Solving the Mutual Annihilation Issue

Hey, I am not a creep!

I am also technically with those guys, just not so extremely with them.

I agree with madsci's post, I mean to a degree. The game needs to have enough time for a player to react to what is happening and instant death doesn't do that, but having too much health does keep the battles cluttering the screen for too long with too many units joining mid battle and too much focus to both micro and gimmicks in it.

I hate currently how, if I see levelers without ground AA in them, and I didn't already get some air, then I can't exploit the obvious air weakness unless I stumbled across them when I was scouting (which is awefully lucky when I do, only happened once though).

I just also don't want it to be where by time the whole packed frontline wave of units clash and some start to die, your factory had already produced those units in the back of the pack, and as the next line and the next line move up, the factory keeps refilling the back of the line. Also, killing a structure to take an act of congress, it should be possible for 2-3 units to do by themselves before interrupted, if those 2-3 units survived the battle where your blob defeated the initial turrets and units in defense around the structure.

We've been watching... waiting for the opportune moment to strike...

Honest answer: my forum browsing habits are very erratic. I sort of bumble around threads and on different days different ones take my fancy. Go figure

Also can confirm, TrophySystem is not a creep. He does smell a bit funny though.

Great post!!! Essentially saying people should micro more though. If you're not stacking artillery units behind your assault units than you've lost

I disagree - it's reducing the importance of accurate micro because it's more forgiving if you make a mistake.

I think you misunderstand what we are talking about when we use the word micro.

see this is where you go wrong.

it's a complete shot in the dark. where have you seen this?

and don't say Supcom, cuz I actually play this game to this day, and I can school you on what really happens there.

Clearly none of you experiment with changing the layout of your troops. . . or formations. Skitters are pretty cheap to deploy, so are air units on patrol... It's a bit micro intensive to make your forces perform a daring flanking manoeuvre, but when you do you generally get some good wins.

It's an example of this: http://tvtropes.org/pmwiki/pmwiki.php/Main/EasyCommunication

Once your troops are actually committed to an attack, that's it, they are committed. It means that you actually have to scout and not blindly send your army to their deaths.

Lol... I uh... said this on page one.


Memory failure.

I use dox, pouring them out in lines on each side of my enemy from my turret line, essentially flanking and surrounding them in fire and drawing their fire all over the place.

I also use walls of tanks where turrets are scare and put nearby engis behind them repairing.

Both those work here. And they will be better once those tasks can be automated by the game instead of 20 player clicks.

I don't see how that is argument that unit's have no tactical formation while in low health blobs.

Please explain then..... @YourLocalMadSci

Godde has made this point in a different thread, but increasing unit HP has the effect of increasing the amount of time that a unit benefits from micromanagement.

As a simple example, if you want to kill a moving worker with a Dox, if the Dox kills it quickly then you only need to give the order to move nearby. However if the Dox takes 10 seconds to kill the worker, then you potentially need to micro that Dox for 10 seconds to follow the worker, shoot at it, etc.

Increasing unit HP increases the amount of benefit that individual units or small numbers of units receive from manual control because that individual unit will be alive for longer. In all RTS games where units are more durable, those units require (or at least benefit immensely) from manual control. And in fact the entire MOBA genre revolves around the principle that direct micro control of a single relatively durable unit is a full-time APM sink.

RTS games about large armies of units should be intended to make each individual unit relatively squishy precisely because it means micromanaging an individual unit is relatively unimportant. Even if you could dodge shots and extend a unit's life significantly, you have bigger concerns like managing your economy to increase your economy's scale in the same timeframe.

Losing a few units of a large group quickly isn't "unforgiving" because you haven't actually lost that much. I absolutely agree that having only a few units, where those units are fragile and expendable, is an issue. But in a large army context, that ceases to be a problem.

With respect to the point someone made about things like an artillery barrage inflicting multiple kills- that is absolutely the point. Directing a coordinated artillery barrage isn't micromanagement, though. That is a strategic decision about when to use a group of artillery, and where you think they would be best used. And even if it destroys multiple enemy units, they have a large number of such units and that isn't "unforgiving" by itself.

The more HP you give to an individual unit, the more advantage can be gained from prolonged micromanagement of that unit. Increasing unit HP will result in more micromanagement of individuals, not less. Large armies of units makes micromanagement of a fragile individual less significant.

This may be arguing semantics a little, but I believe this will help clarify things.

The word and concept of micromanagement was around long before RTSs. It was originally used in business for a type of managerial mistake. It means when managers have a tendency to regulate and control their underlings activities far more than they should, when the correct option is to allow their underlings the chance to complete tasks under their own initiative. For example, a manager tells his underling to write a letter to a client, while a micromanager stands over his underling's shoulder and dictates the letter exactly.

I believe this concept translates well into the arena of RTSs. For me, it means "when the player is forced to manage their units in order to maximise their utility, for matters where they should be perfectly able to make that decision by themselves". For example, suppose there was a game whereby units would never automatically attack enemy units which came into range. Having the player constantly being forced to give targeting orders would be micromanagement. Same goes for things like kiting and dodging. Ordering units to fall back under superior fire-power, or things like flanking and reinforcing are not examples of micromanagement, as there is no way that a unit could "know" the state of the battlefield to the point where it could decide what it should do under it's own "initiative". The former behaviours can be achieved with very simple bits of code handling the most rudimentary of decisions. The latter requires complex machine learning techniques such as influence maps and neural networks to even approach the kind of higher level decision making that a human mind can make (at least for now). This is why Uber have hired an extremely talented AI expert in order to apply such techniques to these high level decisions.

This segues to the next point:

You have the correct premise, but the wrong conclusion. A greater average Time To Kill per unit offers a greater potential to be micromanaged. However, It does not automatically follow that it offers a greater advantage from micromanagement. For example, let's return to the gunships vs. tanks scenario. Let us suppose that due to the increased TTK, a player discovers that if they dodge their tanks around a bit, they can make them survive a little longer against the gunships. This is an insignificant advantage. The tanks are still doomed, and a better response was still a larger scale move such as a retreat, or sending appropriate reinforcements. Let us examine this more closely with a different example:

I have highlighted the weasel word for you. You haven't explained how a longer engagement directly leads to it being advantageous to micromanage, only how it offers the potential to micromanage. In both cases of the Dox taking 10 seconds to kill the fabber, or taking 2 seconds, the command to kill the worker is still the same. Select the dox and then click on the worker. Dodging, kiting, stutter-stepping or any of the other tools in the micromanagement toolbox are still no more advantageous than a simple kill command. What the greater time does allow is for the player controlling the fabber to perceive the "unit under attack" warning noise, move to the site of the fabber, and allow it to either run away, sacrifice it to finish what it was doing, or send reinforcements. If it had almost finished a laser tower when the Dox appeared, then any of these might be viable decisions. This creates internal conflict in the player as to what they might want to do, which is the fundamental factor which underlies decision making in games. If the battle was already over before the player had a chance to react, then that opportunity is lost, along with many others. It is apparent that there is a "sweet spot" for engagement length where the player has sufficient time to make a decision, but not so much that the engagement drags out. The current engagement times in PA are far below this sweetspot, whatever it may be.

So, what factors do make it advantageous to micro? There are lots of design decisions that lead to micromanagement being a dominant game mechanic (for example unit motion vs accuracy, UI limitations, etc.), but there is a particularly egregious one which I would like to take some time to make an example of. Namely, unit special abilities. These are a common mechanics where a unit must be manually told to use a special ability, normally limited by a cool-down time. Lets return to our gunships vs. tanks as an object lesson.

Suppose that our tanks are now given a special ability to fire some sort of anti-air flak shell once every so often on manual command. Obviously, this now creates a situation where it is advantageous to micromanage the tanks against the gunships, as correct use of this ability could destroy the gunships. This allows us to see the true role of extended TTK in the context of micromanagement. If we have the special ability AND the long TTK, then there is opportunity to micro. However although the extended TTK offers the opportunity to micro in conjunction with the Special ability, it actually makes the micromanagement of this engagement more forgiving in that players have more time to react. The worst case would be the presence of special ability and a very short TTK. This creates the instance where only the fastest and most reflexive players are able to use the tanks to their full potential, which is not really the kind of skill set that we are hoping PA promotes. We already have Starcraft for that kind of game, and it would be bloody stupid for PA to try and out micro Starcraft.

If you don't like Special abilities as an example of a micro-generator, then you can say the same about kiting/dodging mechanics, which is why I hope that PA will eventually have some sort of auto kiting/dodging mechanic, is this is yet another thing that lessens the advantage of micromanagement, without subtracting the advantage of broader strategy.

To summarise, it is a fallacy that longer engagement times automatically lead to more micromanagement. What is required for that is a combination of longer engagement times with mechanics which benefit players who constantly baby-sit their units. What longer engagement times do allow is more time to think, plan, and react, such that players can make broader moves such as retreating, reinforcing, enfilading or feinting. These are concepts so important to any game purporting to be about strategy that they need to be encouraged.

Slight tangent to discussion - but this instantly gave me an idea for a mechanic for this:

Each unit has a second 'engage' ring outside its weapon range ring. When set to the equivalent of 'kite' stance (are we having stances?) the unit endeavours to engage units which it detects within this engage ring, but attempts to keep them at the very edge of its weapon range ring.

If that's not clear I can knock something up on Paint.

Please continue the debate, though, this is an excellent read.

In the words of Sun Tzu, the victorious general first wins the battle, then fights it. The defeated general first engages in battle, and then seeks to win it.

I do not consider saying a unit potentially gains from more focused micro to be inconsistent with saying that it does benefit from micromanagement. The only difference is that a player might not micromanage the unit for other reasons. Perhaps a different unit or task is more important right then, or perhaps the battle is already won anyway and micromanaging would be pointless. Saying a unit "potentially" gains from micro is the same thing as saying that a player can micromanage it and gain an advantage. Whether the player chooses to micro that unit right then does not matter.

Giving units longer time to kill will create more opportunity to micro. In your own counterexample, MadSci, you say that a longer time to kill gives the player more time to react, such as to run away with the con or do something else. This is micro. A longer time period creates a higher micro potential for every unit.

You assert that giving a player 10 seconds gives that player more time to react. False. You still must react instantly, but now you have to stay in manual control for 10 seconds. Multiply this by however many units you have active at once and you can see it quickly becomes unmanageable, but the potential is still there. There is actually so much micro potential that it is impossible to capture all of it. What this means is that players with faster hands and better micro can capture more of that micro potential. Time spent in slow reaction time is wasted potential. Players with faster hands and faster reactions actually benefit far MORE from having a whole 10 seconds' micro potential for each unit. Especially if their hands are fast enough to simultaneously micro multiple units. Longer time to kill makes multitasking micro in battle vastly more powerful.

Short time to kill means that a unit that is attacked by an enemy that should kill it will do so quickly and both players can move on to other things. There is no point prolonging an inevitable combat result. When a small group of units gets swamped by a huge group, let both sides take their losses and move on. However the more time it takes for the battle to resolve, the more opportunity there is for the player with faster hands to gain an advantage during that battle. Giving one player "more time to respond" just makes it easier for a player with slow hands to capture some micro potential. Players with fast hands will definitely react immediately, and will capture far more over the longer time period a longer battle provides.

True, if you prolonged the battle tremendously using HP, then it would no longer be inevitable. Two ACU's in SupCom take so long to kill one another that multiple waves of reinforcements can arrive. But, as I have said before, high DP duels are ridiculously silly, to the point of being comical to watch. Certainly not interesting gameplay.

But the increases in HP that people are proposing are not anywhere near substantial enough to make battles so slow that a significant amount of redeployment of forces can occur over the course of a battle, purely because of unit HP making each unit take so much time to destroy. Nobody suggests this because it would be silly. Can you imagine a time to kill on the order of the amount of time needed for reinforcements to rally from the factory? It would be hilarious, until you realized you can just totally ignore the enemy's units and just walk over and blow up whatever you want.



Ultimately the problem is that the only type of army interaction we have is pitched battle. Ideally, mobile forces should be able to engage each other at range using a variety of unit roles. Temporary defensive positions, flanking forces, small groups of units arranged in a semi-stable front, and a variety of other unit types would create the kind of long engagement that will be slow to resolve. An engagement of armies can take a long time to resolve, as both armies more gradually suffer casualties and receive reinforcements, a few units at a time, sometimes in bunches due to a slightly larger confrontation where a larger fraction of the army is committed. Both sides should cautiously attempt to gain a strategic or tactical advantage instead of just blobbing up and attack-moving. Instead of taking every single unit at once and just attacking, it should make sense to try and maneuver for advantage and fight a slower, more cautious, more deliberate prolonged engagement.

The option to just assault obviously remains, but it is an option that should be exercised with extreme caution due to the possibility of an unsuccessful all-out assault costing you a lot of units for very little gain. The attacking force needs to have a very large strength advantage, as well as good intelligence on the disposition of defenders, and make a calculated move to just brute force it. Despite the simplicity of the attack, this kind of all-out assault should require the most premeditation and second-guessing before executing. Raiding with small groups of fast units, sending an artillery volley or air strike, a missile volley from some gunships, these things are relatively small maneuvers that may risk a couple of your units. But taking everything and just charging in potentially risks everything all at once. And there is a good chance your enemy could crush your assault very efficiently.

But a single pitched battle should resolve relatively quickly. When squads of a few Doxes crash into each other at point blank range, one side should just win quickly. In order to win a pitched battle, you need to have deployed the right number and types of units in the right places. Perhaps softened the enemy up from a distance, perhaps tricked the enemy into sending forces to defend somewhere irrelevant. But a single pitched battle is decided before the first shot is fired. Giving the player "more time" is giving the player more opportunity to change the outcome through micro. The victorious general first seeks to win, and then seeks to fight. The defeated general first engages in battle, and then seeks to win it.

The unit AI should be good enough that the units fight just fine regardless of whether the player is paying attention at all. And consequently, it makes no difference if the battle is over quickly because it is unnecessary to give the player time to micro. You should have deployed your units differently if you wanted to win that one battle; micro is not going to save you.

Instead, squishy units makes you cautious with your units. Which, assuming the right units roles are available, means you should try to be cautious and get extremely efficient exchanges instead of blobbing up and brute forcing your way through the enemy because your HP buys you so much time that you don't fear the enemy's guns. Even a very large army should fear and give some respect to even a small group of enemies and engage slowly, or risk losing a lot of units inefficiently.

That sounds like a great game you've envisaged.
Unfortunately it doesn't sound like Planetary Annihilation: The Spiritual Successor to Total Annihilation.

Quoting Sun Tzu is all well and good if we are developing a treatise on real world military theory, but not if we are making a game. Remember, reality has up to three roles in a game - as a consistency check, as a guide to new players, and finally as a source of inspiration. Many parts of military tactics are interesting challenges, such as retreating and supporting units. They force players to make decisions which alter the outcome of the game. Battles always being already won before they are fought is not a fun or interesting mechanic.

I'm sorry, but the logical fallacy is so obvious hear that it needs to be pointed out. In fact, it can be summed up in the classical formalism:

1. Micromanagment requires time.
2. More health gives more time.
3. Therefore more health results in more micromanagment.

I'm afraid that if you cannot see why that is a fallacy, then I cannot make it any clearer. Potential is different from actuality. I could potentially assault my enemy with dozens of isolated Doxes. I could potentially rush orbital without building a single combat unit or defensive structure. I'm not going to do either of those things though, because it would not be advantageous for me to do so. Just because someone can do something, does not mean they will. What they will do is detirmened by their perception of the utility of the action itself, not just how long it takes to do. In fact, you already dismantle your own argument by giving the reasons of why they won't be microing, even when they have the opportunity to do so:

This is why more time does not result in more micromanagment. Because there are too many higher level decisions to be made, that spending time microing is pointless. Or at least, that's the way it should be. Examples of higher level decisions are things like "Should I retreat or advance?" and "Should I support or sacrifice". The current game doesn't give us the time to make these decisions. If you think that these are examples of micromanagment, then I'm afraid we are in semantics territory.

What you have suggested instead is that players should be punished in the harshest possible way for every mistake or unlucky gamble, with no way of recovering their position. That games should be decided as quickly as possible, not by the accruing aggregate of player error, but by the first mistake they make. That defeats should be so swift when they occur, that players should always shy away from confrontations unless they are 110% sure that they will win. This is not a recipe for driving player engagement.

The only way to come out on top in a game where failure is punished so instantly is to give players the tools to see the failures ahead of time. This would mean relying heavily on intel as the key area for engagement in the game, instead of the actual destructive side of the battle. The problem with this is that intelligence has always been one of the weaker sides of this kind of game. There are only really two forms of intelligence gathering - radar and vision. There are only really a two of ways of defeating intel: jamming and cloaking. Anything else requires a direct destructive action. This game does not have the tools required to make intelligence gathering the primary area of engagement, when the core game so far has always been about direct confrontations and battles. If battles should always be quick and painful, then what you are looking for is a game where direct conflict plays second string to the reconnaissance system. A game where subtlety, sabotage and subterfuge are the main weapons, and the tanks are only brought out when the war has already been won. A game where the basic tool of domination is not the armed soldier, but the intelligence operative.

This sounds like an interesting concept, and if you ever find or create it, then I suspect I would enjoy it. But that is a game about scalpels, while Planetary Annihilation is, and always has been, a game about sledgehammers.

The Art of War is about strategy, and in a strategy game the basic concepts should still hold even though the weapons differ. Referring to The Art of War isn't like referring to the Army Field Manual- it's not a realism argument. Basic precepts of strategy also apply to all strategy games.

If you intend to use a syllogism, you have to get it right. This is an improper formalization; your conditional in premise 1 is backwards.

Done correctly, my argument is as follows:

1. More health gives more time.
2. More time creates more potential unit actions before death.
3. Therefore more health creates more potential unit actions before death.
4. More potential unit actions before death means more micromanagement is more useful.
5. Therefore more health means more micromanagement is more useful.

As you said yourself, increasing unit HP gives the player more time to act. More time to give input and affect the outcome. Slower players benefit from this additional time less than faster players.

Sometimes I get the impression we are spending a lot of words just in heated agreement, and this is one of those times.

I absolutely agree that players should focus on higher-level decisions, because that is where game-deciding decisions are made. Large strategic moves like moving an entire army about the planet are obviously in this category. And if you make scale sufficiently large then micromanaging individual units to do things like target fire and dodge shots no longer makes sense.

We also agree on the definition of micro. Tactical decisions like advancing and retreating are not micro. Having to manually control movement or assign enemy units as targets would be micromanagement, and should be avoided.

The disagreement we have is whether ordering a single unit to retreat is micro. Granted, if you are using a blob (which needs to be solved first) then it would make sense that if only every member of the blob had more HP, the blob would survive for longer and the blob could retreat. But the blob is a much bigger problem than anything to do with micro. And the blob needs to be broken up into smaller pieces that are more diverse and leads to a more structured battlefield.

If the blob has been addressed, then you aren't engaging with the entire army at once. When you want to retreat with an army, the units that are retreating are probably not even engaged with the enemy at the time. If you have a front line which has been attacked, and you think it's going to break, the units on the sides of the breakthrough and the units in the rear you want to protect should retreat. But the units that are actively engaged? Retreating with the entire force works just fine even if those units that are actually engaged will be dead quickly.

I really think we are just heatedly agreeing on pretty much everything about micro. And the entire issue of whether HP leads to micro is moot anyway since we agree that the scale of the game should make combat micromanagement irrelevant.

If we solve the "blobs that mutually annihilate" problem, then we can start talking about how to reduce micromanagement, if it is a problem. I agree with you on pretty much everything about micromanagement and what should be done about it. With the one except of when you say high HP reduces the impact of micro.

This is just not even in the same universe as what I am talking about. Games should not be decided "as quickly as possible"- the whole game is not decided by a little fight between small pieces of just two opposing forces, regardless of the outcome.

Neither is the game decided by a single mistake. Even if a single small confrontation goes disastrously badly, the army is still mostly fine. Not only is the game not over, not even the engagement is over between those two armies.

A blob of Ants is currently a very boring army. It only contains one type of ground combat unit. It doesn't really have any interesting differences in effectiveness when used this way or that way. In fact, the best possible use of a group is to grab every tank you can and to just charge directly into battle. That is my main issue. Two blobs of units just collide, and one of them is just gone, while the other one is just smaller.

My main issue is that two armies collide, and one of them almost immediately vanishes, leaving the other to advance uncontested. One of the armies is just gone, while the other one is just smaller. Even if the units take longer to die, you still have the core problem that blobs of units just mutually annihilate; the process just takes longer.

Instead, armies should engage each other by spreading across space, and trying to probe and pick off enemies, to use a variety of defensive and offensive weapons and maneuvers to gain an advantage. And the way to do this is with units that fear each others' guns, not by making them more durable and better at just dogpiling.

Armies should lock into each other, with neither being able to force its way into or through each other until the enemy army is weakened and defeated. When you believe the enemy army is weak enough, then try to break through. Both armies will have shapes and will occupy territory with different groups of units spread across space. Losing just one unit or small group isn't "game ending" even though it dies quickly. I can make diagrams of the kinds of army shapes I am talking about if that would help.

The reason they lock into each other is because of an inability to just march through each other. Their fragility means they have to respect each others' firing ranges. You have to fight to take ground from them, not just wipe them out and then immediately roll straight to Berlin.

I don't see TA the same way you do apparently. The sledgehammer is certainly available, but having a huge economic advantage over your opponent so you can just attack-move with overwhelming numbers is not always an option. Generally this only happens when your opponent is bad and fails to expand. Against serious players TA and ZK games are won through death by a thousand cuts.

Against skilled opposition in any TA-like game, the entire match is a series of skirmishes of varying size, and seldom with every single unit either player owns. After the early land grab phase, the game stabilizes into a structured war with territories held by both sides, and calculated moves to try and take certain areas. This goes on for a while, and then big ticket items start to get produced which tend to destabilize the equilibrium that resulted from neither player being able to outright kill the other.

I definitely recall this exact dynamic in TA between strong players. It certainly is alive and well in Zero K. Assuming your opponent is strong enough to not just crumple to basic harassment, the game stabilizes into a well-developed, structured stalemate of units and buildings. Although the scale in Zero-K is much smaller than in PA; I envision one engagement in PA as being like an entire match of Zero-K, except possibly with multiple such engagements going on simultaneously, either on different planets or in different regions of a large planet.

For an even more extreme example of caution because of fragile units, check out Wargame: Airland Battle. An entire match of Wargame is one engagement of two armies. Confrontations between squads of infantry or tanks are resolved in seconds. In fact, dropping bombs or a direct missile or shell hit on almost anything tends to kill it instantly. But the engagement as a whole can take half an hour or longer. Again, PA's greatly increased scale leads me to believe that there should be multiple such engagements going on simultaneously in different places.

I believe you have lost concept of micro sir........



I only see automated threat response or auto micro if you will just creating a huge dance display with the same one for one exchanges we have now . Yes more time should be involved in conflicts to give the slower newer players a chance.
Please give an example of how intelligence gathering can be improved ??? What else is there aside from radar and visual contact??? Just remove fog of war??? I don't really understand what you're getting at....... scouting will always be essential and so will micro. I still missed where I was confused on micro as well..... taking direct control of my units and making every decision for them is micro or as you put it dictating their letters.


At this current stage in the game a simple attack command is not the same as a move command to the targets area. Currently an attack command send s your unit to the targets previous location even if los is maintained. Then said unit continues an extremely inefficient path of pursuit. While the command has the potential to be better than a little micro in such an instance currently it does not. Therefore a human estimation on your targets heading and a move command is much more efficient. Hopefully code gets improved and the attack command will allow your units to predict a targets heading independently, this will alleviate a great aamount of micro and put the game back to the more macro aspect that is apparently being aimed for.

I would like to add one more point. Adding inaccuracy to weaponry will actually make units take longer to die, while still leaving them squishy such that they die in few direct hits.

The difference is that a unit with a lot of HP has impunity and can ignore enemy weaponry to a certain degree. A squishy unit that the enemy will miss a statistically significant amount of the time still has to fear enemy weapons. Even though in practice the unit's effective durability has increased (takes more shots to destroy it), you are still exposing the unit to risk to put it in range of the enemy.

Enduring wrecks have a similar effect of blocking weapons fire and increasing units' effective durability while leaving units squishy. These things slow down pitched battles while still forcing both sides to respect each others' positions and not just automatically blob up and charge.

By contrast, a unit that just has a lot of HP will just tank shots for as long as its HP holds out, and can move anywhere with impunity and shoot whatever it likes. Squishy units that can score kills efficiently, or which can be killed efficiently, require more strategic care with their deployment, and more tactical acumen to get the most out of the units you have in an engagement. If you want to slow down a pitched battle, then making combat less efficient using obstructions and inaccuracy is a better solution than just making every unit take longer to die in all circumstances.

Correct. Because all you're doing is lowering the average damage per second.

It's a combination of real life military engagements, and the phrase "No plan survives contact with the enemy"

think about Alexander the Great era of history.

Once your troops actually start fighting, they start dying pretty quick. They're committed to the battle, and that point you can't actually withdraw them without suffering massive losses (because you can't turn your back to the enemy and run in full armour.

It means as soon as you've committed your troops to battle, that's it. You have to keep some troops in reserve to deal with flanking attempts and so on.

Essentially low health blobs means that the phase when you are changing your troop layout and so on occurs earlier in the battle, not after your units have engaged.

How about 6 seconds? Six seems like a good number. It's enough time for armies to fully engage, and gives a viable window for wounded units to pull away.

In fact, another game used 6 seconds as its combat baseline. You wouldn't believe how fast paced that game was.