as reward skilled players who do not rely on the aid of the in-game AI.
So, as far as implementation:

• It should be totally voluntary (temporarily turning control of units over to the AI) to prevent the AI running rampant.
• It should be penalized in-game (cost resources and be vulnerable to attack) to reward skilled players who don't need it.
• It should be tune-able and informed by player's goals. Some way should exist to easily tell the AI "here is the base, here is our main attack vector, here are our flanks" and so on, as well as specifying specific strategies.

But figuring out all those details is the developer's job, and none of us will be able to tell them the best way to do it. What I'm asking is that we be given access to the in-game AI, and be made to pay for it in-game to reward skilled players.

Without going into the whole resource thing or how possible things are; I like the concept of smarter units. The main threat when dealing with a macro game is stupid units. It's also the primary reason I generally fell back to using only aircraft in TA; at least they didn't get stuck behind one another due to poor pathfinding.

You want units to do what you want; that's a given. The AI already does the small thing anyway. We don't have to manually guide our units through difficult terrain. We don't have to tell them to shoot something if there's units in range. We can simply select 10 units and order them to go somewhere and reasonably expect them to arrive there after a while. The AI shouldn't be making decisions for you, it should be following your orders. What we'd need then is not an AI that automates stuff; it's highly complex orders given easily.

How about a "mass repair" order, where you simply drag a rectangle on the playing field and everything inside the area sends up a "fix me" signal, whereas a second "designate as repair unit" dragged over a group of engineers toggles all of them to repair mode, making them search out buildings marked as "fix me" in their vicinity?
How about a "mass factories" order where you simply drag an area, and any engineers designated "build unit" will go over there and start filling the area up with all sorts of factories?
Or a "mass units" order that will simply make factories produce stuff? Or, even more advanced, a macro order where you can predefine priorities on units types (ie; 60% bombers + 30% fighters + 10% construction craft) that makes all selected factories start making these things in said ratios?
Or perhaps "mass transport" which will make all selected units look for some form of transport unit in the designated area and load up.

Then add in some order markers such as "designate fallbackpoint" that allow you to place a giant marker on the battlefield where units will fall back to, and then a "retreat" call which will make any unit within X area move back to the fallback point.

Then you can also add a rally point marker on the field, allowing you to attach a factory's output to it. Now you can drag a single, visible object around and it will update the rally point for each attached factory without you having to select it. Likewise, you could have an "assault position" marker that makes units move closer and engage nearby targets.

None of these ideas are terribly complex to implement, all of them are pretty straight forward as to how they will be executed, and they will allow you to manage far larger armies then you could before. This is not a micro game. So why would we want to stick to micro commands?

With a proper interface, the player wouldn't need to worry about turning over any high-level control to AI in order to make the game manageable. Consider a Starcraft player being told the kinds of things they'd have to manage in Supcom WITHOUT mentioning strategic zoom, order queuing, etc. In their eyes, with the UI they consider to be the standard, the tasks they'd be expected to do would be enormous, and they would consider Supcom to be "too big to play properly." But those of us who have played and know what tools are at our disposal already know that the game is very manageable.

This is the same for PA. We only know minimal details about the UI and how it will work together, so managing such a big game seems like a monumental task. But from all the UI features I've seen being discussed, I suspect that managing these large armies will be very natural, without the need for helper AIs.

However, on the lowest level, with regard to unit micro (such as kiting, swarming, etc), I feel that AI would work just fine, and in some cases be desirable. Of course, these reactions should be set by the player, but Zero K does provide a nice example of how this system can work well with minimal loss of player control.

I think we're saying the same thing at this point. You are calling it "proper interface" and I'm calling it "helper AI" but it comes down to the same thing: Programatic abstraction of micromanagement in order to ease the burden of command on the player. No one wants to have to manually type in x,y coordinates for movement orders, or manually do pathfinding. We agree that there is some minimum amount of AI that should always be present.

But I'm not saying that we should raise the minimum level. I am suggesting that it would be helpful to increase the maximum amount of AI control available in-game, and for an in-game cost. At least one person (myself) would like to be able to give control to the AI, up to and including the total strategic level. I might want to go make a sandwich, or have a nap, and see where the AI takes my base design and forces. I may want to just watch an asteroid's path around the sun while the AI manages the war. I'm saying make this the maximum (not minimum) level of control that I can give to the AI. And, since (I think) this would give a significant in-game advantage to a player who used it, make it cost in-game resources.

As I said before, if you're better than the AI, great. Good on you. Most of the time, I find that I am barely on par with the AI, and would like to have it fighting on my side. I constantly found myself frustrated in TA, SC, and SCII with the amount of fiddling required to set up, build, and manage the war effort. Yes, it is very possible for skilled players. But the AI can do it all by itself, and I want to be able to ask the AI to do the job while I enjoy the parts of the game I find enjoyable... Like watching things explode.

This is not at all what I'm saying in any way, shape, or form. I am definitely opposed to the idea of AI giving any sort of commands beyond that of micro. Anything on a tactical or strategic level should be left exclusively to player decision, unless the player is playing a team game with an allied AI controlling the same army.

Ok, how about this? I want the ability to have an AI teammate (during "single player" or "multi player" mode, it makes no difference) (embodied in a non-commander unit or building, since commanders are pretty powerful), to which I can issue commands and give authority over portions or the entirety of my forces. In this sense, we are still basically suggesting the same thing. Actually, this second player could be an AI or another human subordinate player, since this seems to be the planned model for multi player team matches anyhow, especially with seamless connection drop/resume architecture.

The only thing I would add is some form of in-game resource consumption for the privilege of having two (or more, why not?) "players" under my control and/or on my team. If you think there should be no cost for adding multiple players and easing the burden of command, great! I'll gladly take the aid of AI or human teammates for free if I can get it. This just seems unfair in some sense, even if these separate players don't add more commander units to the fray.

That line of thought actually opens up another. If AI players have various "values" such as easy, and hard, why not also rank players using this same system? The IN cost of a player could be linked to their battle record... which of course would tend to reward grief accounts etc. Eh, just a thought. Also, if there is a command hierarchy, then it would be possible to have an AI at the top tier, with multiple human subordinates playing out the details. Hmm... maybe I'm over-thinking this.

Re: Micro, AI, and cost: Refocusing the discussion

I think this is a really bad idea. Do not put an ingame price on smart units.

When I say AI ally, I only mean it in the sense that in TA/FA you could put an AI on your team of 2, and your opponent could be a team of an AI and a live opponent. As for them controlling the same army as you, sure, if Uber allows that type of thing in their plan for multiple players to control the same army. Not sure I'd want it, as I'm pretty sure I could play better than that AI, but hey, whatever floats your boat.

But it should not be a gameplay mechanic.

I like what dudecon is saying (along with the other normal UI and optional helper enhancements)
Effectively allowing a AI or human helper. (human coop seems to be in?)
Now before people start freaking out. Most of you where saying the AI sucks so I do not want the same people saying that it will some how be unfair now :/

If done well it is just delegating some work (quite a few games have modes where you can delegate local control to a general/governor). You can always over ride the local control.

Remember, there is going to be a game mode where you may have to keep track of more then one planet or even system of planets. A bit of local control and the ability of them to "yell for help" ie direct control may even be required given the new scale.

This could allow the AI to do mundane things and even do some basic defending when given lease to. Remember this could be a mixed bag (just like having a human partner) Both may not develop/defend the area as well or in the manor you may like BUT it is better then the base/planet just sitting there scratching.
I know we played with things like this in SupCom. "shake-en-bake" fire bases where you give an order and pick a layout and the builders make a base for you . . some enhancements may even be build a factory with orders to build a small defense army. Again this is less then idea in all cases but this may be ok given the time saved having to build every last bit (again which you could if you wanted to)

Again all you uber multi player gurus that only goal is to smash nubs . . should not have a problem with this since the AI is Soooo dumb and you all are so uber. And since it would be optional I could see people using or not as needed later in the game when the scale of the map and action is getting intense.

Still optional, still easy to empliment as it will use AI code that exists, just needs UI and scope tweaking.

Any constructive criticism?

I have read quite a few AI threads here and it seems that there is always people that just are straight on against any AI helpers.

It seems that it doesn't help if people say that it's optional because if it's in the game you somehow have to use it even if it's optional.

I'm also one of those people that really can't do the APM thing. And for me it would really be a wonderful boon to have an AI to help me do the mandatory things.

I understand that the pro players want to play games that don't use these helpers so it really depends only on player skill and not an AI that helps/hampers you.

So why it's so damn hard to allow these helpers if there would be an option to either not use them or just disable them all together when you create a game.

edit: not a native english speaker. semidrunk at the time of writing.

TBH my only concern is we get QoL stuff like engineer patrols right. It is annoying when you want an engineer to only collect resources and it instead decides to repair stuff.

Another QoL enhancement I'd like is Dwarf Fortress style zoning. I should be able to specify a zone for resource collection. Maybe with a specified patrol route. Maybe with an AI generated route (though this AI route will probably be bad). Then I should be able to attach engineers to it and take them from it. With the engineers automatically adopting the settings of the control zone I created. That way I can control zone the region behind my front lines as a resource collection area and just let the engineers get on with it.

Then I can alter and advance that control zone as more areas become filled with carcasses and other areas clear.

This stuff is currently ad hoc in games. This is an interface issue rather than an AI one.