[D] Swarm host is bad

At the very first presentation of HotS and the Zerg units at BlizzCon 2011, the swarm host was presented as a siege unit, that both allows the Zerg to be aggressive even when the opponent is turtling, and presents a strong enough danger so the opponent shall be forced to move out of their fortress and deal with it directly. To make it feel "zergish", Blizzard decided that this would be best represented by a neverending stream of free units.

Later on, in the Battle Reports for the HotS Beta, the actual role of the swarm was less that of a direct siege unit as more of a support siege unit, that would add free units as a buffer to other high-damage-dealing ranged units of the Zerg arsenal.

The swarm host is a unit that is available with the infestation pit. It is a unit that can burrow without burrow upgrade researched, and when burrowed, will spawn locusts, which are like brood lings. Those locusts die after 15 seconds (25 seconds with an upgrade that costs 200/200), and new locusts will be spawned every 25 seconds. Each swarm host creates two locusts at once.

In its first iteration, the locusts that spawned had according to Dustin Browder around 90 hp. When the first numbers were publicly available, they had 110 hp instead, and it is probable that they did so in the presentation video at Blizzcon as well, judging from them being able to tank at least two direct hits plus AoE fire from nearby tank hits. My guess would be, that they had even more hitpoints back then. Also, they started with being melee units, so they were basically brood lings on steroids.

In the battle report, the locusts had changed already. The flaws of a melee unit were obvious, as they are to every Zerg player when going mass Zergling while forgetting Zergling speed. Also, locusts could attack air units. For the HotS closed Beta, the locusts are already in the third publicly shown iteration, having only 65 hp, but slightly more range and attack to boot. (not sure about the damage though, since in the battle report they did two-shot probes. Might have been upgrades). Plus, the locusts could not hit air anymore.

Note: I do not really care for the actual numbers, as they can be changed easily. Tomorrow swarm hosts might cost 100 minerals less, or more, who knows. I am just trying to talk about what is wrong with this unit besides raw numbers. I will probably mention numbers a few time, where i deem it necessary to reiterate my point, but raw numbers are never the target of my discussion argument.

First of all, a unit with an attack that can be denied without cost already directly contradicts the concept of a siege unit. Sure, siege units can counter siege units, but they should not be able to deny completely all damage when giving the right conditions (which are easily met in this case). It takes away the whole concept of your opponent having to come out of his fortress to deal with you.

Second, locusts are short ranged units, which again leaves them with the problem to connect with the enemy. Added in the factor that they can also be killed completely and them having only 65 hp (and light armor if i'm not mistaken), Blizzard has to make up for those fundamental flaws by giving them an overpowered strength in return, in this case the attack damage. It is over the top, which makes the swarm host even more of a hit-or-miss one time one trick pony. Either your opponent shuts it down mostly/completely, or he suffers great losses. And due to the easily abusable fundamental flaws of the locusts, there should be hardly any cases of the locusts doing massive damage before they get shut down.

Third, the locust adds to the death ball concept. While area-of-effect units like the lurker forced to split the death ball, the locusts get exponentially stronger the more you stack them, and exponentially weaker the more your opponent stacks his AoE defense. Unfortunately the power of your opponents AoE grows way faster than the power of stacked swarm hosts, and soon they do no damage at all. Which makes you think about why you build them in the first place.

Forth, the swarm host is a huge investment, and will mid-game drain a ton of gas. But since one of its obvious weaknesses is its lack of anti-air and except for Queens all mobile anti-air costs a lot of gas for Zerg, you are bound to lose in case of your opponent opting for flying units. Yes, it does get created by the same building that allows you to build infestors, but you will have to get the upgrade for the duration of the locusts (they will travel at least 5-10 seconds to the opponent, leaving them with 5-10 seconds to attack. The upgrade will double or triple the effective time your locusts hit the enemy, which makes it a crucial upgrade to get. Not to forget, increased action range if you so desire). That is 200 gas +100 for every swarm host you get. As has been pointed out, they suck alone, they do better when stacked. And the same way you cannot go mutas and infestors at once right now, you will not be able to go infestors and swarm hosts at the same time off 2/3 base. Which is like the only time you would prefer those over Brood Lords.

Fifth, the swarm host does not allow for control of space on the map, it just forces your opponent to take control of said part of the map to collect your gas heavy units and go. To put it in other words: The swarm host is not a shield, it is lika a sign that says "hit me here". The locusts do not put up much of a fight in a direct combat situation if there is AoE present, and even without AoE there are easy ways to shut them down (air, forcefields to keep locusts away, kill locusts then swarm hosts, or just attack when they time out and new ones aren't there yet). In any way, the swarm host was concepted as an inferior direct combat unit, which would be okay if it would fulfill its other roles. Also, the swarm host has limited mobility and will therefore hardly escape, plus you cannot really afford to lose them. Which makes them a unit that has to be defended with all your strength, while at the same time not adding much of firepower themselves. Assuming no huge gaps in macro ability between your opponent and you, he should be coming out ahead of this just because you have invested too much in "dead supply". They are like corruptors with no Air unit on the field - they can corrupt, but they won't win you any direct fight.

Sixth, the actual unit design of the swarm host is redundant. Its concept was to give Zerg a Brood Lord pre Brood Lord, which had to be worse than the Brood Lord to compensate for earlier availablity. This lead to problems that were not forseen (as in siege units with zero range and no jump/throw ability to get to the enemy? Back in 2011 at Blizzcon i was already sure that this should never work in a real game, but did not care enough to make a thread about it, as it was only a sneak peek. And now that the first iteration of Beta started, i expect the units to be thought of by Blizzard as viable to be implemented in release, but it is still early enough that major changes can be made, as in replacing a unit completely), that are not and never will be cleared by just slightly increasing the range of the locust. For constant siege and mobility, the Brood Lord far surpasses the swarm host. For raw damage, the Brood Lord has more dps assuming the opponent does get colossus/tanks (which there is no reason not to, as they are the hard counter), because Brood Lords get at least damage done with the initial Broodling attack from the BL itself.
That in itself is not that surprising, but the direct comparison with the infestor shows the flaws of the design of the swarm host once again:
For sieging up pre-Brood Lord infestors are better because infested terrans still have more range and can be launched across a distance, plus infestors have way more burst potential due to the energy mechanism. Also, they are more mobile and posess other spells of which at least one is useful as well (probably one one of the best spells in game), when it is less useful to just get a few free units, and more useful to get AoE entangle + damage. And they can hit air.

Seventh, they do not fill the last gap in the Zerg arsenal no one talks about. I am not speaking of a lack of a direct damage AoE unit, nor for map control units or siege units. I am talking about a unit that needs instant detection to be dealt with. This is the thing i honestly miss most from Zerg in WoL, a unit that deals damage while cloaked. Any continuous damage from cloaked Zerg units right now can be mitigated by simply parking your army on top of it (i said continuous, you see what i did there? ). Some burrowed roaches? Send half your army in your main, half your army in your natural, wait for scan to be ready again. Infested Terrans? Shoot them down. There is no Zerg unit that deals damage that cannot be mitigated while cloaked. A dark templar can still do damage, even with 30 marines sitting on top of it. So can Banshees and Ghosts. But park your army over burrowed Zerg units, and they won't be able to lift a finger. A lurker could.

As i already pointed out, the locusts suffer the same problem every low-ranged unit suffers: It will do nothing if it either cannot reach its target or gets shot down before it does.
This already highly limits its use as an actual siege unit. If used against a small ramp, forcefields will completely deny the locusts, although on ramps with more than one forcefield required to wall off this is not always endurable. Furthermore, if the opponent is set in a defensive position, it cannot be broken by swarm hosts, as the locusts have only 65 hp now and will die to any siege tank or colossus fire without doing damage at all. Therefore under the condition that siege tanks or colossi hit the field already, the swarm host is sure to not do much damage while sieging a base.
Second, the locusts cannot be used for much of a buffer for the army anymore for the very same reason: They get swiped out by a single volley of 2+ tanks/colossi, and do not even allow other units to close the distance, let alone retreat unharmed.

Also, since locusts cannot hit air units, and anti-air pre-hive has not become any better at all for Zerg, any air-to-ground-unit will shut down the swarm hosts without resistance.

This leaves only the following niche for the swarm host:
- in ZvT the time before siege tanks are set up and no banshees are out
- in ZvP before colossi are out with thermal lense and no void rays are out
- in ZvZ when there are no Mutalisks (and Brood Lords) out, and only when turtling behind a wall of spines or when having a huge army advantage (since the Zerg army is highly mobile, and if not used behind a wall of spines, the swarm hosts will just get overrun along with everything else by superior enemy forces)
- in any matchup when harassing bases/positions that are too spread out and do not have the heavy defensive focus of the opponent

To go into Detail for each of those situations:
ZvT: Currently, besides the rare case of pure Bio play only (in which case Zerg might want to get Banelings/Infestors/Vipers instead of swarm hosts) pretty much every other unit combination Terran goes for includes rather fast siege tanks. Plus, a huge portion of Terran openings include getting a fast Banshee. The only situation where swarm hosts should be able to do any damage is when the opponent is way too greedy, in which case the swarm hosts do come way too late into play, are too immobile, plus they do not punish weaknesses fast enough when opportunity presents itself. The nibbling-away-at-the-opponent seems to not be viable once the opponent is set up.

ZvP: What are the current unit compositions the Protoss players aim for? Stargate harassment, (fast) colossus (into push or late game), sentry/immortal two base play, blink stalker two base play. This leaves you with the following options:
- you will face a defend-or-die situation when barely finishing saturating your three bases against a two-base-allin, in which scenario infestors tend to come too late (and swarm hosts for that matter). Plus, swarm hosts are aweful for defending any of the two-base pushes, as forcefields wreck locusts, colossi do as well, and blink stalker can either blink on top of swarm hosts or just blink elsewhere and keep on killing you.
- the Protoss will somewhat turtle. If he really goes for a third without colossus, you can try to punish him with swarm hosts, but then again this is the same scenario as in ZvT vs a very greedy Terran: About any other option to punish too much greed will be more effective than slowly nibbling away at the opponent.
- you go one/two base yourself and try to attack the Protoss. In this case, it plays out similar to the Destiny Ling-Infestor bust on two bases, with the slight differences that you cannot throw your units over your opponents wall, plus you are only chipping away at the wall. Slowed down due to forcefields (which the Ling-Infestor bust circumvents using the range of IT), your opponent should be able to hold out long enough for colossi to come out, in which case he will have superior economy an be ahead in tech, effectively losing you the game. I am fairly certain that the upgrade for the locusts will cut too heavily into your swarm host count, and without it, your locusts will do only about a third of their potential damage (if you substract travel time). Which makes braking walls hard. Again, i theorycraft, but for me it all comes down to the conclusion that this is not really a viable strategy. And against FFE into Stargate or really fast colossus you are screwed anyway.

- In ZvZ, the currently used strategy at lair tech is often about getting out a couple of Mutas (~6-10) to clear the map of overlords, then switch to infestors to help in roach vs roach battles (as key fungals + infested terrans from infestors with much energy give an incredible boost to army strength). Against mutas, building swarm hosts first will auto-lose the game, since Queens will be the only affordable anti-air along with Spores (no Gas), therefore forfeiting any map control while leaving you with units that are only useful when constantly nibbling away at your opponent. Against infestor-based openings, your opponent will have more burst power than you have (his roaches and infestors will kill your ground force, and the locusts do not give you immediate burst but constant free small fries). So if you did not get miraculously ahead, you forfeit either the map and/or direct combat strength, which will in direct return lose you the game by economy or in combat. Therefore, swarm hosts should in theory be only not game-losing when you are either way ahead and use them to keep your opponent down, in which case you might as well just go steamroll, or in a situation where you can get a ton of static defense in front of your opponents base (to make up for your lack in direct engagement power). Which is cute, but highly unlikely, since your opponent should see your spines and swarm hosts wiggle over the map once they leave your base. And he will laugh.
Later on, Brood Lords should be prefered over swarm hosts anyway, due to attack range, being a flying unit, moving while attacking, creating free units more frequently, brood lings being faster etc, you know

This leaves us for the last possibility of use for swarm hosts, the case where your opponent is spread too thinly.
In this case, again, other units might be way better at punishing this (as a Ling runby costs less and will probably do more damage than few locusts trickling in now and then, and an infestor hitsquad will also give you more bang for your buck while being more mobile themselves, plus costing less), and if you commit too many swarm hosts while being spread out yourself you just make sure to lose them the fastest way. Sure, swarm hosts CAN perform this role, but there are other Zerg units in the arsenal that can perform this role way better.