Defining micro

Oh yes, of course. I did always intend that units should have a "unit AI on/off" unit state. Not having that would be bad.

Let's say I have a bunch of hammers in ZK. The battlelines have been drawn and neither me nor my enemy are willing to do a full attack just yet. I have radar coverage of my enemy's lines. It is too early in the game for the enemy to have defenses that outrange hammers.

Now, here is where the auto-kite AI comes into play. I order the bunch of hammers to fight somewhere to the front of enemy lines. They automatically move forward and attack radar dots when they reach maximum range. They stay there and out of enemy defense ranges. Good!

I leave my lines to micro manage ticks to support an ally. This goes well, but ticks are really high maintenance so I really cannot take my eyes off them for even one second. (they're a suicide EMP bomb with large AoE, friendly fire and stationary cloak, so "one mistake and bye bye army" applies to both me and my enemy)
After a bit, I come back to my lines and set some new hammers from repeat factory build to join the artillery. I notice the enemy has tried to send light raiders to attack my hammers, but the unit AI skirmed the raiders and the hammers got behind my light defences before his raiders could reach them.
More good!

I go back to my ticks. After all, it's all quiet on my front. I shouldn't be required to do the almost trivial job of "hammers two steps foward, tower down, one more step forward... ok, ten steps back now!"

That's the kind of situations I see autoskirm working in.

I don't generally use autoskirm when I am concentrating on a heavy push. Like if I decided I had enough forces, I would send my Zeuses forward, followed by warriors and glaives, with hammers trailing behind. No autoskirm, just plain old move.


I do not want PA to be a game where one has to do those things. I see them as almost trivial and the overall movement (like deciding to help an ally, Square law and all) is far more interesting.

Babysitter is a bad figure of speech for this, it's like saying the commanders in real world are the babysitters of their soldiers.

Personally, I find it is a bit strange that Cola_Colin argue against the automations but support the order delays, they both make the fast reactions (which he think are essential for the rts games) less important, and the order delays achieve this by limiting the controls of the player, which is a harder way.

Limiting the game mechanisms might take away some potentials that are more interesting than the repetitive controls that can be performed by some short scripts.
For example, cheap suicide bombs would need some micro to avoid chain explosions when attacking their targets, they won't fit a strategic rts game well without some auto micro.

As far as I'm concerned, in a nutshell:

You should have to pay attention to your troops, and AI should never decide for you what to do with them. If you leave a group of units to idle they shouldn't do much anything, there's a notification for you if they are attacked or such, so you can tell them to attack, withdraw, or whatever else.
But making your units do what you want them to do, should be relatively easy - for that tools should be top notch, that making them do something is fast as possible. And follow my orders precisely without trying to be "smart" about it.


This stuff about setting huge contingencies and retreat zones and whatever seems really awful and counterproductive to me. You spend time and assume what will happen and what will be needed in advance, just so you can neglect them from your attention, and then afterwards wonder what exactly happened in the battle.



And honestly if you're not aware of what goes on everywhere on the battlefield, your strategy suffers too. So don't try to give any crap about wanting a game where good players don't have to keep an eye on minimap or whatever.




TLDR:
Requirement for attention - good, requirement for manual dexterity - not so much. I want to directly control mostly everything but I want that to be as convenient as possible.
Convenience is king but automation is not convenient.

So what do you think then of attack-move? This is one of the most basic examples of automation in RTS. You are giving control over to the unit AI to decide when the unit fires or moves.

Of course, the conditions are very simple and transparent, and there are only two possible options for the AI to chose from. Even Starcraft has this: "If no enemy is in range, then move. If an enemy is in range, then stop and shoot." And while this particular automation isn't really necessary in TA-esque games as it is in the stop-and-shoot games like Starcraft, it's been an RTS staple feature for over a decade.

Something like retreat zones is more complicated, but not fundamentally different. Say you can set a zone, and set your units to retreat below a certain HP threshold. The unit AI will decide what to do based on your parameters. If a unit is above the threshold, then continue. If a unit falls below that threshold, move to nearest retreat zone and wait until repaired, then return.

In each case you are not giving it purely a command, but rather setting a behavior.

So, is attack-move a bad feature, or is automation not bad in itself but at some point simply becomes excessive? And if the latter, then what are the distinguishing characteristics of excessive automation?

It's not as simple as you make out. Does it stop and shoot, does it resume moving once there are no targets? Does it continue to move to where you told it to go while shooting? Does it depend upon the unit type? Is the stopping to shoot the only difference between attack move and move? If something is told to move does it allow units to shoot the enemy on the way to the destination?

While we're on the topic retreat zones seem like needless complication. People talk about the gigantic scale this game will have, and how they can't possibly manage it all themselves, yet they want individual units to be repaired easily. Just build more units, if neither side has access to this feature its the same as both having it. It's just more UI to manage that could be better spent elsewhere. Not to mention needing to manage repairers for retreat zones, it's only gonna add more mindless tasks to do not less. The theory sounds like a good idea, until you actually examine the consequences. How is it compelling to have a percentage of units repaired for some unknown additional economic cost at a point when you could just spend that on additional units for the meatgrinder? I'm not seeing the appeal.

While I agree with your most of your points, this one bugs me not cause it's false, but the way you have worded things gives the impression that you hold a bit a grudge against SC, by disregarding in that this case is over a decade old, even though you don't mention the flaws of the other games from the past you mentioned; you could have pointed out that say red alert 1 wouldn't let you cue units in the build cue? Hell you could only produce out of one barracks at a time. But I'm going to hold it against it, even though red alert 2 address many interface issues of previous titles

My point is, Starcraft has a sequel that is a couples years old now and it address many the problems that are still present in the original (not to say SC2 is flawless). But eh, petty issues are for me :roll:

Heh, dune is ancient, though you don't seem upset by any flaws it had.

As for the whole thread, I'd more simply separate things into a few categories.

Micro: Unit control. From precise movements to accurately casting abilities. Different games use different mechanisms to increase micro requirements - compare DotA to Supreme Commander, both have micro, but very different flavours.

Macro: dealing with economy, building your base efficiently/effectively and management of unit production.

Then the would be a subcategory for each of Micro and Macro, listed as frustrating mechanics and interface flaws. That's where most of what the OP makes comparisons to; things that we aren't going to enjoy.

Because then it becomes a game of attrition, or who has the bigger economy, who can throw more units out. With retreat zones, you can save units that would otherwise be uselessly tossed. Similarly to manually retreating, you're not always going to want to keep throwing your units in, when you can use them better elsewhere.
The difference here is that you're ordering your units to fall back by proxy. This is great for smaller skirmishes where you may not be available to watch your units' every move. And if it's a big battle that you don't want them to retreat from, just disable the retreat order.

As for repairing, in Zero K the units automatically healed up when in a retreat zone. No extra engineer units needed. (Although, if you wanted them to heal faster, you could always assign one. In PA, a repair turret would fulfill the role nicely too.)

Starcraft attack-move is exactly as simple as I make out. Yes you can complicate it, but you can also endlessly complicate the process to make a peanut-butter sandwich if you so desired.

It is counter-productive argument for you anyway. If you want to make a contrast to why retreat-zones are too complicated to be a good feature, you shouldn't try to make a staple feature also sound complicated.

These are the starcraft community's definitions but I'm not terribly fond of them myself. I think they are trying to describe the difference between an attention investment with short-term payoff, and an attention investment with long-term payoff.

I feel the "truer" definitions, based on the prefix definitions and business applications for the terms, could be:

Micromanagement: Characterized by close attention to and tight control of details. It can be thought of as a 'close' or 'low-level' view of the game. Mostly includes simple, direct and specific actions. I.E. Move, Attack

Macromanagement: Characterized by broad attention, ignoring details and focusing more on the large picture of how parts work together toward the final goal. It can be though of a 'wide' or 'high-level' view of the game. Involves more complicated and abstract decisions. I.E. The hypothetical retreat zones

I wouldn't say these describe two distinct seperate styles of action but rather the two extremes of a gradient. Most of your actions will fall somewhere between the two. For example, attack-move isn't a pure simple command, but somewhere a little along the line more toward macro.

It won't make a bit of difference to the war of attrition, if you can set retreat zones so can your opponent. For it to be a benefit to the player with less economy it has to have a disproportionate effect between the two players. Something I cannot see how would work.

That would only be the case in a symmetrical battle, peewees vs peewees, let's say. If a group of artillery bots were attacked by skirmishers, then it would be in their advantage to retreat until you dealt with the skirmishers, or they moved elsewhere. I'm sure one could argue that this would be offset in the long run by a similar situation with the opponent's artillery bots, but since this is done in different points in time, it has a very different effect.

One can't disregard the value of a good retreat. I'm sure when playing TA or FA, you have retreated your units from disadvantageous engagements on several occasions, in order to preserve them for later battles. This would be the same thing, except for those battles where were elsewhere.

The "order delays" make it harder to micro units like you do in SC2. You still have to move your units by yourself at all times. With some kind of automation you might just give on AI-command and afterwards ignore the units more or less. Thats quite a difference.

Then why you don't oppose order queuing and patrol too? They both make you ignore the units more, I know they are commands, but like you said, complicated automations can be implemented as commmands too.
Anything that reduce micro will give the player more chances to ignore the units, maybe just for 1 or 2 seconds, maybe for longer.

Because those dont take away the need for the player to constantly watch his units. You may queue up alot of stuff and end up cancelling it all, since something came up that wasnt planned. Thats quite a difference to some unit-ai that deals with everything on its own.

We are not suggesting a unit AI that deals with everything on it's own.
We are suggesting automization to further enhance the capabilities of strategic decisions and overview.

Say we have unit 'Tank' that will beat unit 'Kite' if it can get in range. Unit Kite however is faster and got longer range than unit Tank. With player micromanagement Kite beat Tank.
The Kite player might be able to counter Tanks' at 1 front but 2, 3 or even 4 fronts? At some point Kite will stop being a counter to Tanks because the player attention and micro requirement is too large. There is also a risk that the Kite player will lose battlefield overview and have to prioritize other tasks that have to be done rather than kiting.
You might say this is okey because the Tank player simply outplayed the Kite player as he forces the Kite player to fight on several fronts.
Personally I think kiting is trivial micro that shouldn't be forced on the player and you still have to pay attention to the Kite as it might run into cliffs, get sourrounded or otherwise wander off into unfavourable positions.

How much micromanagement and attention that is needed to keep kiting can be balanced with unit acceleration, range disparity(if the range difference is low the micromanagement requirement is high), turn rate and speed for example.
If kiting is automated the balance can be much tighter between Kite and Tank as larger numbers and more fronts doesn't necessarily require more player input.

There is also the possibility to exclude kiting mechanics and balance the game in other ways.

Nobody here said the unit ai could be perfect and should make every tactical decisions.
In a rts game, sometimes you might set some units to patrol, then move the camera to another place until you hear some sounds of the weapons or notice some red points just appeared on the minimap, if that count as constantly watching, then set a AI behavior for your units until you need to change it is constantly watching too, you need much less attention for this kind of watching.

This, it is okay and the whole point of units like that. Since we will only have one faction in PA you will be able to chose: Am I good enough to use the kite-units? Or should I rather use some close-combat-tanks so I dont get outplayed? That sounds like a very interesting decision to make to me.

I might end up constantly watching the ai. The question is: how good is the ai?
Either it is very bad and you practically have to change so much of its commands that it is more work than just direct commands, or it is very good and you end up only watching the ai play the game, which isnt the point of the game.
Finding a line between these two extreme is very hard and not worth it.

You can't generalise like that. We are not talking about a general AI but unit AI.
No. It is not hard. Retreat automization and auto-skirmishing(kiting) in Zero-K are perfect examples of this.
If Kite as in the previous example automatically skirmish(kite) the experienced Tank, player will not engage the Kites since he knows Kites counter Tanks if they have room to skirmish. He will have to find another way to counter them like forcing them to near cliffs where they can't run, bombing them or using other units.

Once you know how the auto-skirmishing works you don't need to watch Kites automatically skirmish the Tanks. You can focus on other tasks like preventing your weak Kites from getting bombed. That is more strategic I'd say.

Then the ranged units would become less and less useful when the the amount of units become more and more, so late game in a big map might has less variety of units, and I don't like that.

It is good as long as it reduce the punishment for not paying attention to the micro, so the player can invest more attention to some more interesting stuffs, like the tactics in the battles or the bigger picture.
If it is really good, you would only need to make some tactical decisions for your units in the most cases, like the decisions about attacking priorities, retreat, timing and formations etc, those things don't need to be changed frequently in a battle.

The tactics in the battle and the bigger picture is exactly == make sure that my army moves intelligent == giving move commands as soon as they are needed, which can be frequently. micro means "evade shots by moving funny" That's something that nobody does with bigger armies, so why add any automation for it, it isnt needed anyway. But the constant move forward, move backward, hesitate to attack, retreat, attack, move right, move left, split into groups that I do with my whole army while playing is the whole essence of controlling the units. That's the interesting stuff players should put their attention to. That's not low level micro. That's the whole army control. Automating it would be very wrong.

But that's wrong. You have to frequently change that stuff. Or if you don't have to, the battles in the game are simply boring.

Maneuvering troops as the battle unfolds = fun. But just giving them a predetermined plan and then letting them have at it is really not awesome.