Cogmind > Strategies

Alpha 11 combat strategy

(1/5) > >>

Decker:
This post discusses the combat strategy for Alpha 11. First I examine the causes of death, then I analyze the item availability rates and churn rates, which explain why some items are more valuable than others. Finally, I compare the merits of the main combat utilities and the weapon types. Per Kyzrati's request, I avoid discussing the specifics of the less useful utilities.


Causes of death

You can win a thousand fights but you can only lose one. Before we dive into technicalities, let's take a moment to look at the big picture.

Usually we can distinguish between two types of death. Death by stripping, and death by core attrition. The former happens when you run out of essential items - power, propulsion, weapons, matter. Then you're no longer combat-capable and you're forced to flee half-naked. The latter happens when all your slots are filled, your inventory is mostly full, but your core gets destroyed through sheer attrition. In other words, you fought for too long.

There is some tension in avoiding either type of death. If you equip more storage units, then you're less likely to be missing an essential part at a critical moment. However, equipping storage units reduces your mobility and your combat effectiveness. These are two of the key factors that prevent a death by attrition.

Deaths by attrition have many causes, but fundamentally it all boils down to taking too much damage for too long. For instance, this can happen when you cannot find the exit and yet you've explored most of the map. This can also happen because you're pinned down and unable to make forward progress. By the time you finish clearing an ARC dispatch, another one is already on its way. In either case, moving and killing fast are your best shot at survival.

The different phases of the game call for different builds. Materials and Factory favor high storage and high coverage. Research and Access favor mobility, killing power and combat avoidance. The dirty little secret of the late game is that the less you fight, the better off you'll be. It's worth considering dumping storage, going stealthy and/or sacrificing combat potential for exit-finding capabilities.


Item availability rates and churn rates

There are two primary sources of items: stockpiles (including haulers), and items dropped by destroyed robots. Stockpiles typically provide items of better quality, since their items are undamaged and are often prototypes. They are thus the preferred source for restocking.

However, stockpiles can only be tapped into when you're not actively fighting. When you are fighting, you must survive on the items in your inventory or those from the battlefield. This distinction has far-reaching consequences for the viability of using certain items. If your build relies on a certain item that is not commonly found on the battlefield, then you can easily get into trouble if you're forced to fight for an extended period of time.

This issue can be mitigated somewhat by having a large inventory. You consume resources in a burst while fighting, and then you restock from stockpiles when there's a lull in combat. However, inventory space is a very limited and precious resource. Hoarding one type of item is necessarily done at the expense of other vitally important items. The lost opportunity cost must be considered carefully here.

Furthermore, the time it takes to restock one type of item is governed entirely by its statistical distribution in and outside combat. Stockpiles are entirely random and there's no guarantee on how long it will take before you find what you seek -- perhaps never. Conversely, items that drop in combat are much more reliably available. For a combat-oriented build, fabricators are not reliable sources of items for a variety of reasons.

The upshot is that items that are not commonly available during combat may become liabilities when your build depends on them. This does not mean that those items are useless, but rather that they should be thought of as nice-to-have rather than must-have.

The counterpart of the item availability rate is the churn rate -- the rate at which you burn through an item type while fighting. The churn rate is dependent mainly on five factors.

* The coverage of the item versus the coverage of the other items you have equipped.
* The integrity of the item.
* The number of items of that type that you have equipped.
* Your effectiveness at dishing out damage (a good offense is a good defense).
* Your effectiveness at defense (mobility, dodging, resistances, force fields).
An important metric here is the integrity/coverage ratio. A low ratio generally means that you'll be burning through that type of item quickly. That's a bad thing for several reasons. Replacing items in combat and restocking thereafter take valuable time. To compensate for the high churn rate, you'll have to hoard several spares in your inventory, and there is a lost opportunity cost to this. Finally, after fighting for an extended period you'll inevitably be left with a depleted inventory and a slot to fill, which is a dangerous situation to be in.

There are many items that have a low integrity/coverage ratio. Unless the coverage is very low (<= 5) and so the low integrity doesn't really matter, they are best avoided. Whatever benefit they provide is not worth it if they'll be gone after taking a few shots.

It is useful to consider the effect of the churn rate and the availability rate on the slots you choose to evolve. It's safe to evolve many propulsion slots since propulsion parts are ubiquitous and many have high integrity. Stacking propulsion parts for use as makeshift armor is one strategy that can work both for a stealth build and a combat build. Conversely, it's dangerous to overdo weapon slots, and to a lesser extent power slots, since those have significantly higher churn rates.

The churn rate makes some kinds of item more valuable than others. Cogmind sheds reactor, propulsion and weapon parts like a human sheds skin. In a combat-heavy scenario, a good gun might carry you through half a level. Thus finding a really good gun doesn't usually have a major impact in the long run. However, a good set of combat processors can last for half the game, with a bit of luck. If you can get anything fabbed, it's a good idea to prioritize building combat utilities.


Firepower

Your firepower is governed approximately by this equation:
  firepower = # guns x gun damage x cycler speed-up x hit rate x core-analyzer/crit/armor multipliers

Notice that those factors contribute multiplicatively with each other. Thus the fastest way to increase firepower is to pick the next biggest increase in some category. I've organized the following table to make this easier to do. Each value reflects the relative increase in firepower compared to the previous state. Of course there are also other factors to consider, which I discuss after.

Weapons (after 1) +33%, +38%, +24%, +17%Imp. weapon cyclers (-25) +33%, +50%Adv. targeting computers (+8) +16%, +14%, +12%Adv. core analyzers (+10) +15%
To compute the effect of targeting computers I assume that the base hit rate is 50%. Targeting computers are more useful against hard-to-hit targets, and less useful for bigger and closer robots. Core analyzers work best against robots with low core exposure, and against very small robots where half the weapon damage is sufficient to destroy the core. In the table, I assume that they're being used against big targets with a core exposure of 25%.

Math for the interested reader.
Spoiler (click to show/hide)Equipping the next gun yields (2/300)/(1/200) = 1.33, (3/325)/(2/300) = 1.38, (4/350)/(3/325) = 1.24, (5/375)/(4/350) = 1.17.
Equipping the next cycler yields 1/(1-0.25) = 1.33, (1-0.25)/(1-0.5) = 1.50.
Equipping the next targeting computer yields (50+8)/50 = 1.16, (50+16)/(50+8) = 1.14, (50+24)/(50+16) = 1.12.
Equipping the next core analyzer yields (0.25 + 0.75*0.1/2)/0.25 = 1.15

Kinetic weapons

Arguably the best type of gun. They make short work of grunts, programmers and behemoths, and they're mostly neutral against the rest, with the notable exception of hunters. Their main drawback is recoil and matter consumption. Having treads nullify most of the recoil penaltly. Since kinetic guns consume the least power and generate the least heat, they can be stacked without requiring a ton of heatsinks. They also tend to have more integrity than other types of guns.

All in all, they have lots of advantages for only one major penalty, matter consumption. This is usually not a big issue provided that Cogmind keeps a spare matter storage unit at all times in its inventory. If Cogmind has 2 weapon cyclers, then kinetic guns become even more valuable due to their moderate resource requirements.

Energy weapons

Energy weapons don't have much going for them nowadays. They're good against programmers, and mostly neutral toward the rest. Their heat injection property is no longer an asset. Generations of melted robots has resulted in all robots now being equipped with good heat dissipation capabilities. Their main selling point is that they don't consume matter, so they can be mixed with kinetic guns to reduce matter consumption, or to reduce the churn rates of other weapons.

Their main drawback is the prohibitive heat generation. The most common heat sink for a combat build is the advanced heat sink, with a dissipation of 19 per turn. With four weapons, that's a dissipation of 66 per volley. Some cannons generate more heat than that. Disposable heat sinks make that more viable, but realize that with 200 heat you get a penalty of 6% to both hit chance and dodge chance. Another drawback is the high power requirements.

EM weapons

These guns are very effective against everything but special robots and programmers. Their damage tends to be lower than other guns, but they make up for it with their corruption capabilities. Since a robot gets terminally corrupted at 100 corruption, late-game EM weapons such as HERF cannons are especially deadly. Another benefit of corruption is that it prevents the self-destruction of parts in the caves. This is based on the system corruption of the robot before it dies, so EM weapons should be placed first in a volley when used for that purpose. Furthermore, EM cannons don't punch through walls, so big EM guns don't accidentally compromise a defensive position except through occasional explosions.

For all their strengths, EM weapons also have a lot of drawbacks. EM weapons are most useful when stacked since it's the fastest way to corrupt robots. However, reconverting a full volley to/from EM wastes a lot of time and matter. Furthermore, firing a full volley of EM weapons puts tremendous pressure on heat dissipation and power generation, so the build has to be specialized for that purpose.

EM damage also fries power sources, heat sinks and processors, and causes explosions. Both these effects work against Cogmind. While the explosion can damage other robots, it often damages Cogmind itself and nearby walls.

Since EM weapons work best stacked, it helps to buffer EM guns until there are enough spares to make it worthwhile to switch to EM. Weapons of other types can be collected in the mean time. Ebb and flow.

EM works especially well against hunters, since they're quite susceptible to corruption but tough to kill in other ways. However, using EM against hunters typically fries their advanced targeting computers, and hunters are the primary source of those. So it's a dilemma whether to use EM against hunters. EM can also be used to take out an ARC before it deploys, with some luck.

Launchers

EM/EX launchers are both very effective. They're simply the best way to clear a group of 3 or more robots, e.g. on a deploying ARC. Programmers and sentries resist EX, but not enough to prevent explosives from being useful against them. Beware though that some special robots are very resistant to EX.

Launchers are precious because enemies don't drop them. Hoard them when you can! Furthermore, they don't leave much salvage behind and consume lots of matter, so carrying a spare matter storage unit is essential. Finally launchers tend to attract suicidal engineers who rush in the middle of combat like flies in a fire. This is bad because they often call for reinforcements before dying.

Disruption guns

Toys. Avoid.

Guns vs cannons

Cannons increase the raw damage output but typically suffer from poor resource efficiency, high recoil and low salvage potential. Non-EM cannons also punch holes in wall, which is a major disadvantage when it's not done on purpose. Cogmind kills best when the enemies are choked in a tunnel or in the entrance of a room. A blasted wall lets the whole batch in. This effect is lessened when Cogmind kills very efficiently, so cannons are more of an advantage for an already-strong Cogmind looking to become even more deadly.

Empirically, (improved) KE penetrators seem to work better than cannons in most situations. This isn't helped by the special energy cannons that are apparently universally terrible save for cold nova cannons.

Discussion

Equipping another weapon or weapon cycler increases the resource requirements over time by the percentage listed in the table. Conversely, the combat utilities increase Cogmind's resource efficiency, save for the power penalty. In other words, when Cogmind equips a targeting computer it will kill faster, thus saving matter while keeping heat generation constant. For that reason, targeting computers are more valuable than their damage percentage increase would suggest.

Weapon cyclers are extremely good. Their benefit extends beyond the firepower increase. They reduce the volley time so you can sometimes double-shoot before the enemy has time to fire again. They also reduces overkill for the same firepower.

Target analyzers are helpful against mundane robots, particularly when combined with flak guns and targeting computers. However a crit-heavy build doesn't help when enemies are immune to critical hits, which is when a high firepower is most needed.

An advanced armor analyzer provides a solid 90% chance to bypass armor. This is good when fighting sentries and big robots, but it doesn't help with mundane robots.

The optimal number of weapon slots is probably in the range [3-4]. It's a complex situation. The pros and cons of investing in more weapon slots are as follow.

Pros:

* This is the most reliable way to increase firepower. The availability of the combat utilities is quite random, but you can always find plenty of guns around. This is especially useful early on and after being stripped.
* Weapons have the best coverage after armor and decent integrity.
* For the first half of the game, it's the most effective way to increase damage and to increase coverage. The early utilities are weak, don't contribute to coverage, and there are very few slots, so investing in weapon slots is doubly effective.
Cons:

* Weapons are heavy.
* Additional weapons increase recoil, volley time, and waste due to overkill.
* Weapon churn can become a problem.
* The more weapons you need, the less picky you can be. Lesser guns don't increase your firepower as much.
Personally I'd go with 4 kinetic guns, 2 targeting computers, 2 weapon cyclers, and some heatsinks for a late-game build. This is enough firepower to slice through anything, and it works reliably and universally (swap weapons for hunters as needed). While it's possible to increase firepower further, it's probably best to focus on defensive parts (armor + propulsion) and storage beyond that point. Other combat utilities can be used to plug holes, but given the choice, it seems to me that utilities that are universally useful are better than those which aren't.


Final tip

Shit happens. Therefore, carry one or two flight units for emergencies. Speed 35 and low coverage are the characteristics you're looking for. When you're about to lose, equip any spare propulsion part you have, equip the flight units, dump everything that slows you down in increasing coverage order, and get the hell out of there. With the flight bonus, high speed and propulsion armor, you are rather resilient.

Enjoy the game!

gammafunk:
I haven't read it...but I intend to..and this will finally be the key for me breaking into and then dominating the late game. Thanks, you're a good green name!

zxc:
This was a fantastic read. Great work!

Kyzrati:
Very well put together, Decker. Excellent points all around.


--- Quote from: Decker on November 04, 2016, 04:10:50 PM ---Per Kyzrati's request, I avoid discussing the specifics of the less useful utilities.

--- End quote ---
Clarification for everyone else: Beginning with Alpha 12 a large number of utilities will be adjusted, with the most significant changes aimed at those which in the past have been clearly suboptimal or even outright useless (many of these have simply lived on since the 7DRL). It will be a project spanning multiple releases (so not everything will be updated right away), and I'll be calling for opinions on suboptimal utilities later, but first I have a lot of ideas to get in there myself.

Edit: I also like how this guide is spoiler free :D

PleasingFungus:
What's your opinion on different propulsion types for combat purposes? Specifically treads vs legs, I suppose.

Navigation

[0] Message Index

[#] Next page