Why an artificial combat triangle?

[Kohan]

The Rock-Paper-Scissors scheme currently employed appears to be almost totally artificial.  I have nothing against the (supposedly) unorthodox direction of this triangle, but it bothers me some that the combat triangle is essentially "Well, um, missiles do thirty times as much damage against bombers.  Because... we say so."  (Of course, this is just one example.)

I am very fond of combat triangles that occur naturally, based upon the way that gameplay elements actually interact.  But if you were to take a game, say, Team Fortress 2, and you find that the sniper class is not weak enough against the spy, wouldn't it seem a bit of a poor solution to just double all of the spy's damage against the sniper?

I'm not trying to outright say you're wrong, but I do have my stance.  I would like to know the reason behind the implementation of this artificial combat triangle.

Thanks for reading.

[darke]

http://arcengames.com/forums/index.php/topic,1956.0.html

[x4000]

It's shortcut simulating complex internal factors of the construction of the various ship types.

If you take real-world examples: a Chevy Malibu and a Porshe 911 and a Hummer, let's say. If you were to fire bullets of varying sorts (differing calibers, even amongst "shell" type ammo) into those various vehicles, you would get different damage characteristics to all of them, resembling more what you see in AI War based on a mixture of natural RPS interactions and the "artificial" bonuses underneath.  For example, placement of engines, size of passenger compartments, material and amount and position of armor, plus other internal safety features, all make for an extremely complex real-world situation.  And, for that matter, as changes are made to each line of car with each calendar year, you'd see different performance characteristics per car per year in order to make all of those realistically modeled.

So, what AI War is doing for the sake of sanity and CPU efficiency, is generally modeling the interactions of all of the various shot types from various craft against the various hull and construction characteristics of other craft.  It can be assumed that the caliber and other factors is a bit different even between "shell" or "laser" shot types that are otherwise the same, and so forth.  So that's what is being modeled -- it's very different from a "job class" for human avatars in a game like TF2, because those refer to skills and there is very little difference in shooting a big skillful guy or a small medic guy with a sniper rifle -- they both die pretty much the same.  Giant spacecraft and a variety of weaponry are a whole different animal.

[Kohan]

Ah, I see.  So then, is damage multiplication a result of ammo-versus-ship-type, or ship-type-versus-ship-type?  Or are those two the same thing?

[Nibelung44]

Sorry X4000, but no, you are defending your system and design, I can understand that, but don't it hints about 'something' which is not suboptimal, if regularly people complain that your system is very artificial?

Your cars example seems to work, but is not describing the artificial nature of the coefficients you set, which go well beyond things like type of ammo, type of protection and particular durability of ships. There are many cases that are artificially tweaked, and in the end, we have to learn by experience the ships interactions between each other, and the tooltip "weak against  / strong against" is just of little help. I just lost yesterday my 320 teleport raider because they jumped nearby a lightning turret which killed them in one shot. This was not predictable by reading the damage of the turret, the resistance of the raiders and even the tooltip said nothing about that. I painfully learned the experience...

--
I'll just repeat an example which clearly demonstrate the rock-paper-scissor (and artificial) nature of the system:

Cruisers (Missile Firing) are strong against Bombers. MLRS (Missing firing) are weak against Bombers. Same weapon class, same target, different effects. Does it makes some sense to you? If there is one it is too subtle to me, sorry. [and it seems it is too subtle for some others ]
--

Now you have 100+ ships design, so theoretically it can goes to a 100x100 interaction matrix that you set (and that only you a few know well enough to be at ease with, sorry to reveal that most people don't master the game as well as his designer and old veterans  ) and that players have to learn by experience.

If you are bored that people continually post such topic, perhaps there is something to do, don't you think so?

[x4000]

It's really a ship-type-versus-ship-type sort of thing.  Even if two ships have the same "shell" type of ammo, it's assumed that they are subtly different in construction of ammo (caliber, what have you).  In the case of lasers, different construction of the firing mechanism, power levels, this that and the other.  And for the hulls on the other end, I've pretty much covered that already I think.  So the net effect is that every ship is assumed to have a subtly different ammo type, and a subtly different hull type, and the simulation is how all of those things interact.

Of course, the general ship classes (autocannons, tanks, whatever) all share a huge number of characteristics amongst their levels, even if the balance at each level isn't quite the same (same as the relative performance characteristics of a 98 civic and a 2000 civic are different by a bit).

And, similarly, the different ammo types (shell, ammo, missile, etc) all have some similar characteristics, especially when it comes to specific immunities (immune to sniper shots, or minor electric shots, or whatever), or when it comes to absorption of shots (absorbing lasers on deflector drones), or when it comes to counter-shooters (counter-missile turrets, etc).

So, there are some overarching meta-classes to make things a bit simpler compared to what they otherwise would be, but each unit is subtly unique all the same.

[x4000]

Sorry X4000, but no, you are defending your system and design, I can understand that, but don't it hints about 'something' which is not suboptimal, if regularly people complain that your system is very artificial?

No, not really.  The fact that mostly new players complain a bit about something minor, while experienced players comment that it would break the game to change it, and reviewers love the game without ever mentioning this, does not make me think that this is suboptimal.  Rather, it makes me think that this is something very new that challenges existing assumptions but does so for very good reasons that have been extremely well thought out, and that some people will simply not like that because it is different.

If you are bored that people continually post such topic, perhaps there is something to do, don't you think so?

Yes, I'll refer them to these two posts, which have my arguments pretty clearly laid out.  I'm not going to break the game to satisfy a very small minority of players, I'm sorry.  I'm very open to feedback, but I've put quite a lot of thought into this over quite a lot of time, and none of the arguments about complexity or what have you are convincing in the least.  Yes, this is a complex game.  I should make it overly simplistic to be like every other RTS?  There are many other games with those other characteristics.  The interactions between units is well known and predictable there, between games even, which is what I think you are advocating basically.  AI War does some things that are very new, and that are generally well loved by its audience.  I have to accept the fact that I'll never please everyone, that's simply a fact of life with any product.  But breaking the game for the majority to please a small minority is not a good idea in any form, I'm sorry.

[darke]

If you are bored that people continually post such topic, perhaps there is something to do, don't you think so?

I think the simplest thing to do is to add another base ship class with subtle variations as to what's better then what, so people will stop making silly comparisons to the tired old rock-paper-scissors cliché.

[Kjara]

I'll still argue that one thing that would cut down on the complaints about the arbintaryness would be to have the good/poor against lists be dynamically generated rather than static, and be a bit more complete, so its easier to find useful info out of it(also there are still unit classe that don't have useful info).  Ie, laser mkIII turrets have no info, and laser mkII's only list three units that they beat, and none that they lose to.

[Nibelung44]

Well, I understand that you don't want to break the actual system, if most of the players can cope with it. Me too by the way, I learn by experience the matrix that you set, and as it sinks gradually into my brain, I believe it is becoming more natural now (even if I still believe it is artificial...). So now that I know to never approach Lightning turrets with Teleport Raiders, or that fighters destroy bombers but bombers destroy fighters it is more ok than before...

I'm a developer myself (in DB though), so I thought of something... I believe you are a big proponent of encapsulation and indirection in your code. So in the end, when you query the damage coeff of ship A vs ship B, it boils down to going to a single function I guess.

So why not having at least as a debug option that some people can enable something like that (a code similar to this one, I could do it in a few minutes in the apps I build): if the option is checked, instead of reading the coefficient matrix (of damages coeff of A vs B) why not just just polling the weapon type and armor type and make a quick and dirty code (temporary if you like, this should go to an ini file or something like that for further modding):

(pseudo code)

if ammo = shell and shield then Coeff = 0.5 else
if ammo = shell and no shield then Coeff = 2 else
if ammo = laser and shield then Coeff = 2 else
if ammo = laser and no shield then Coeff = 0.5 else
if ammo = missile and shield then Coeff = 0.75 else
if ammo = missile and no shield then Coeff = 1.5 else
if ammo = energy bomb and shield then Coeff = 4 else
if ammo = energy bomb and no shield then Coeff = 1 else
Coeff = 1

the special cases cover 95% of the ships, the rest stay at 1. (code done in 5 mn chrono I believe...)

If you want to commit 2 more minutes, some special cases should be done for energy bomb against structures, and that's it. In a few minutes, you can propose a debug option for people to try out, which is vastly simplified but much more logical. Why not allow people this kind of experience if they want it, if this cost you a few mns and a checkbox in the game interface?

[quickstix]

Here's my two cents.

Why I like AI War so much is that I know what units are good for, and what they're not good for. The basic structure works so well because it's well, basic. It's well thought out and makes sense in a kind of elegant way. In other RTS games, I sometimes find myself struggling to pick between units because the differences between them are so minor that they really don't matter when you group them together and rush them. I think the minute we start to make the differences between units less distinct, the minute differences between them matter less, the game will become boring because all you will need to do is rush one or two types of units. What makes AI War great is the fact that you really need to think about what types of ships you will need build and unlock, often a game making decison.

This is just my opinion, but I felt I needed to voice it because this is an issue new players bring up every so often (as well as the whole semantic issue of 'what a fighter should do') but it's such a core gameplay concept that changing it would change the whole game, and I honestly think AI War works so well because its been designed very carefully with these concepts in mind, which is why many people, including myself, enjoy it so much.

[Revenantus]

if ammo = shell and shield then Coeff = 0.5 else
if ammo = shell and no shield then Coeff = 2 else
if ammo = laser and shield then Coeff = 2 else
if ammo = laser and no shield then Coeff = 0.5 else
if ammo = missile and shield then Coeff = 0.75 else
if ammo = missile and no shield then Coeff = 1.5 else
if ammo = energy bomb and shield then Coeff = 4 else
if ammo = energy bomb and no shield then Coeff = 1 else
Coeff = 1

If you tie damage coefficients directly to weapon types, then any changes made to those coefficients will echo throughout the entire combat system. It will become impossible to do fine tune balancing between two unit types. Let's assume you make a number of changes to the coefficients and unit health, trying to find the ideal balance without screwing up anywhere else. Say you find this elusive configuration that matches exactly the ideal balance you had in your head from the beginning; where was the gain over setting that balance manually? To be frank, the whole thing strikes me as an elaborate way of losing development time.

It is also completely unrealistic to expect this to quick to implement, because it would be a miracle if you could find such a simple generalised system that in any way approximates what we have currently. It would require huge amounts of analysis and testing to reach the point where the game was playable. Not to mention that such systems are incredibly high maintenance, especially with the number of units in AI War where analysing every ship interaction every time a balance change is made would be unfeasible.

Again I'll reiterate that my views here don't make me averse to making superficial changes to the game. For example, if we were to give the MLRS unit a different name and weapon type, hence changing its identity, to something that might seem more suited to its role, that may well be pretty cool. This works for me because it doesn't involve actually changing the balancing itself, or unnecessarily tying damage coefficients into ammo types.

[Vade]

I think the current system is great, the RPS or triangle system what ever you want to call it promotes balance and is one of the most balanced systems there is. Some may say it is unrealistic or artificial but neither of those factors are that important when it comes to a game since obviously a game's goal is to be first and foremost fun. This game is very balanced, very fun and also very challenging at high levels and that is why I like it so much and you can tell a lot of thought went into the system and how to balance the units.

[Ktoff]

The current system works really nice, but it is not very transparent.

An example are fighters and cruisers. In my experience fighters die off much quicker than cruisers, but this is not reflected in the unit stats. (i thin mk1 fighters even have more health than their counterparts)

Dreadnaughts are described as good against forcefields, but a raid starship takes a forcefield down much quicker and wipes out the rest of the defenses in the process. (also Dreadnaught: attack:~1000/s; raid starship ~28000/s)

You can learn what work well and where, but you cannot find it out before trying...

Edit:
Or the Spire versis Zinth Starship (not tried any of them yet) Both have same amount of shields speed and range but the spire is more expensive (both in terms of knowledge and resources) takes longer to build, has less health and does less damage:
19800/shot*6 shots/2.7s~43500/s for the spire starhip and
9800/shot*14shots/2.3s~60000/s for the zinth starship

Just looking at these values and neglecting the fact that unlocking the spire lets you field 2 more powerful starships the spire look inferior to the zinth starship on paper at least. I might yet learn the advantages of one starhip vs. the other but it won't be clear (not a bit) before i try.

[laxrulz777]

A) I don't mind the current system (pre Zenith)
B) Let's acknowledge that the system is complex (and somewhat opaque)
C) Let's also acknowledge that there are some things you can ONLY learn from experience (i.e. T-Raiders die a bloody horrible death to turrets and can't damage turrets worth a flip, T-Stations take out turrets like a knife through butter but get absolutely shredded by special ops stations, etc)
D) Can we also acknowledge that the (already somewhat complex, somewhat opaque) system is getting MORE complex with the expansion
E) Chris has said multiple times that nuanced unit balance isn't important because of the assymetric nature of the system. He's acknowledged that there units that are appreciably weaker then others and that that is part of the skill testing of the game.
F) I also fail to see how a changed system would "break" the game. It would certainly change the relationship of the units but that doesn't have to mean "breaking". And if it's a system that lets me properly guess that I should teleport my 50 strong strike force of Teleport Stations (that have been decimating dense, mixed turret nodes) near a lonely special ops base because that lonely special ops base will CRUSH them (making me cry a little on the inside) then all the better.
G) The tooltips only help so much. They have limited space showing and often don't indicate the thing that I want to know about. What's better for taking out turrets? Cruisers, Fighters, Bombers or Laser Gatlings? I guess I'll just have to experiment? What turret should I build for a wave of incoming Fighters? It's not MLRS which is what you would have intuitively thought. Turns out it's lasers (I think... hard to remember now).

I'm not suggesting a change but I think an out of hand rejection (which is more and more what this appears to be) is, perhaps, detrimental.

Lastly, and this is probably a question for Rev (since I believe he built the charts), the simulations assume that the units head towards each other from some distance, yes? Are the numbers significantly different if I warp to a wormhole on top of someone (or have a force waiting for them on a wormhole they're coming through)?