There Be Dragons: Slaying the Deathball

This thread has a issue that the devs should take into consideration. definetly. its just not fun to see everyones blobs fly around destroyign everything.

A few clarifications for common misunderstandings by respondents in this thread.

First, a large group of units by itself is not a deathball. This large group must have an explosive combat effectiveness in order to be a deathball.

There also appears to be a common misunderstanding about the absolute strength of larger forces. A larger group is always going to have a higher absolute strength than a smaller one. However the key point is that the larger force's efficiency should be lower. The marginal gain from adding each additional unit should be less than from adding its predecessor.


And veta's claim that TA is a high HP low DPS game, yet Starcraft II is a high DPS low HP game is just bloody bizarre. It is the reverse. TA had low HP and high DPS, and Starcraft has high HP and low DPS. This is readily verifiable fact. Veta is just completely wrong, and evidences a very thorough lack of understanding and comprehension.

Even in Starcraft, those units that have high DPS and low HP are not used in deathballs. Marines must be dropped around the map, and cannot fight a Stalker-Colossus deathball, or an Infestor-Broodlord or other deathball directly in a massed engagement. Zergling harassment and run-by tactics, as well as gambits like hellion or banshee harass, or oracle harassment. These units are high DPS and fragile compared to other SC2 units, and they are not used in a deathball.

Units with high HP are used in deathballs, such as ultralisks, thors, and colossi because it takes a long time to kill one member. And units with long range and/or splash damage are used in deathballs, such as siege tanks, broodlords, colossi, and tempests, because they can deal more damage in less time when grouped, maximizing their utility from their range advantage or overlapping splash.

Certain units in TA have relatively high HP compared to others, and certain units in Starcraft have relatively high DPS. However the way the game is actually played TA units were destroyed quickly, and units in Starcraft are more durable. The fact that Starcraft units are more durable actually leads to deathballs which stack DPS to overcome their high HP quickly. Simply because battles are over in 5 seconds does not mean the constituent combatants have high DPS and low HP, especially if overlapping splash damage, and other mechanics that run away (like spellcasting) are at issue.


For those who suggest that splash damage discourages deathballs, first it doesn't necessarily make it less efficient for large forces to engage together. And secondly it creates the possibility of mandating a deathball of splash damage dealers.

Brood War got rid of the deathball not by introducing splash damage alone, but by introducing highly positional units like lurkers and siege tanks. Three lurkers at the top of a ramp can hold against an unlimited amount of MM. You don't need a deathball of lurkers (and in fact gain little for doing so). And this prompts the MM player to split their units across the map as well. Just splash damage may produce a Protoss Colossus case. Or worse, a SupCom 2 mobile artillery case.


As for territory control, I completely agree that giving players an incentive to spread economy and industry out across the map creates an incentive to spread your forces. However that incentive pales in comparison to the reverse incentive created by an opposing deathball. If your forces will be completely combat ineffective against a ball, then it destroys any incentive you had just to be out on the map with them. Except for expendable vision, you will keep your combat forces concentrated for the entire game. Because to do anything else, even only losing a few units for free here and there, jeopardizes the entire game when eventually the deathballs collide.

A good strategy game does create the possibility of taking advantages for free, and of losing advantages for free. However it is best to avoid having this happen when large forces and small forces fight. A better dynamic would be harassing an army (not just economy), or picking off outriders or cutting off reinforcements.

This was mentioned on the first page, but not strongly enough:

If a game is to have greater strategic value, it should allow for skirmishing and pitched battles, but also for a war of attrition. You should be able to hold an area and fight for it using the economies, leaving the actual forces you deploy as the executive detail. A war of attrition would pit each players' multitasking and economical management skills against each other in a very different although also valid way. Plus it wasn't done well or at all in most RTS games, possibly with the exception of some matches in Company of Heroes, even though the pace of battles in that tended to shift around just a smidge too fast.

Wait... what?

Can you show us those facts then? Because its absolutely not true. StCII is a lowhp, high dps game. Stuff, even in low numbers, dies in mere seconds. If units would die even faster you'd call it instakill.

Wait, marines aren't used in deathballs? MMM was so devestating especially because you could do such an easy deathball with them. And your examples of colossi and Infestors? Those were splash units especially designed against deathballs of small units.

Starcraft units are durable? In which world? Even the highest hp units such as Ultras, Battlecruisers or Carriers die extremely fast, even when fighting against their own type.

And marines vs. zerg battles that are over in seconds without any splash in play? Or marines vs. zealots, or any early game units...

Broodwar and normal StarCraft didn't had Deathballs simply because its pathfinding was so bad that deathballs weren't possible by definition. You just couldn't concentrate enough units in a small enough area to do enough damage so death balls weren't the only way as it is in Starcraft 2.

Simple example, Siege tanks in normal SC vs. zerglings and the same thing in SC2. In SC1, unless you microed them terrific your zergling would get destroyed in no time. In SC2 on the other hand, the extremely similar units work out differently. You can mass zerglings so easily and their pathfinding is so good that they can reach a siege tank force and take them out, which wouldn't have been possible in SC1.

An interesting side-effect of the simulation aspect of TA style games is that unit maneuverability becomes a very important factor in the responsiveness of deathballs. Since all but the quickest units are still limited by turning speeds, acceleration, etc, the larger the group of units there is, the less easily it can react to battlefield situations. Referencing my previous example, a small group of units in formation is a bit harder to flank because they can reorient relatively quickly and bring their full firepower to bear on the attackers. Larger groups have a longer distance to move for the units that are on the opposite side of the attack, meaning it takes them a LOT longer to reposition to most effectively combat the attackers.

In deathball-heavy games, unit response and repositioning is quick enough that the damage inflicted is usually not a sufficient reward for the cost of splitting troops. In a TA-style game, this type of attack is far more rewarding, especially if multiple groups are involved. Several small armies composed of less total units than the deathball can actually do far more damage, simply because the deathball just can't react fast enough.

(This is, in fact, one of the key components to a military tactic known as Defeat in Detail.)

So provided that PA follows the mobility concepts of TA and Supcom, and makes formations effective as those in Zero-K, it will have already gone much of the distance to solving the deathball issue.

ledarsi talked about 2 types of Deathball :

1. Specific Deathball combinaison
2. Massive quantity of unit Deathball

2 Can be prevented (people have covered this point pretty good in this or other topics)
=> range, HP, non stacking unit (for air), friendly fire, splash damage...

1, however, can not. Whatever you do, one day, a smart/lucky/smart&lucky fellow will hit the jackpot by finding THE ultimate combinaison.
You nerf this combinaison ? Something else will be found.
Nerf ? Don't care, here is the new solution.
It's like trying to nerf a faming spot/build in MMORPG. Not possible.
That's the terrific true with competitive/multiplayer games, where victory or epeen are displayed to the world : "fame" hungry people will always find a way.

I think you are underestimating the impact that spread out resources-points would have on deathballs. A deathball is very immobile and the deathballing-player would quickly have extreme trouble with small raids all over the place. This in turn would lead to his own army being rather small. So deathballs make no sense when you have to hold a large area. So people will spread out and only gather forces when they want to do a big attack, in which case a rather big army will of course be better.

I really don't see the point of your long explanations. While I disagree on some of your SC2 and SupCom2 things and quite a few other things there is not really a need to discuss. PA won't have deathballs anyway, since neither TA nor FA had deathballing and I can't see why UBER would change this on purpose. Should it turn out that first versions of PA's balancing end up with deathballs UBER will probably try to patch that away.

Siege tanks are in no way similar between Brood War and Starcraft II. Even ignoring their higher gas cost and lower damage, they cost 50% more supply than Brood War tanks. You can have 50% more siege tanks in Brood War. Increase the number of anything by half and it will perform much better against anything, even if it is tanks against zerglings.

You want evidence that TA was a high DPS, low HP game? Very well. Let's take a look at four units from TA, and four units from Starcraft to compare their HP and DPS cost efficiency. You will find that TA strongly encourages building cheap high DPS squishies, and that Starcraft encourages building expensive higher HP units.


Total Annihilation: Crunching Numbers

Peewees cost 45 metal and have 300 HP ('maxdamage' in profile). However each EMG shot deals 11 damage (default). And the EMG fires in bursts of three with a reloadtime of 0.31... (for some reason this number has 14 digits) seconds. So a Peewee has about 100 DPS for its 45 metal, with considerable inaccuracy. However it has only 300 HP for its 45 metal. For each metal you're getting a little over 2 DPS, and bit over 6 HP. A peewee is very obviously a low HP, high damage squishy where you are primarily paying for its weapon. But as you can see it is actually an extremely efficient source of both DPS and HP, despite its low HP.

A similar unit to the Peewee that differs markedly in terms of being a high DPS squishy is the Warrior- a "medium" kbot. The warrior costs 248 metal and has 1300 HP, for 5.24 HP per metal, which is excellent. I seem to recall the TA warrior has an EMG and a plasma cannon, whereas the BA one simply has a laser. Either way, their DPS is functionally similar. The BA Warrior's laser deals 55 damage and fires once per 0.3 seconds, for 183 DPS, which leads us to 0.738 DPS per metal. Note how much more this unit costs than the 45 metal Peewee, and how much less DPS efficient it is. Smaller, cheaper units are more efficient in TA.

By contrast, let's look at an exceptionally durable unit from TA. The Bulldog. The Bulldog heavy tank costs 844 metal, and has 4200 HP. Its weapon deals 270 damage (with small splash) per 1.12 seconds for a DPS of 241.07. For each metal you're getting a bit less than 5 HP, and only 0.286 DPS. The Bulldog is very obviously a durable unit with a weak weapon. However note how inefficient and costly this unit is compared to the Peewee or other cheap, high DPS squishies. This cost-inefficiency is why TA is primarily played with high DPS, low HP units.

Now how about a real DPS squishy that isn't efficient? The Penetrator costs 2985 metal, and has 5500 HP. It has a weapon that deals 12,000 damage which only fires once every 10 seconds, for a whopping 1,200 DPS. This unit gets 1.84 HP per metal, and has an incredible DPS, but its DPS per metal is actually only 0.4. Still, they don't call it the "blue laser of death" for nothing- 12,000 damage will straight up gib almost any target in the game, a neat functionality which you are also paying for in inefficiency, as well as its triple range. The Penetrator, if it were more efficient than units like the Peewee, could be a deathball unit. But is is vastly less efficient and serves as fire support to an army of efficient main combat units.

Long story short, weapons are cheap (and efficient) in TA. HP is expensive. As a direct result, TA emphasizes high DPS and low HP units. For cost-efficiency, the small high DPS squishies rule TA.

(unitdata from Balanced Annihilation 7.75- I couldn't be bothered to unpack TA's HPI files to get at the .fbi unit definitions. However they should be largely the same as in BA)


Starcraft II: Crunching Numbers

In Starcraft II, HP is relatively cheap and DPS is expensive. Even very expensive units often have quite low raw DPS for cost, but may compensate with other features such as range, splash, or special abilities.

As with TA, we start with the cheap high DPS squishy; the Marine. A marine costs 50 minerals, has 45 HP, and attacks for 6 damage once per 0.86 seconds, or about 7 DPS. This means it has about 1 HP per mineral, and deals about 0.14 DPS per mineral. Note that these numbers have different significance between games; minerals and HP are both "larger" than in TA. Numbers for TA are in hundreds, or thousands, not tens or hundreds. So the systems differ, but the same cost analysis works internally on each.

Proceeding in the same order as we did for TA, the Stalker costs 150 minerals and 50 gas, and has 160 HP (80 shields and 80 hull). Its weapon deals 10 damage (+4 Armored) for about 7 DPS (about 10 vs Armored). Now the binary resource system, the damage types, and the nature of regenerating shields makes this unit harder to number-crunch than TA units. However it should be intuitively obvious that this is a high HP unit, but isn't the most efficient HP source. And it's a low DPS unit that is a very inefficient source of damage.

Vespene gas as a second resource complicates the calculation, but as long as your calculations are uniform it will produce results which can be usefully interpreted. For now let's simply add the mineral and gas costs together and consider that the unit's "cost." So a Stalker "costs" 200 Schrute Bucks and has 160 HP, or 0.8 HP per cost. And it has 7 DPS (10 vs Armored), so that means it has .035 DPS per cost (.05 DPS/cost vs Armored). As expected, it's an alright HP source with high HP, and it's just a really bad DPS unit on both counts.

Now for a seriously tanky unit. Consider the Ultralisk. The Ultralisk costs 300 minerals and 200 gas, and has 500 HP. It has an attack that deals 35 damage per 0.86 seconds (with 33% splash) for about 40 DPS. Adding its costs together gives us 500, so the Ultralisk has a cool 1 HP per cost, and 0.08 DPS per cost. Obviously the unit is hard to kill since it has 500 HP, but it is actually only about as HP-cost-efficient as the Marine.

Now, the Colossus is an expensive DPS unit, much like the Penetrator. It costs 300 minerals and 200 gas, and has 350 HP. Its weapon fires twice at 15 damage every 1.65 seconds, for 18.2 DPS. Therefore it has 0.7 HP per cost, and 0.0364 DPS per cost. The Colossus is the absolute least DPS-cost-efficient of the four units we have analyzed in Starcraft- worse even than the Stalker. Although the Colossus also does full, multiple-target, overlapping splash damage, it is still incorrect to say it is a low HP unit with high DPS. The Colossus is a high HP, high DPS unit. Furthermore, its DPS cost efficiency is actually dreadful, even though in quantity it kills the enemy in a short period of real time. The Marine, by contrast, is a high DPS, low HP unit. And the Marine is actually more HP and DPS cost efficient than the Colossus, but is countered by it (different story).

The Colossus is a high HP, high DPS unit that relies on its range and splash more than its absolute DPS. While just one Colossus is an expensive and fairly durable DPS paperweight, a group of them will deathball together and kill the enemy before they sustain any casualties, or at least only nominal casualties (i.e. chaff zealots in front). Despite its very poor DPS cost efficiency, Colossi when grouped can achieve enough stacked DPS to prevent the enemy from having enough time to get through the HP of a single Colossus. That is why the Colossus is a deathball unit.

You should have noticed a trend in the above number crunching. That TA makes cheap units more efficient, and that TA makes DPS units more efficient. Whereas Starcraft makes cheap units relatively less efficient (although still more efficient), thus encouraging greater use of larger and more expensive units.



On Immobile Deathballs

This is an assumption which is false. Many RTS games balance strong units by making them slow, and while this is a good balancing factor used in moderation it is bad to have as a universal rule.

It is simply not the case that there were no deathballs in SupCom/FA. T3 armies and experimentals were almost exclusively used in deathballs. Early game was very dynamic in SupCom/FA, but higher tech units did benefit immensely from being deathballed.

It is also not the case that deathballs will always be slow. Air superiority fighters in SupCom/FA, and to an even greater extent in SupCom 2, were a deathball unit that was highly mobile. However they only attacked flying units, meaning you could seize air superiority with ASF's but not win the game with them directly by attacking the ACU.


On Marine Marauder

In Starcraft II marine marauder "deathballs" are a non-issue in serious play. MM is highly efficient in small numbers, but you only see huge armies of bio units A-moving into the enemy in very low-level play. If you are at the level of play where you lose to "deathballs" of marine marauder, then I guess your concerns make sense, even though you could learn to deal with MM.

However at high levels of play these units get crushed by real deathballs. Multitasking drops and harassment are the way these types of units are used- you will never win a straight up fight using MM against a genuine deathball. The same goes for a variety of other relatively high DPS units, including zerglings, mutalisks, hellions, banshees, phoenixes, oracles, and so on.

I don't see the argument that leads to my assumption being false. Deathballs are bound to be slow and being slow means that you can not hold a large amount of mass deposits that are scattered all over multiple planets. So a deathball is not a valid strategy to hold your resource, and therefore you will need to have raiding and smaller skirmishes all over the place.

Air is pretty different from ground and I can't see how to not make it deathball all the time without proper collisions between airunits. They just have zero pathing issues and are just as fast in large groups as they are alone.

Experimentals in deathballs? T3 army in deathballs? wtf? your idea of deathballs really seems to expand over to any kind of "big army is more effective than small army". Which has nothing to do with deathballs.

Where do you draw the line between a deathball and a "big army"?

Whole thread: tl;dr

A deathball is an army that has enough DPS that it can prevent itself from sustaining any casualties, or at least only nominal casualties.

The actual number of units is not important. A Starcraft deathball might be only 30 units, whereas a vanilla TA Samson deathball had hundreds of Samsons. Very large groups of low-DPS units can be used to overcome an individual's low DPS, but a small number of exceedingly high DPS units might also be a deathball if the enemy can never close to range. In fact, very large units are more conducive to being used in numerically small deathballs. For example, Fatboys.

So 10 Mantis are a deathball relative to a single enemy Mantis?

It really seems pretty arbitrary to me where you want to draw the line between "deathball" and "big army", because a big army vs a very small army (big and small relative to each other) will always only suffer minor to none casualties. Do you want to get rid of big armies all together?

I can't remember which unit was the samson in TA, but I definitely can remember that houndreds of units in TA clumbed together very inefficient. There is no way any army of 100s of units in TA would be able to deathball. That is not possible in FA either.

I think I see where the main confusion is arising. You might say that 10 mantises is somewhat similar to a "deathball" when fighting just a single Mantis, because the larger group kills it and loses nothing. However higher absolute strength is not the same thing as having strength that runs away with scale such that no force except another deathball can inflict casualties. 10 Mantises is not a deathball. The other player can easily build, deploy, or reposition a force that can at least inflict casualties on it if not outright beat it, even if only by adding a couple Mantises and still fighting outnumbered just to shrink the force. I have no problem with 10 vs 1 Mantises winning with zero casualties. It is when no force except another deathball can inflict any casualties, due to a force's strength running away with scale, that there is a problem.

In a nutshell, suppose it is 1 vs 10. But they don't fight, and both sides add 5. Now its 6 vs 16. The larger force's advantage has shrunk, not grown as if it were a deathball.

The Mantis is a very poor deathball unit because as the number of Mantises in a force increases, the marginal utility from each additional Mantis drops. For small numbers of Mantises, grouping makes more sense than sending individual Mantises out on the map, for precisely that reason; a force of Mantises will kill a single one for free. But the important question is whether the Mantis' strength runs away with scale. Does having 500 Mantises mean 50 Mantises will inflict zero casualties? They are outnumbered to the same proportion. But the Mantis doesn't run away, and 50 Mantises will perform better against 500 than 1 did against 10.

The 11th Mantis adds less strength than the 10th one did, not more. The 11th Mantis undoubtedly makes the group stronger. But as the number of Mantises increases the marginal gain from each new Mantis drops, until eventually that Mantis would be more useful in a separate group than adding a 200th Mantis to a single group. Instead of a single group of 200, it would actually be stronger to have two groups of 100 (and probably to subdivide further still). At a certain point a very large group of Mantises becomes too inefficient to use.

For deathball units, every additional unit adds more power than its predecessor, not less. And this explosive power gain means you are essentially never going to want to split your units. And it makes the deathball effectively impossible to defeat without using your own concentrated deathball.

A deathball is a blob of units that can kill enemy units before it takes significant losses in DPS. Not casualties in general, in fact the nature of the StarCraft 2 deathball is to take damage with gateway trash units (stalkers, zealots) and never take a major hit to its DPS dealers which are low hp high dps Colossi, thus allowing you a maximally powerful army while constantly replenishing fodder.

This is an important distinction to draw as when a player has no poorly designed high DPS units surrounded by fodder this dynamic doesn't work. I think what you have a problem with is the low hp high damage colossus unit, not the Stalker which kills its equivalent in 16 shots.

Also you pick a random unit from TA to exemplify your point of low hp high dps, the reality is most of TAs units took more that a few seconds to kill their equivalent. And nobody was marching armies in full of peewees. What would be more intellectually honest would be to take a staple unit from TA, like a tank and then consider that its EHP is augmented by the the wrecks that formed on front lines.

Let's take an example from Forged Alliance since I can't find a detailed unit guide for TA atm and you probably won't appreciate Zero-K units which are even higher hp to dps:

http://faforever.com/faf/unitsDB/unit.php?bp=UEL0202
Here's a T2 Pillar Tank - 1500 hp, 54 dps, 28 seconds to kill another T2 Pillar Tank with perfect accuracy

Generally a squishy unit is anything that kills itself in under 10 seconds, you want a wide variety of units but your staple units should leave time for tactics, I don't think it would be unreasonable for all staple units to take 30+ seconds to kill each other when inaccuracy is factored. Honestly it might be better if it took staple units 30+ seconds to kill their equivalent because if PA is as epic as it is anticipated to be you will have a lot of these minute-long battles around the system simultaneously.

Balanced Annihilation is obviously very different from actual Total Annihilation. Here's the stats from unmodded TA:

Peewee
HP: 250
Damage: 8x3 every 0.4 seconds (60 dps)
Cost: 53 M, 697 E
http://units.tauniverse.com/?p=u&v=158

But using the Peewee is not a good example in any case; they died too quickly in any major fight to actually bring their DPS to bear.

Bulldog
HP: 2102
Damage: 147 every 1.4 seconds (105 dps)
Cost: 467 M, 2994 E
http://units.tauniverse.com/?p=u&v=193

Warrior
HP: 850
Damage (cannon): 60 every 2.13 seconds (28.17 dps)
Damage (emg): 12 every 0.2 seconds (60 dps)
Cost: 196 M, 2236 E
http://units.tauniverse.com/?p=u&v=89

Warriors were also buggy and didn't always fire both weapons.

Penetrator
HP: 650
Damage: 1300 every 6 seconds (216.67 dps)
Cost: 2604 M, 17477 E
http://units.tauniverse.com/?p=u&v=107

This was a very bad unit. Large explosion (nuke) when it died and very slow turret turn. It's also incredibly different to the Penetrator you describe in Balanced Annihilation.

The most common unit played was the samson (and core equivalent). Lets look at stats:

Samson
HP: 650
Damage: 40 damage every 2.5 seconds. (16 DPS).
Cost: 119 M, 1027 E
http://units.tauniverse.com/?p=u&v=31

It would take 40+ seconds for a 1v1 samson battle to finish, assuming no shots missed. Very obviously High HP / Low DPS. And while it would appear a Peewee would kill a Samson in 1v1, the truth is that once the fight became a few samsons vs a few Peewees, the Peewees would lose very quickly as 6 samsons could 1-shot a Peewee, and so several would die before they got into range.

Samsons would take 40+ seconds to kill other Samsons with perfect accuracy, what you're really looking at is a unit that would take much more than a minute to kill its equivalent on a wreck-laden battlefield. Those are some high EHP to DPS units and indeed make for strategic gameplay.

Units like the Peewee are tactical by nature. Their HPPS ratio and doesn't permit them to be used strategically in a battlefield. Peewee units have their place but it isn't as a unit that makes battles memorable or epic.

Peewee's were great fodder and scouts.

Samsons and Peewees would allow the longer ranged Sampson to benefit from the quick scouting of the peewee for little cost.

And isn't the way to deal with a death ball a AOE weapon?

Edit: Finally read through the OP, and other then the indication that you more or less insulted every RTS game that isn't TA or SupCom, it seems alright.

The fix? Don't have samsons, don't have a unit with a fast fire rate, tracking missiles, long range, AA targeting and be cheap enough to rapidly produce.

My fix? IF we are talking in TA terms then reduce the fire rate of the samson, and make the unit move slower, leaving them easy to swarm and thus be ineffective.

Thus making the easy to stack unit, poor at stacking.

I didn't realize Balanced Annihilation differed so substantially from OTA. To the extent that a bulldog's cost is double its cost in OTA. The main reason I used its numbers instead of Total Annihilation's is I have ready access to its actual unitdata files, whereas I would have to install and unpack TA's HPI files to read those.

Still, the basic trend is the same. The Peewee is still a squishy high DPS unit that is highly efficient for both HP and DPS, the Warrior less so, and so on down the list. The cost-efficiency numbers I wrote before are obviously incorrect for vanilla TA, but their relation seems to be preserved.


I will cede the point on the Samson being low DPS. It is a low DPS unit. However the thing about the Samson and Slasher is they are the poorest-designed OTA units for the reasons I describe in the OP. They have low DPS individually, but due to their superior range and stackability they can mass together into a kiting blob. As a result missile truck spam was a bit of an issue with vanilla TA in some circumstances.

As you say raevn, 6 Samsons can 1-shot a peewee, and can do so from range. More trucks can 1-shot even bigger units from range, and they can move and fire to kite even extremely large, heavy units. They are low DPS units that stack, and as a result hardcore MT spam was a recurring gameplay issue in TA. Uberhack eventually just axed the unit by making MT's only attack air. It is my assertion that this type of stacking low-DPS units into a ball is exactly the kind of play that should be avoided, even though it had some limited presence in TA, it wasn't unresolvable.

Like I stated earlier- implement full friendly fire as is in the Spring engine, and Samsons can no longer all fire from a blob, as missile trucks required a straight line of fire.

While true, the actual use of the units changes to the point where the comparison is moot (eg. Peewee's not being used once larger forces are assembled).

I agree completely that this is a problem, and I understand that deathballs are undesireable. My only real point I'd like to contribute is that low HP, high DPS units should not be the way to solve this, due to the fundamental shift in gameplay towards tactics that would occur, and the removal of most strategic options with regards to manouvoring. Making samsons low HP high DPS wouldn't actually solve the issue anyway (it would actually make it worse by increasing front-loaded damage), as their specific cause for deathballing is being all-round units (AA and anti-ground), long range, homing and effective in groups.