Okay, I know this point has been beaten to death before in this forum, but I thought I''d share this last post from my other thread in here. I wanted to discuss the issue of micromanagement in RTS games and why some people want it, versus the concept of leading troops. Perhaps a better word for leadership would have been, delegation. That''s my point really...the best solution to easing micromanagement is to delegate alot of actions to AI objects which in turn handle certain events. The player will merely act as a sort of guide or choreagrapher rather than the boss that looks over your shoulder and tells you that you are doing it all wrong and you have to do it like this. From my previous post: In a way, you hit the nail on the head of one of the most controversial aspects of RTS games, namely leadership vs. micromanaging. In one camp (me included), you have people that are tired of having to micromanage each and every unit to tell it what to do. In essence needing to hold every units hand and manage every little aspect of play...from telling which units where to go, when to attack and defend, and even what units to build. The other camp says, "hey, that''s better than having the computer make decisions for you". They want the fine level of control that comes with micromanaging, and knowing that if you click on something, it does what you tell it to do, Even the first camp is not too keen on the idea of losing control of your forces. How many times have you been frustrated with a game because the units didn''t do what you intended them to do? But I think micromanagement, and delegation of responsibilities are mutually exclusive. As a player, we want ultimate control of everything, but when we look at this from the flip side of the coin, do we like being told what to do by overzealous bosses? A micromanaging boss is one of the least effective kinds of boss there is. A true leader understands that his subordinates must be trusted with tasks set forth for him so that the boss can concentrate on higher more abstract things. And this is exactly what I want to get at. Not only do I want my game to be a warfare simulator, but really a leadership simulator as well. Not everything is going to go your way, and it''s the sign of a true leader who can get past unforseen situations and plans gone awry. That''s why I want the Commander and Communication system in place so that the player not only does not need to micromanage, but must learn to cope with situations where things do not go his way. Some people will say this is not fun, and skews too much in favor of realism. I say, "how do you know?". So far no game has really attempted what I''d like to do, so I''m hoping that like anything in life...there''s no guarantee that it will go the way you wanted. But that doesn''t make life any less fun. It keeps us on our toes, and it also presents unforeseen opportunities. So what I’m curious is, how many people here are dead set against the notion of having control taken away from you to let an AI object take control of events? As I mentioned in the above post, I think that even those who don’t like micromanagement are not keen on the idea of losing control...they just want easier ways to handle actions (for example, better control functions, handling economics in a differnt phase, or better gui’s for interfacing with units). I think the issue that I have with RTS’s is that they really just teach you to be the micromanaging boss that we all hate. I’d like to see a strategy game that is about leadership rather than control.

I''m all for delegation! The only problem is that you have to guarantee that the AI subordinates will be sufficiently intelligent to make good decisions when properly trained. There is no greater frustration than leaving the computer to take care of some menial task and having it make such a huge mess (at the most critical moment, no less) that you lose the contest.

Micromanagement is tiresome, especially in a truly expansive real-time conflict. Played Freedom Force yet? Too much hand-holding! I got tired five minutes into the first "real" level. I''d prefer for the heroes to generally do intelligent things (like, oh, I don''t know - beat up the bad guys?) but respond to my instructions to do something else. They may not respond immediately if it isn''t given as a directive and may instead factor such standing orders into their planning for the next several iterations (ie, "We need to shut down that PowerSphere" elicits an "Okay, I''ll see what I can do" while "Shut down that PowerSphere, now!" results in an "Aye-aye, Boss Dude!")

Keep the good ideas coming. Can I sign up to beta test when you start implementation?

[ GDNet Start Here | GDNet Search Tool | GDNet FAQ | MS RTFM [MSDN] | SGI STL Docs | Google! | Asking Smart Questions ]
Thanks to Kylotan for the idea!

A key to the idea of "good" AI ... is that it seems "good enough" in a given context ... it doesn''t have to be as good as a player would have been, as long as both sides have the same AI and AI bugs ... BUT it can''t just do stupid thing after stupid thing, raising the player''s blood pressure 10 points and causing them to bang their desk in frustration ..

Example .. the "Stars!" battle simulator moves all the pieces on the field for you, based on preset battle plans ... and the games choices are good, but not great, their are situations it does not handle really well ... BUT it is very consistent, and once you''ve seen a battle play out different than you thought it should .. you can play to change your fleets next time, or change their battle plan ... because the AI is doing CLOSE to the right thing ...

Another Example of games with this ... are games with AUTOCALC features .. like "Lords of Magic" ... These features are GREAT .. except that you can always do better, and you know it ... so the ONLY way to use these, is to agree BEFORE the game starts that ALL battles will be autocalced ... no matter how important they are ... and me and my friends have the ONE DO OVER rule ... each player can load the game ONCE, to take back an outcome that he feels is totally different than expected ...

If you''ve played Earth 2140 (and 2150) there is some auto AI around ... and if you''ve played Majesty ... it is ALL auto AI ..you can NEVER tell a unit he MUST do something ... you simply assign bounties to things ... and they decide what to do. The Unit AI in Battle Realms is also very interesting .. they will not run away from some battles, and they pursue battles of their own accord. That game also tried to get more realistic by making hurt units attack and flee much slower .. which is a great IDEA .. but it didn''t work out so well .. since in practice it means ... all outgunned forces are doomed to die to the last man, since you can''t run away fast enough once you get hurt ...

quote: But I think micromanagement, and delegation of responsibilities are mutually exclusive.

I don''t think so. Well, I really really hope it doesn''t anyway.
I''m currently working on(well, that''s one way of saying it) a strategy game that allows for either extreme, you can control every unit, create an ai using the script language, and/or use high level portions of the script or the default-and C3 lines can be cut(along with supply lines).

Roland-
I think you can do one or the other, but I can't envision how to do both at the same time...at least not without stepping on the toes of the subordinates. And that's what I mean by the micromanaging boss, the micromanager bsically gets in the way of the responsibilities of his workers. I suppose you could set up a game to do both if you wanted the control and the ability to delegate responsibility, although I guess that would be an extra workload. So now that I think about it, you are right...it is possible to have both aspects in the game if that's what you wanted.

Oluseyi-
When I was thinking about leadership, I realized that the AI would have to be VERY good to take control and make decisions based on certain hard coded knowledge as well as adaptive learning (either through training from the player or from node strengthening via GA's or NN's from its own experience). I really wish I knew more about AI programming...as it stands, I'm still learning the more advanced concepts of C++ But you are absolutely right, without a sufficiently good AI to hand off responsibilities to, my game idea will flop. As for beta testing my game, well, you may have to wait 3-4 years on that, hehe...though any help or input is always appreciated

Xai-
I was wondering how good, "good enough" is. You're right, auto calculations can definitely make things easier, but often lack a capability to support a strategem that a player has in mind. I suppose some scripting "rules" could be given to AI commanders that give them general guidelines and a framework from which to work with so at least they follow a certain track that the player would like.

[edited by - Dauntless on April 13, 2002 10:12:55 AM]

I'd vote to forbid micro-management completely, but then again, I don't like the AI idea too much either. It's not good if the AI is stupid (obviously), it's not good if the AI is too smart and makes the player feel unnecessary. Advanced group orders (formations, tactics, ammo consumption, movement speed etc.) should do it. Shogun style. An "You've asked for it, you've got it" type of system.

Btw, I hate it when games force me to micromanage. Age of Empires come to mind. Having to order a knight to dodge catapult fire ?? Huh?

[edited by - Diodor on April 13, 2002 12:32:39 PM]

One problem that people have mentioned is that the AI may not do things as well as you could do them by micromanaging. In some cases, a player might even be frustrated because his AI lieutenants are doing something in a way that is not objectively bad, but different from the way that the player would perform the task personally.

One thing to notice is that these statements are still true if you substitute "human subordinate" for "AI" in the above paragraph. It is the nature of leadership.

Another thing that the game designer can do to help with this is to design the game such that the AI is doing tasks that only the AI has the ability to do... not filling in for the player because he simply can''t micromanage everything. This at least sidesteps the problem of the player being frustrated because "that''s not how I would do it". Of course, the AI still needs to be competent.

Diodor-
That''s precisely what I mean by micromanaging. Units and their commanders should have an innate set of "rules" to cope with certain situations. These rules can be overriddn by orders however. Like you said, I hate having to babysit units because they are too stupid to think for themselves.

Hamdoon-
AI and human''s can equally do things that a Leader did not intend them to do. So, you''re right, the role of a leader should be to guide subordinates along the right track rather than tell them precisely where to go, and to put the left foot in front of the right foot ad nauseum. What you suggested about having contained roles is what I had in mind.

In a nutshell, commander objects have a rank object included within them. This rank class determines capabilities that the commander has. Ranks are structured by a number designating what level in the hierarchy they are (the higher the more powerful they are) as well as a operational level. Operational level is a bit harder to define, but I grouped it into 4 categories, though I may include more. These include, Field, Operational, Echelon and Support. Field officers are in direct control of units, and tend to be lower grade officers. Operational officers are mid-rank officers that can deal directly with OU''s and with the Echelon officers above them. Echelon officers are the high ups...the 1-4 star types thta do alot of the high level strategic planning. Support officers are a specialized catch-all that includes support roleslike Intelligence, Quartermasters/Logistics, Medical, Artillery, Electronics Warfare, etc.

Why have both a rank and a type? Different militaries have different levels at which officers have their roles. For example, Russian field officers are much like NCO''s in the West. They do not have the same level of power that Western field officers have. So even if they have the same rank, their capabilities are different. I want the military doctrines and "configuration" if you will, to be customizable by the player (for the hard core grognard).