Defining micro

But that's the main point of his post, units should do those simple common sense stuffs themselves, like they should attack enemy units in range without the player telling them to do so.

There is a balance between micro and macro in every rts games, the lack of automation would make its balance more lean to micro. If the game reward good strategies more, it would divide playing styles with strategies more.

this isnt necessarily true, depending on what automation we're talking. automated micro of, say, unit tactical positioning could actually increase the balance towards rewarding player micro.

Not sure if this is what you meant, but one thing that players always manage to do in online games is find exploits of one sort or another, and make the game behave in ways that weren't intended.

Obviously it depends on how it was implemented, but I could see for example auto-kite behaviour being used to cheaply push enemy units into heavily-defended areas. A bit of distraction elsewhere on the map and it would be easy to for example make a valuable detachment of rocket bots run straight into some turrets without the player noticing until it was too late.

Because AI can never be properly emergent (nor would you want it to be in this case), it will have blind spots and players will find and exploit those blind spots.

somehow I double posted in between my posts!!

Just like in chase AI behavior where units never chase the enemy beyond a certain radius, kite AI behavior should work similarly. But within those constraints if you can abuse the system, I'm all for it; in fact, that's how the system should work. Hell, the same basic idea works just as well against people as it does against AI.

An interesting point. Though I think this is possibly a real life tactic that could fairly be employed. These units are following orders and trying to avoid fire. If you used fire to push enemies into a kill zone I would call that tactical and valid.

Also, it's up to the player again to say do or don't take that course of action. If I told them to kite I would expect them to keep kiting until they were dead or I tell them to stop. This is exactly what people are talking about when they say they don't want the game to play itself.

The main difference I see here is not the action taken but the way in which the action is issued. If I issue it that's fine. If the unit makes the decision for itself then that is not OK.

I disagree. I think they should do them if I tell them to do so. For example if that unit has been issued a hold/return fire command, then I expect it to carry out that command until I issue another one. Likewise, if I say fire at will then I expect it to shoot. Simple things that are common in RTS games that have evolved over the years provide these functions adequately. I see kite and strafe actions as a natural progression of this and it makes perfect sense to me.

Games like DoW2 have other commands, like engage at range. Engage at range>close range (using ranged weapon)>engage in melee, close range asap (no ranged weapon)>engage in melee.

All things that could easily be included in this game without too much trouble.

To address your concerns, wolfdogg, in Zero K the unit ai for kiting/strafing has to be turned on in order for them to actually do this automatically. Retreating to a retreat zone is very similar, except you can also specify how much damage the unit has to take in order to retreat. So you can tell a unit to only retreat if they've taken 90% damage, for instance. This is much like how unit stances are set up in TA, and since the devs have implied that something similar likely be implemented in PA, it wouldn't be too much of a stretch to implement the ai and retreat controls.

On top of that, in order for the units in ZK to use the combat AI, you had to issue a specific command, the "fight" command. So even with the AI on, the units only do as the commander orders them to. I think this is an important aspect that I probably should have mentioned earlier.

Of course, the devs will need to make sure that all these features don't result in UI clutter, but considering the FA UI, I think we'll be fine on that front.

I feel much the same way about FA. I keep picking up a skirmish game now and then, starting to understand the mechanics, but I just can't get to where I'm actually enjoying it. Mostly because of the scale and teeny tiny ant units/icons. I've given it substantially more chance than 10 minutes though... several hours at least.

Would having a big menu of potential actions for a unit to take be just as bad as StarCraft-style micro? Presumably each unit would have a default behaviour to avoid this, but how would you decide what the default behaviour should be?

What if the default tank behaviour fitted your overall plan, but the default missile launcher behaviour did not, and so you had to manually adjust each missile launcher? That would inevitably favour one tactic over another, would make one way the de facto standard way to do things.

Meh, I think it can be done simply with multi-toggle buttons. It's been done enough in other games successfully. You don't need to see all the buttons all of the time. Especially when they don't apply to the selected unit/units.

That's what control groups are for and commands like Ctrl+Z etc.

EDITED: People have encountered and solved most of these problems a long time ago. Those commands were from TA for example.

yep, that's precisely what i meant. there will always be circumstances when automated kiting will be taken advantage of by a good player, like your example of tricking an AI into moving troops into a disadvantageous position.

i dont think anyone is disagreeing units should shoot at everything in range automatically. i was referring though to suggestions that units should automatically hide/retreat/kite etc, because they will never do those movements as well as if i were commanding them HOW to do them. players that rely on automated systems like this will get beaten by players that do it themselves, (RTS) games are too nuanced to teach AI how to play properly tactically.

That's only true for high apm players, and only true for small amount of units in one or two fronts.
Pathfinding in some old rts games are worse than manual pathfinding, so you think those games better not have pathfinding at all?
Some simple tasks like kiting aren't very tactical, even monsters in diablo 3 can kite well.

Wrong on both counts. You clearly don’t understand the nuances of troop micro or how hard it would be to get the AI to do it effectively versus a human opponent. SO many decisions like this are a line judgement call, something which is very difficult to teach an AI to do. Check out the AI in supcom for instance. Just terrible.

Again, you’re not thinking completely enough and a comparison to diablo is absurd because the AI only has 1 “unit” to consider. All you guys that favour deep levels of automation in regards to micro simply think “oh, in instant X my units should do Y”, without considering the larger issues at play. Action Y could lead your units into an easily-laid trap for instance. It is very difficult to teach AI to do the right thing at the right time. And even if you could, it would become so easily predictable that half-decent players will anticipate and wipe you off the map. You’re the same players that after release would start to complain that your unit/army AI doesn’t act intelligently enough in certain situations. Yeah, no kidding!

Again, I have to point to Zero K on this. Despite the fact that the units have kiting/strafing AI, they do NOT get drawn into traps. This is due to the fact that they ONLY kite/strafe when you give them the "fight" order. This order is essentially an over-glorified attack-move order, so had you given the order as an ordinary a-move (or, in the style of TA/Supcom, an ordinary move order), you would have already been sending your units into a trap in the first place. The difference is that now, if they find themselves walking right into a previously unseen laser cannon, they'll be smart enough to pull back out of its range and attack, or retreat from it. The end result is that it's MORE likely your units will survive some sort of trap, especially since they'll be smart enough to run away without your command (provided they are allowed to).

Conversely, if you WANT to send your units into the trap, perhaps as a distraction, you can disable the attack/retreat AI and do an ordinary a-move.

And for goodness sake, at least try out the current implementation, if only from a "mock up" standpoint. Then argue from the position of having actually tried out a working model.

that was my point, that’s where the trap would be, not at the previously unseen cannon but at the units’ retreat point.

Yeah that’s fair enough, but im not grabbing a game just so I can see an AI micro troops worse than I can do myself. If people want it, and as long as it’s optional, then that’s fine. I look forward to smashing the hand-held noobs

As a long-time player of RTS games, I personally feel that PA should be a Real-Time Strategy game....


Starcraft, DoW 1and 2, CoH always felt like Real-Time Tactical games to me. The scope of the game was small, units tended to be more valuable and took longer to produce and direct unit control and placement was essential to whether you won or not.

Basically, RTS should be mainly about the War, not the Battles.

I want to be a Commander, the exact details of how a unit survived or died are irrelevant in the scheme of things. Units should have some degree of autonomy, micromanagement of units is a nice idea, but I feel more suited to a RTT than RTS game. As a commander of any army, you would expect a degree of initiative on the part of your troops. Otherwise we're talking a situation where every single bullet fired requires an explicit order from on-high to do so. At the other end of the spectrum; I don't expect a unit to have a personal life and bug me about certain food requirements :lol:

At the very least we should have units that play to their strengths, as has been discussed at length in this thread. Small units that maximise speed and agility, larger ones that have greater range that avoid close contact but attack bogies on the radar etc etc. If it bothers purists so much, just allow the unit-AI to be a toggled condition on game creation.

Strategy should be about logistics, economy management, intelligence gathering and countermeasures, base layout and construction, preferences of unit types and what combinations suit a particular theatre of war or to counter enemy strategies and lastly, broadly defined unit placement. Given the context, and that each unit we would make is a robot with its own AI, lets try for a little realism and imagine they have more than just an onboard Arduino...

Regarding player micro vs. automation:

Of course player micro will always be better. No one expects otherwise.

The point is that PA will potentially involve battles among multiple fronts on multiple planets. No player could pay full attention to everything that's going on.

Without automation this means that any unit that requires some basic micro to be useful will not be worth building, unless it has some small level of initiative.

The strategic question then becomes: Where should I concentrate my superior micro and where should I allow unit automation to control my units?

High APM and multiple monitor players will still have an advantage in microing multiple fronts, but they always will, automation won't change that.

Another reason for automation is that by varying the way different units are automated, unit behaviour becomes a characteristic that factors into whether or not to build that unit to meet a certain situation/needs.

I ran a test in Zero-K, using the fight command with automation, 30 glaives can kill 30 rockos with 18 glaives left, using the fight command without automation, all glaives got killed and there were still 20 rockos.
I doubt any normal level manual micro can have such a impact on the outcome, and the units in PA are supposed to be a lot more than 30 and in multiple planets.

Exactly. Try pulling off some fancy stutter-step against that. It won't work very well, and as far as I'm concerned, that's a good thing. In a grand strategy game, apm should make absolutely no difference (after a certain point), and micro should be a figment of the player's imagination. What should matter most is defining the high-level strategy of the game.

If you look at games like War in the Pacific, Hearts of Iron III, Unity of Command, and my favorite example, the Close Combat series, none of them have any micro, and APM makes absolutely no difference. (Indeed, some are turn-based, but . . . technicalities ) But they are some of the best definitions of strategic gaming you'll find. (And tactical, in some cases.)

Now PA is definitely a different type of game than the wargames I mentioned, but that doesn't mean it should be made to rely on the stuff that waters down games like Starcraft, making it a game that relies more on physical skill and muscle memory as it does on actual strategic thinking. Instead it should evolve the genre and push it towards what RTSs should have been about in the first place.