Smart Combat, Unit AI - Why not?

If you take that to the extreme you would need to let an AI do everything. That would lower the APM barrier so much that everyone could play . Obviously that's taking it too far. The subjective question is: How much AI is too much?
Quickly reacting to pull units back is something done quite a lot while play FA. I've played thousands of 1v1 that way. Of course I don't like the idea of changing anything about it. Pulling my units back when they are in danger of dieing while standing around, kiting my opponent when I have the range advantage, all those are things that add to the "feeling" of playing FA. I like that feeling. Controlling my units like that is not tedious to me, it's fun. Totally subjective, indeed. I guess we can just wait and see what UBER will actually do with PA.

I'm resisting the urge to bombard you with arguments cola, you asked nicely and we both want the same thing: PA to succeed. The smart combat/unit AI stuff can easily be added and removed from my understanding. Based on my experience in other games I'm convinced it would objectively improve gameplay rather than damaging it.

Regardless, such smart combat needn't be forced on anyone, as Zero-K's googlefrog explained it's inferior to human micro. If it is a subjective matter as you say then would you cede this should be up to the community as a whole and not the preference of either of us? I think both of us could agree both approaches should be given a try in Alpha/Beta, what plays better and what the community prefers should factor into Uber's decision.

I think more will like smart combat than not, but I'm not dogmatic. If I'm wrong and it is in fact worse for gameplay I would change my tune. So, if you agree it's worth looking at, I think I can agree leaving it out wouldn't break the game.

If you were to throw arguments at me, I would throw some at you. We would keep discussing and in the end just stay with what we are used to. That's the result of other discussions so far. Discussions about taste are hard to win So let's just wait now what Uber is gonna do and try that out.
It's up to them to come up with something now, possibly under the influence of us and the whole community.
Trying out a lot of stuff during alpha can never hurt.

More complex unit interactions allows more room for micro too, and without better automation a large scale RTS game that's supposed to be macro oriented can't have really complex unit interactions.

Yes and no.

Yes: I feel that the stances that existed in TA won't accomplish what the OP is getting it. The stances in TA basically controlled how units left on their own reacted to being shot at, but the OP wants to take it a step further, so that units that are sent on the attack essentially micro themselves to dodge shots and kite units.

On the other hand: as I indicated in my previous post, I'm not sure that what the OP has in mind is necessarily optimal. I feel that micro should be reduced -- perhaps not as much as the OP has in mind, but certainly reduced. After all, players will own multiple planets simultaneously -- they can hardly afford to micromanage skirmishes on their own.

I'm of the opinion that combat in Planetary Annihilation should most definitely not be fully automated. That won't reduce micro at all; it will simply shift the micro from combat to base-building. The player who is best at SimCity will win, and I'm not sure that's what the devs are going for. For another, it has the potential to alienate players who enjoy maneuvering for tactical superiority.

Here is where stances come in. I feel some additional stances might smooth things over.

For example:

• Movement Mode: This stance will modify the behavior of the unit while moving
  • Normal movement: When moving, the unit will follow a direct path between Point A and Point B
  • Swerve: When moving, the unit will follow a sinusoidal path between Point A and Point B, thereby throwing off the aim of units attempting to lead the target

If you've ever seen high-level SupCom 2 play, a lot of engagements boil down to commanders frantically clicking to micromanage their armies in a zig-zag course to throw off artillery. It's pretty simple, really: artillery has a long flight time, so it aims far ahead of any unit it targets. If you're going in one direction when the artillery opens fire, all you need to do is change directions to avoid getting hit by the artillery.

This sort of micromanagement adds a large number of clicks that, while necessary, could be done away with.

I'm not sure about a stance in which units endeavor to maintain maximum range to their targets; to me, it would feel too much like the game is playing itself.

That said, perhaps the community feels differently. After all, we'll be monitoring engagement on multiple planets; at what point do we throw up our hands?

What say you?

I'm calling it a night so without responding to specifics I'll say I don't buy that. We've seen it employed in other games and using the Zero-K example, that game has developed a reputation for its emphasis on strategy and tactics.

http://springrts.com/wiki/Games

Again, I don't buy it. If you've played Spring games that employ smart combat it's usually the opposite, players are relieved of tedium and spend more time maneuvering, maneuvering different unit types and maneuvering in different ways. As was explained previously, when micro and APM aren't barriers the overall variety of unit interactions can increase.

This may be presumptive but it seems many detractors are coming from the FA perspective. In FA there is an excess of tedium both in using units and managing the economy. If we took FA and reduced the effort required to use units at a high level it would likely shift the focus further onto economic tedium, as you suggest. That however is not intrinsic of smart combat. It is a function of FA's over emphasis on economic tedium, tactical trumping, and the lack of variety in top tier units. Tactical trumping refers to a unit suddenly rendering existing units on the field obsolete, e.g. T2 PD trumping T1 Spam. PA on the other hand will have a less tedious economy, flatter technology and a greater variety of viable units. This means that the economy won't always demand your attention, units will not suddenly become obsolete, and you will have a larger variety of units and tactics at your disposal. There are major design differences between these games and what does not work for one may still work for the other.

It seems I don't have to convince you of the merit of combat AI as you have made your own suggestions. If it is a matter of degree or subjectivity then I think we can agree different approaches are worth a try in Alpha.

Dafaq is the post above me about?

Anyway, I think micro should be reduced, but not in this way. You're not reducing micro, you're merely giving it into the hands of the AI. That's not a solution that's getting around the problem, handing it to someone else. Uber has some great plans to reduce micro for sure. Size of the map will be one thing. Maybe enhancing projectile speed so that units simply won't have time to move away would be nice as well. Try dodging a bullet for example.

Right, that is the idea though. Instead of reducing the variety of unit interactions by making it difficult to dodge like Forged Alliance or simply making maps too large for non-pros to micro, smart combat mitigates the value of micro without removing interactions from the game. I would encourage others to try games where this approach has been used, regardless if you are for or against smart combat.

Looks like that post above is some kind of bot spam.

Concerning the OP, all I have to say is Yes. Just Yes.

"When a unit encounters an enemy whose fire it can jink (a swarmiee), it will automatically move in a zigzag pattern to close in, then (depending on its range) circle strafe the target to continue avoiding its fire."

As posted above, it allows you to focus on building the units and getting them in there, and not trying to keep track of every battle. Very important in an epic style game such as this where (hopefully) you'll be fighting not just multiple battles at once, but on multiple worlds as well.

Smart combat is some kind of attack management (although a pretty low level one), Sorian has said he don't that kind of things several months ago, not sure if he has changed his mind.

A dumbing down of gameplay does not make a better game. certainly not where competitive play is going to be emphasized. Perhaps in a mod, but not in the ladder matches/main game.

I'm really starting to dislike this bandwagon term. It doesn't mean anything.

Unit AI would not 'dumb down' the game. It would simply shift the meta. Instead of focusing on unit micro, you focus on other things. It may make things more accessible, but it in no way changes the skill cap. Competetive players will simply find other things to focus on to make their skill better.

EDIT: Sorry, nothing against you at all, I've just seen that term for a good 3 or so years, and it starts rubbing off after a while.

If you define turning tedium over to the UI as dumbing down, maybe. Auto-repair, auto-reclaim, attack-and-stop walk, area commands, target priorities, multi-point patrol, formations, etc. were not always part of staple RTS user interface. These are recent innovations that you have likely experienced and appreciate if you play modern RTS. Smart combat is in the same vein by automating additional types of mundane commands.

Smart combat can be particularly valuable to PA as it can shift the focus from unit engagement to battlefield management and strategy. It is also important to note the difference between engagement (e.g. swerving, swarming, jinxing, dodging, kiting) and battlefield tactics (e.g feigning, retreating, pitching, skirmishing, flanking, ambushing, delaying). Smart combat deals with the former and the player deals with the the latter. Let the game handle how individual units engage, let me handle the battles.

Furthermore, APM and micro should not dictate unit balance. If you are looking for a game where micromanagement dictates how powerful units are I would recommend StarCraft. APM barriers can manifest themselves in subtle ways and can undermine unit balance. An example of an APM barrier in Forged Alliance is the Seraphim T3 Siege Tank, it was the weakest T3 in high level play because it was easily kited and the strongest T3 in low level play. Smart combat can abate this by allowing PA to balance units around unit characteristics (e.g. range, movespeed, projectile speed) instead of attempting to balance unit characteristics around APM. Forged Alliance ultimately had to water down a lot of unit interactions like dodging and swarming to balance gameplay.

You don't have to take my word for it though you can see for yourself here.

Zero-K is a great example, it is an independently developed RTS with a reputation for emphasizing tactics and strategy.

Wouldent this be the opposite of dumbing down the game?

Dumbing down a game is this (at least according to me): You move focus from any type of intellectual parts of the game onto more action oriented things (aka the game becoming more dumb).

Making units smart and removing some (if not all) micro would (i belive) move focus from action and onto the more intellectual parts (like strategy), so i wouldent call it "A dumbing down of gameplay".


If it works as im imagining it wouldent remove control from the player, it would shift it (Instead of controlling micro you would be more focused on controlling macro).

But it would have to be tested to see if its viable (its easy to say that it would work, but we dont realy know until its been tryed).


But there are other problems with the idea, the resources required for any type of advanced AI (This can realy snowball easily, im talking about computer resources here), how to make a AI as efficient (or more) then a human at micro (havent been done yet in any game as far as i know), etc.

With the many limitations the idea may be unfeasible to try for real, at least for now.

I'd be inclined to agree garatgh. The only way I can see micro being reduced from a practical standpoint, rather than a theoretical one, is by changing the game in such a way that Microing your units provides such a minimal increase in their effective "power" that it's just not worth doing.

A high hitpoint to damage ratio achieves this goal. If a unit can "tank" the damage then there's less reason to micro them than to just build more units. Why bother saving the life of a single extra unit when I can have 50 more on the battlefield in less than 20 seconds.

This is the Smart Combat code for Zero-K, reposted earlier in the thread

Indeed. Although I disagree that micro cannot be mitigated with smart combat. I would go so far as to say it can reduce APM barriers and increase the variety of viable unit interactions. And we need different unit interactions to make things deeper than massing 1-2 units and A-moving into your opponent.

BA and Zero-K have a characteristic based balance that is the result of smart combat, varied unit interactions, and high HP to DPS ratios.

Cursory outline of unit interactions:
http://zero-k.info/Wiki/UnitClasses
Riot & Assault < Skirmishers < Raiders < Riot < Assault < Raiders
All-the-above < Static Defense < Artillery < All-the-above
*This does not include scouts, mobile bombs, anti-air/naval, support, or shield/stealth/jam/cloak

I never said that AI isn't a way micro could be mitigated... but it is the most CPU intensive way of doing things.

Unit AI is basically something that spam move commands like a really high apm player, if those starcraft micro bots with more than 15000 apm can't cause any lag in game, I don't think Unit AI would necessary be a very CPU intensive thing.

This relates to a very important point. The game mechanics and interface should be designed at the same time to complement each other. You should not complete the game mechanics and later attempt to make a powerful interface to deal with them, the interface is likely to negate some of the mechanics.

Here is an example of a set of mechanics which would be nullified by a powerful interface. In the RTS Cossacks, these two mechanics are in place:

• Players can instantly kill most units that they own (self-d without a timer).
• Some units (cannons) and structures can be captured by most infantry and cavalry units.

The smart thing to do is to destroy your own units just before they are captured by your opponent. This takes a lot of micro, it can be alleviated with smart unit AI. Such unit AI would cause the units to kill themselves just before an enemy enters the capture range. This is actually implemented for the skirmish AI for this game, it is almost impossible to capture any of the AI's units because it always kills them just before they are captured.

The unit AI has removed the micro requirement but some game mechanic has become trivial. The mechanic of captureable units has been replaced by units which die when an enemy unit is near. This is a case of the interface fighting against the mechanics and it is basically a waste of development and CPU time.

This is why it is important to think about mechanics and unit AI at the same time (that is, if you are making a game with the aim to reduce micro). Whenever you propose mechanics think about how this mechanic could be automated or bypassed by unit AI. Likewise, whenever you propose some unit AI think about whether any mechanics are made trivial. If unit AI destroys a mechanic then the mechanic was probably just a micro sink which added little strategic value. In this case consider removing the mechanic instead of writing the unit AI, it saves time.


There are a lot of non-trivial mechanics. For example units fire simulated projectiles which must fly through the air and collide with their target. This allows fast units to dodge the projectiles and this often requires micro to achieve. Say you wanted to remove this micro requirement, it is possible to make some unit AI to jink and dodge shots (I have done so). This AI will not be flawless (sometimes you cannot dodge) so there is now a certain chance for each weapon to hit each unit.

There is now a mechanic and some unit AI working with this mechanic. Maybe it would be easier to remove both and replace with a simple equation. Remove simulated projectiles and have each shot either instantly hit it's target or miss with some probability (as far as I can tell this is how most of the projectiles in Company of Heroes are handled). But this simple equation fails to capture the depth of the simulated projectiles mechanic; it is easier to hit a clump of units, it is easier to hit something moving towards you than moving sideways, close units are easier to hit than far units and jinking while moving is slower than moving straight. More importantly, players will always have to choose whether to jink or move in a straight line. Jinking is slower and safer, rushing in closes distance faster but is more dangerous.

The 'jinking units' mechanic is too deep to be replaced by a simpler mechanic or equation. Any equation which captures this depth would be a massive monster of an equation and it would be very unintuitive. It is much easier for a player to see their robots jinking and dodging actual simulated projectiles. So at least in this case I think the correct way to reduce micro is to add jinking unit AI.


Just to be clear. This is not an argument for reducing micro. I assume you want to reduce micro and argue that adding unit AI is often a good way to do this.

It is also good to realise that unit AI is not the only way to reduce micro. There is an equally powerful tool; not adding the micro requiring mechanic in the first place.

Have people here played Warzone 2100?

If not, it can be downloaded for free, from WZ2100.net

This has just the right amount of unit AI in it IMO.

The only real difference it has from TA is that units can be told to retreat after taking a certain amount of damage.

A great system for probing enemy defences no doubt, even better for setting attacks going while you manage another front, or make a cup of tea.

Really I'd recommend everyone giving it an hour or two of their time then coming back to this topic!