EPT
A couple caveats-
1) This is written within the context of changing back to true Mech Play. This argument should be viewed within that argument. If you don't agree with the original argument of what Mech Play is, then this argument will not make sense at all.
2) I don't think so-called smart tank ai causes mob battles any more than adding overkill will automatically get rid of mob battles. But I do think overkill pushes in the right direction.
3) This is not an all or nothing argument where SC2 has 0 and alternatives have 1000. It's more a continuum where SC2 has some, but there could be more
Purpose:
To convince you that so-called SC2 Tank 'smart ai' is actually not as smart as Overkill. If not that, then at least I'd like to contend that neither are 'smarter' per se but rather different design choices with different implications.
Laying the Groundwork
Before we get to the main argument, we need a little common ground. Can we agree that certain unit designs allow or promote/encourage certain behaviours? For instance, baneling splash promotes splitting marines as the ideal response. If the banelings had no splash damage, the Terran player would be much less inclined to to split their marines. Furthermore, the marines speed and manoeuvrability allows the Terran split in a hurry. If marines were slow and sluggish, then it wouldn't be worth it to try and split the marines. You'd be better off firing a little longer until they die. Especially if the marines were given extra hit points.
Another example. The existence of air stacking allows Thor splash to be effective against mutalisks. The existence of splash damage, promotes the behaviour of trying to split up the mutalisks to avoid splash damage. The existence of air magic box allows the mutalisks to execute this behaviour.
![[image loading]](http://i.imgur.com/Ps4ao.png)
So in more general terms, certain unit abilities or unit designs forces a reaction or promotes a certain behaviour. Further unit design on the unit reacting allows the player to react in a competent manner. They have a tool that they can use- air magic box or fast moving marines for example.
The Argument
First what is Overkill?
In the case of tanks, if an enemy comes into range tanks, all the tanks that are facing the enemy will fire even if it's too much damage. So if 4 tanks were in range, but only 2 tank shots were needed, all four would fire anyways. However, turret rotation time will stagger the fire, so if the 4th tank was facing the wrong way, 3 tanks would fire and the 4th would now be facing the right way if more units come.
Now right away I can already hear the protests. "We don't want to fight against the computer" "automatic maximized damage is superior to wasting all that extra damage, why do you want to create additional, arbitrary difficulties?"
Automatic Maximized Damage.
My first counter is that it's not so simple as automatic maximized damage is 'smarter' than overkill. Damage output is a trade-off especially when we're dealing with AoE damage. When we're arguing that no overkill = smart ai, it assumes we're still talking about the same unit stats and abilities. We're trading damage to make maximized damage easier. When AoE of is more easily available, then the damage must be reduced unless we want the battle to come down to a couple AoE units.
The easiest thing to point to is the nerfing of damage to both EMP and storms due to smart casting. Tank fire also got nerfed due because of so-called smarter ai.
Ghost Example
However, smart-casting on spellcasters is a can of worms I'd rather leave alone. The best example of the trade-off between damage and more easily guaranteed maximum damage is the ghost-nuke.
It's better because it's less controversial. The skill of using ghost-nuke is pretty comparable. So we don't need to have the 'fighting against the computer' or 'takes more/less skill in ______ game.' They both handle the same, but SC2 made the nuke more easily available. BW had the ghost and nuke buried way up the tech tree plus nukes cost 200/200 and 8 supply. It was not easily available, but it brought down the health of any unit to 2/3 of max health or 500 whatever was greater. 2 Nukes could take out a Nexus or an EMP and a nuke could kill it. 2 Nukes could take out any building actually and most only needed one.
![[image loading]](http://i.imgur.com/F82oI.png)
one nuke
In comparison, SC2 made the ghost technology much more easily available just by moving the tech down, getting rid of the supply and dropping the cost to 100/100. So suddenly it became much easier to drop nukes and even drop multiple nukes on the battle field. So as AoE can now easily cover the field (due to cost/availability), the damage was subsequently nerfed to 300 (+200 to buildings.) The damage radius may also have been adjusted, but I'm not sure.
![[image loading]](http://i.imgur.com/WTHpt.jpg)
couple nukes
So in BW only a BC could withstand a direct hit (with 3 hit point left- 6 with armour upgrades) while in SC2, the BC, Carrier (with 150hp), Archon, Ultra, and Mothership can all withstand a direct hit.
Now I actually like that ghosts and nukes are easier to get in SC2. However, the main point is when it became easier to lay down the AoE on the map, the ability itself changed. It's not so simple as "I want my units to always fire the right amount of damage because that's 'smarter.'
It's a trade-off because now that it's easier to evenly distribute the damage, the damage got nerfed. Not only that, but the cost of the tank and supply went up in SC2. So in order to get this 'awesome tank' that will nearly automatically maximize its damage, we get less damage, they cost more, and we get less of them.
![[image loading]](http://i.imgur.com/R3JMH.jpg)
The Tank
![[image loading]](http://i.imgur.com/cMXsM.png)
Min 150
Gas 100
Supply 2
Damage 70 (75% medium 50% small)
30 Tanks = 60 supply
![[image loading]](http://i.imgur.com/xYOxh.jpg)
Min 150
Gas 125
Supply 3
Damage 35 (+15 armoured)
30 Tanks = 90 supply
Not the same tank
We're not even talking about the same units that are supposed to be smarter. It's not a simple 1:1 translation where the only change is the unit is now smarter. There's a chain reaction of changes without overkill, which makes this change more a design choice rather than improving unit ai.
Automation vs Manual
To really hammer home why automation is not inherently better than allowing manual control, I'd draw your attention to combat in traditional MMO's vs FPS (or even FPS RPG's like Mass Effect 2.) If we're going with the idea that every shot hitting every single time is smarter ai, then MMO's combat system is 'smarter' you just have to manage cooldowns and it'll auto hit every time. Even if you're using a gun like Star Wars Old Republic.
SWTOR
However, if you get rid of some of that automation and give the player manual control, then skill shots can exist like in FPS. Not every shot will hit every time, but there's a subsequent increase in player skill that's required to make the hit. Furthermore, the damage is a lot more powerful because not every shot is guaranteed to get maximum damage.
Mass Effect 2: Skill shot fps
And when it comes down to it, if you manually target units with any other unit there will be overkill unless the player chooses the minimum number of units to one shot their target. Aka better unit control rewards the player with the most efficient kills. Why should the tank be any different. The player that is better at positioning tanks will be the one with the most efficient kills.
What is the Difference Really?
Now we're going back to the idea laid out at the beginning- that certain abilities promote/ encourage a certain behaviour. And that other abilities allow the units to execute a desired action.
The problem I have with 'smart-ai tanks' is not so much what it does do as much as it what it doesn't. You do see tanks spread out for a variety of reasons, but they don't force particular reactions or promote a variety of behaviours. Whether the tanks are clumped together or spread apart, there is little easily discernible difference in maximizing damage. The tanks will evenly distribute it's damage to maximum efficiency automatically.
![[image loading]](http://i.imgur.com/KLHwp.png)
Right away some of you are thinking. Good that's exactly what I want. But remember, as a trade off we're losing some damage. With overkill, we can keep our damage and give increase the value of where tanks are placed. I'm not saying that currently tank placement has no value. Far be it from that. But I'd suggest that we can increase the tanks power as well as increase decision making for tank placement.
With automatic maximized damage, tanks can be spread out (within reason) and clumped together and the damage output remains the more or less the same. This also means, that the opponent can send out all their troops at once or stagger their line and it won't make much obvious difference. The tanks will only fire as much as is needed.
That's what I mean by I have a greater problem with what 'smart-ai' doesn't do. It doesn't force the players to adapt very much to play better.
So what is so interesting about Overkill?
1) More Incentive for Terran to Spread Out
With overkill, the opponent can bait your clumped together tanks with one zealot. The entire tank line fires and then the opponent can close the distance to get at the tanks. Therefore, Overkill promotes/ encourages the player to stagger their tanks.
Now the tanks must be spread out to in order to be used to maximize efficiency. However they cannot be too spread out or the opponent can blitzkrieg through a weak part in the line. There then becomes an interesting tension between clumping the tanks too much so that too many shots are wasted and spreading too thin and making the tank line vulnerable to penetration.
![[image loading]](http://i.imgur.com/koZU5.png)
i) RED- first salvo from front row
ii) BLUE- goliaths coming in from behind
iii) GREEN- back tanks unfired due to positioning ready to fire on incoming goliaths
2) Turning Splashed Damage Against
There is even greater cause to spread out your tanks with overkill. With no overkill, if you sneak zergling or zealot right up beside the tank, just the right number of tanks will fire and the enemy unit will die. The surrounding tanks will suffer damage from splash, but the bare minimum. With overkill, the same unit dies, but because tons of tanks are fired, the nearby tanks are absolutely obliterated by all the stacked splash damage.
![[image loading]](http://i.imgur.com/TPYy3.png)
So this further pushes the Terran to spread out tanks to minimize splash damage. Furthermore, it makes it even more imperative to that the Terran throws cannon fodder in front to prevent units from getting next to the tanks because they will get absolutely destroyed. So already we're starting how these changes would further promote further positional play
3) Dropping on Top Tank Lines
With overkill, the Terran's opponents are further encouraged to try more drop play. Zealot bombing works because the shuttle gets the zealot over the cannon fodder and when the zealot is dropped next to tank(s) all of the nearby tanks fire at it, destroying the nearby tanks by stacked splashes due to overkill. There are other problems preventing zealot bomb style play- Terran air superiority with marine-Viking, but overkill opens up a greater possibility. It's cute tactics like this that pushes away from ball vs ball or multiple mini-ball concaves.
![[image loading]](http://i.imgur.com/7gBh2.png)
Dropping zealots out of shuttle to make tanks kill each other.
4) Incentive for Terran Opponent to Spread Out
Now is when we really start cutting into the deathball. Before, tank position didn't matter (as much) and Terran opponent's army formations only mattered so much as to minimize splash as best as can be managed. But without overkill, the opponent has less control in dictating the battle through better execution. No matter how the opponent attempts to spread out their units, the tanks will still hit to evenly distribute its AoE damage. There is a difference, but it's a much more subtle difference.
But I don't think that a spectator sport should settle for 'subtle' when we can go for obvious/ blatant skill differences. This is a visual medium and obvious skill is more spectator friendly than subtle only the greats will pick out. There should be subtle shifts for players to gain advantages, but there should also be obvious shifts, where it was clearly evident that better execution allowed the player to win. As opposed to better composition.
So with staggered tanks, an oncoming clumped army (deathball or a concave of miniballs) will be shelled to pieces from the alternating barrages. With increased tank power (and more of them), the Terran player with better positioning, will wreak havoc on the clumped army.
Therefore, this promotes more spread out play from the Terran opponent. To counter better positioning, they need to stagger their own troops. Spread them out. Not just into mini-clumps, but to actually spread them out. Or maybe flank them with fast cannon fodder from one side and then send in the main force in a concave on the other. (Especially with Turret rotation and if Tank rate of fire was slowed.) The Terran opponent, can even bait Tank fire with a couple sacrificial units before sending in the main force.
![[image loading]](http://i.imgur.com/tyhS6.png)
Hm, this is actually kinda hard to capture on screen cap. But that's because the battle ranges across multiple screens which is rather what people are often pushing for when they complain about SC2 battles.
5) Appeal to Race Distinctives
One of the distinctives about Terran Mech Play was there crazy powerful splash damage. Between Siege Tanks and Spider Mines, armies could just melt to smaller Terran merch armies. However, that same crazy powerful splash could be turned against the Terran. Spider Mines could be pulled into the Terran's own army and Tanks' splash could be turned against their own tank lines to wipe themselves out. Overkill was part of the equation that allowed Tank damage to be both the Tanks' best weapon and it's own worst enemy. Losing Overkill loses part of the distinctive of the Terran Mech Play. Sure there's still splash damage that can hurt tanks, but the difference is less extreme and thus minimizes the Terran distinctives.
Overkill is a chain reaction. Tanks need to be staggered/ spread to get maximum damage, this in turn forces their opponent to also spread their troops even more.
Further Tools
There are a couple tools that allows the players to respond better to Overkill.
Group Movement.
I'm still unsure why people seem to think SC2 unit clumping, with all the pushing and shoving is superior to alternate solutions. Forget about buggy unit ai. (The dragoon thought it was smaller than it actually was, which is what caused that set of problems.)
Can we agree that giving the player multiple options in moving their troops around is better? For instance, if they had tools that allowed them to spread out a little better, allowed them to clump up, and also to stay in formation relative to the the other units in the group? Why would we want just one of these options?
1) Spread Out on the Move
One tool that could be added is when you right clicked your units and the group travels a distance, the troops spread out in a longer line rather than keeping clumped up. This has both an advantage and a disadvantage. The long line, spaces out the units (the zealot conga line) which allows the player to execute their attack better to counter the tanks overkill. With conga line can be run perpendicular to the tank line and then attack moved in so that the troops will spread out once they get amongst the enemy, thus minimizing splash and overkill from the tanks.
![[image loading]](http://i.imgur.com/l5f40.png)
i) GREEN- Right click after long distance travel
ii) RED- a-move forward to in a spread out line.
The drawback is that the army is more vulnerable on the march- it's more spread out. But this actually gives greater variance between an army on the move and an army set-up, which I think creates a more interesting dynamic. It gives advantage to the player who has set up in position. It also means the army on the march needs to form up form up before attacking. In addition, harassing the army on the march is a little more viable as the units are more spread out.
(I don't like appealing to real life very often because gameplay matters more, but it actually makes more sense that an army on the march is more spread out into long lines than to move in a great big ball. It's more incidental than an actual reason.)
So the army can still gather close together, but we add a second option that allows the units to spread out more. There's also a more significant difference between army moving an army prepared.
2) Magic Box
The other tool for more interesting interesting troop movements is to fix ground magic box. This one should just be a no-brainer. (Air already has it, so why not make it work better for ground?)
Longer description if you don't know what this is:
http://wiki.teamliquid.net/starcraft/Magic_Boxes
Ground magic box exists to some extent in SC2, but it seems relatively smaller/ less useful. With magic box, players can clump their units if they want or spread and keep in formation if they want.
It requires skill to use as it depends on where the units are in relation to each other. If the units are too far apart when you select them, they'll clump. If they're too close when you select them, they'll clump. But if you select them within between, they'll move in precise formation without shuffling around. This allows for very controlled, predictable movement. This in turn allows for very precise micro control. (See Bisu versus spider mines- there is no way that's possible if units had no option except to clump, shuffle and shove. But we still keep unit clumping (especially if you select your entire army.) But more than anything it just expands the players options.
Maximum Area Selection
![[image loading]](http://i.imgur.com/KaYGV.jpg)
Minimum Area Selection
![[image loading]](http://i.imgur.com/eLFFC.png)
What this means is unlimited selection is still in place so the game is the same difficulty for low level players, but for medium level players and up, there is an even greater benefit to moving small groups around rather than the one big mob (assuming they want to spread their units.) Yes splash may have to be bumped up to compensate, but balance shouldn't get in the way of giving a player better tools.
How does group movement apply to Tanks and overkill?
It's quite simple really. Overkill + more powerful tank damage means that the Terran's opponent encourages their opponent to spread out their troops more. However, this by itself is not of much use if the player does not also have the proper tools in order to spread out. Giving the player 2 methods of spreading out their units (and keep them spread out) as well as keeping the ability to clump allows them to respond exactly they way they want.
Conclusion.
If I haven't convinced you that Overkill is necessarily 'smarter-ai'/ the better choice, then perhaps you can see that it is more a design choice rather than the ai being 'smarter.'
There is a tremendous trade-off when splash damage can be evenly distributed with greater ease. The cost is in damage, price, and supply so that not nearly tanks can be made and the ones that are, aren't as impressive.
I don't think Overkill will solve Deathball attacks per se. But I do think that Overkill forces the Terran to spread out even more which in turn forces their opponent to spread out even more. However, it's not enough to simply promote a certain behaviour if the tools are not in place that will allow the player to respond appropriately. That's where adding a couple extra options in troop movement allows players to spread out their units better.
Overkill was part of what made the Terran race distinctive. Tanks and mines could lay waste to enemy armies with powerful splash damage... or be turned against themselves. There still exists friendly fire damage, but without Overkill, the distinctives are minimized.
TLDR
I'm really, really sorry. This was supposed to be a short addendum.
Now I can see why I didn't add this to my original blog as it is equally long.
