AI War state of the game

[Wingflier]

This post is made to outline what I consider the major underlying problems with the current state of AI War.

AI War is a game that promises a unique experience with every game, and every newly seeded map.  It does this by randomly generating many elements of the experience such as buildings, ships, enemies, tactics, fabricators, ARS unlocks, factories, and many other aspects.  The idea is that the player has to face a completely new and dynamic challenge every game, and react accordingly:  That is the premise that AI War is founded on.

But does it deliver on that promise?

The more I've played recently, the more I've realized that even though each scenario and situation can differ wildly, they must all be handled in a similar way.  The ships you unlock, the technology you choose, the tactics that you use - I've realized that in every game, these stay almost exactly the same with little variation.

Where before there was strategy and critical thinking before each battle, it has now devolved into an extremely repetitious process of taking every planet in the exact same way, with little resistance, and little chance of failure.

I'm talking about blobbing.  I'm talking about sending my entire army into a planet, alerting that entire planet, then killing them all 1 by 1 as they slowly attack my inferior blob in pieces, and ultimately lose a war of attrition as a result.

In the ideal game state, alerting an entire planet should be a huge no-no unless you have a massively superior force.  To bring an entire blob onto a planet with superior strength and technology (in the form of Guardians and such), alerting them with your sheer size, should bring swift death upon you.

This is a strategy game isn't it?  When did it stop becoming a strategy game?  At what point did mindlessly moving all of your units into a vastly superior enemy army become a "good idea"?  It should not be.  That is not good design, and the game suffers for it.

The entire idea of the game is to out-think and outplay your vastly superior AI opponent is it not?  If you could win purely by blunt force and direct combat, then the humans never would have lost the war to begin with.  So why does the game reward you for brute forcing your opponent then?  I can't think of a good answer to this question.

Many people object to the idea that the game promotes blobbing and brute-force tactics.  They may say "What about the counters?! What about AI Eyes!"

The first thing I want to say is that the fact that AI Eyes even need to EXIST is proof that there is something incredibly broken with the game.  If you can send your entire force into ANY moderately-defended planet and not get crushed, there was already something wrong with the game's design to begin with.  Let's call AI Eyes for what they truly are:  A band-aid to a bigger problem.

And what is this problem I speak of?  Lethality.

AI War's fundamental problem comes down to a problem of lethality.

Did you know that it takes 28 seconds for a Fighter to kill a single Bomber?  TWENTY EIGHT SECONDS.  That is enough time to get a cup of coffee and come back.  Why?  Why does it take so long?  And the saddest part of it all is that is the fastest matchup of the entire Triangle Counters, and indeed one of the fastest counter-matchups in the game.

But you may be wondering, "Why is lethality so important?"  Why is it so critical that battles end faster than they are now - some of us like the game the way it is.  It's important because at the current state of lethality, AI War has basically boiled down to a war of attrition.  Ask yourself:  When you take on much heavier planets with forces many times your size, how is it that you can continually not only survive, but even easily win the fight?  It's because the enemy, sending their forces at you piece by piece, never actually deals enough damage to punish you for alerting the entire horde at once.  In addition to that, you may be constantly healing yourself during the battle with Engineers, Mobile Repair Stations, Medic Frigates, etc., further increasing the longevity of your (supposedly) massively inferior force.  So you may ask, "How does increasing the game's lethality discourage these brute-force tactics?"

It's simple:  When the lethality is cranked up, you can still blob to an enemy planet and wait for their entire force to come upon you, but the difference is that each unit will be dealing more damage before it dies.  A Beam Guardian in its current state may damage to a few units before you kill it, but when you crank the lethality, making it do 5x as much damage per shot - you'll lose a large portion of your force before it finally dies.  A Lightning Guardian now may hurt your entire army by a laughable amount - but when it starts taking 10% health off of your entire force with each attack, you're going to regret alerting several of those at once.  The same goes for Starships and even the smaller fleetships as well.  It doesn't matter that they also die faster, what matters is that they inflict more damage before they go out.  In other words, increasing the lethality removes your ability to win a battle of attrition against a superior opponent.  Even though he is still sending his force at you in pieces, each "piece" does its fair share of damage before it dies.  Eventually you become overwhelmed by the sheer force of the opponent, where before you could basically take on an endless army as long as you were fighting a battle of attrition.

In my opinion, the lack of lethality of this game is the main reason why people can continue beating it on 10/10, even after we buff it time, and time, and time again.  It's not because these people have superior strategies, it's because these people have perfected the same strategy you use for all difficulties, and in fact, for most scenarios.

Increasing the lethality of the game would require the player to play much more carefully and tactically than before.  No longer would it be a good idea to send your entire force to an even moderately defended planet, as alerting them all at once could very well cause you to lose the game.  Instead, the player might want to spread out his full force across several planets, making surgical strikes and dealing moderate damage over time, instead of attempting to overwhelm his opponent with brute force.

Upping the lethality of the game would obviously cause some major balance concerns, which is why the hull-types and armor system would need to be restructured (in my opinion) before any major change like this took place.  However, I think increasing the lethality of the game is the future for AI War, and an attempt to go back to the fundamental principles around which the game was designed:  To handle each scenario differently, and to continually change your strategies based on the situation.

[keith.lamothe]

Not really going to jump in at this point, but two thoughts:

This is a strategy game isn't it?

Yes, but the blobbing you're describing is a tactical issue, not a strategic one.  AIW would be a strategy game even if planetary attacks were resolved automatically/abstractly (from as simple as risk, to as complex as Dominions 3, etc).

AI War's fundamental problem comes down to a problem of lethality.

I've heard this from someone else recently, but the objection in that case was that the lethality is too high

[Lancefighter]

I would be interested in seeing a comparison of mk1 ships during 2.0 and now, to see what has changed. Outside of the Fighter beats bomber 100%.

[keith.lamothe]

2.0?  Did the fighter even counter the bomber then?  At some point the triangle was bomber -> fighter -> cruiser -> bomber.

[Lancefighter]

oh  right, that was 3ish, wasnt it.

In either case, the comparison would still be interesting >_>

[Faulty Logic]

But does it deliver on that promise?

Yes.

The more I've played recently, the more I've realized that even though each scenario and situation can differ wildly, they must all be handled in a similar way.  The ships you unlock, the technology you choose, the tactics that you use - I've realized that in every game, these stay almost exactly the same with little variation.

Not in my experience. My fleets have been:

Speed-based: using exp speed boosters, tractors and engine damage to chop the enemy into bite-size pieces.

Carrier based: a core of SBS, TDLs, and Neinzul enclaves, with other ships there to keep the enemy off the carriers.

Long-range bombardment: Z bombards and missile frigates, fighters to intercept enemy bombers.

Brute force: throw everything under an umbrella of shield bearers.

Starship based: don't unlock any fleetships, use the cheap mkIs as a cheap screen for the starship core.

And that is all base-game. Golems and spirecraft each add a huge array of fleet compositions/tactics. (FS is allowed to win by brute force, of course).

The champion and its minions form a fast, heavy-hitting strikeforce to augment your base fleet. The champion abilities allow for more tactics.

Where before there was strategy and critical thinking before each battle, it has now devolved into an extremely repetitious process of taking every planet in the exact same way, with little resistance, and little chance of failure.

Taking a planet in itself shouldn't be that big a deal. Dealing with the consequences should be. And yes, there should be some challenging planets in there, and there are (the first couple you take, eyes, inter-p boosters, raid engines, fortresses, planets with tons of snipers etc).

I'm talking about blobbing.

Here we go again: strategic blobbing, or bringing your entire mobile fleet in to conquer a planet, should confer advantages. If you have sufficient defences without your mobile fleet, then you have earned the right to blob.

Tactical blobbing is more of a problem, and here I agree with you. But there are counters, enough so that you cannot do it "mindlessly" and expect good results. Take a self-destruct guardian, for example.

This is a strategy game isn't it?  When did it stop becoming a strategy game?

Yes. Never. The strategy comes from which planets you take, when you take them, and how long you hold them. And that is different every game. A "tactical" problem for me is, say, a CPA attacking in more than one place: how will I deploy my fleet? Or: exo attacking the FactIV: do I try to hold it, and risk losing outright, or do I turtle up at my homeworld (or bust out the warheads)? Planetary problems are sub-tactical issues.

At what point did mindlessly moving all of your units into a vastly superior enemy army become a "good idea"?

First, mindless will lose. Second, even a mkIV planet rarely has anything approaching "vastly superior".

The entire idea of the game is to out-think and outplay your vastly superior AI opponent is it not?  If you could win purely by blunt force and direct combat, then the humans never would have lost the war to begin with.  So why does the game reward you for brute forcing your opponent then?  I can't think of a good answer to this question.

Just because we can win planetary battles does not mean we are winning by brute force. Brute force would be taking far more planets than necessary. Do that, and the game will hammer you as advertized.

That is not good design, and the game suffers for it.

Lancaster's law is a fundamental fact of any wargame, not bad design.

AI War's fundamental problem comes down to a problem of lethality.

I strongly disagree here. Fleet-to-fleet combat lasts minutes at most. This seems about right. And having higher lethality would mean those sub-fleets on the AI planets would die against your blob even faster, without the small chance to support each other they have now. (also, doesn't Blitz/normal/epic style do this, if you want to change the lethality?)

Increasing lethality would, I think, increase blobbing. Tactical strike teams would be cut down before they could do anything useful.

Finally, I do support making the planetary-level stuff more interesting. I just don't think there are any fundamental problems with the game.

Dynamic Defence Guardians
AI Defence Fleets

[Wingflier]

Yes. Never. The strategy comes from which planets you take,when you take them, and how long you hold them. A "tactical" problem for me is, say, a CPA attacking in more than one place: how will I deploy my fleet? Or: exo attacking the FactIV: do I try to hold it, and risk losing outright, or do I turtle up at my homeworld(or bustoutthe warheads)? Planetary problems are sub-tactical issues.

I'm on my phone right now so I can't reply to all your objections, but I think this paragraph is the crux of our disagreement. 

I agree, CPAS, Exos, and sometimes waves are the REAL challenges of the game, and even then that's only because the AI is hitting you with overwhelming force from multiple angles, not because the game is asking you to change your strategy or be particularly creative.

Even then, these major events only happen one every hour or so at most. In my opinion, playing a 14 hour game in which the challenge and threat is not ever-present, but only really becomes dangerous once every hour or so is not good design. I think that as the player, every choice you make should affect whether you lose in the next few minutes.

I can not, in good conscience ask my friends to play a game with me that lasts upwards of 12 hours, and in which we are only really challenged a handful of times in this huge duration. This is not good pacing.

There are all kinds of games we can play where every moment and every decision can end in disaster. AI War's current pacing is just off, and in my opinion it's because planet battles are no longer dangerous or challenging.

[KDR_11k]

The reason blobbing doesn't work well in real life* is that frontlines are wide and concentrating all your forces on one point leaves the rest of your area exposed. The enemy general will exploit that and encircle your forces, denying them the supplies they need in order to keep fighting. The AI is passive. Only its free ships are allowed to engage in counterattacks, as long as not enough of those are around the AI cannot punish you for moving all your mobile forces out into enemy territory. So the AI's true force strength is just the strength of the threatfleet (and stuff like hybrids) which is usually far below your own strength. It can defend against your forces with the special forces but those cannot counterattack. And even then your territory can be fortified like the goddamn Marginot line and you can be sure that Belgium is not a valid path. The AI's harder attack options, waves and CPAs, come with announcements so you know perfectly well when you can attack and when you should stay home and help defend.

Players do blob in competitive RTSes but until the game is decided they each have a blob of about equal strength and just rushing past the enemy blob would get them annihilated, he who first loses his blob is exposed so the tactics are down to inflicting more damage on the enemy blob than it does to yours.

Mindless blobbing and attack moving is common against the passive AIs found in most RTS campaigns (they don't expand, they just sit there and take it).

Higher lethality would not change the blobbing. The current piecemeal forces can already kill one or two of your units if they get in range but most get gunned down by your frigates and other long ranged ships before they actually move into range. That beam guardian won't do more damage to your troops with higher lethality because it will be shot to pieces much faster. Lethality amplifies the effects of speed and range, that's pretty much it.

* Assuming strategic blobbing here, i.e. concentrating your whole army on one location, not tactical blobbing, i.e. having everybody move in a tight formation that gets annihilated by explosives and machineguns. AI War does not provide the necessary tools to cope with AOE attacks so tactical blobbing is okay in it too. Making AOE common with the current game design would be a disaster.

[doctorfrog]

Getting a real sense of deja vu with this thread. But taking a genuinely holistic look at the game by the community could still bear some good fruit.

Also there's a joke somewhere in using gigantic walls of text to discuss blobbing.

[TechSY730]

I don't know about you, but the new threat fleet and special forces mechanics have made attacking planets a much trickier and interesting for me, or at least for the planets that the AI consider worth enough to use these to defend with (which tend to be planets I want, especially in the mid game)

[HitmanN]

First, let me start off by saying that I haven't played a proper game of AI War since something like 4.0. Started LotS couple of times, momentarily tried AS, and so on. Always had too much other stuff to focus on to get back to the game proper. So my in-depth knowledge of current balance is pretty limited.

AI War's fundamental problem comes down to a problem of lethality.

I strongly disagree here. Fleet-to-fleet combat lasts minutes at most. This seems about right. And having higher lethality would mean those sub-fleets on the AI planets would die against your blob even faster, without the small chance to support each other they have now. (also, doesn't Blitz/normal/epic style do this, if you want to change the lethality?)

Increasing lethality would, I think, increase blobbing. Tactical strike teams would be cut down before they could do anything useful.

Finally, I do support making the planetary-level stuff more interesting. I just don't think there are any fundamental problems with the game.

This is exactly what I feel too. Lethality is pretty ok. For me it is as long as I play on Epic. Anything faster is just out of my hands on tactical level. For me to use anything else other than blobbing, I need time to perform those strategies, maneuver the ships, observe the results and change the tactic while the results are happening. The less time I have, the less incentive there is to try anything creative. Blobbing isn't ideal, but it works decently, so it's easy to rely on that in such situations.

And yes, planetary-level stuff could be more interesting. My biggest issue has been that despite the AI having some variety in its ship usage, it seems to be a jack-of-all-trades on most planets.

For instance, if AI has ships A,B,C,D,E in use, on an AI planet with 15 ships, I'd tend to see combinations like:

AAA
BBB
CCC
DDD
EEE

or

AA
BBB
CCCC
DDD
EEE

...and so on. The AI always has a bit of everything. This does mean that it makes sure it has no weakness. Which in turn becomes its weakness. The AI seems to be powerful only in total numbers, and rarely in having upper hand in specific scenarios and tactics.

What I would like to see on the AI side is more variance in the ship distribution. Combinations like

AAAAAAAA
B
CCCC
D
E

or

AAAA
CCCCC
EEEEEEE

Because player is limited to certain number of each ship type, the AI could occasionally have a planet that simply overpowers the player's favorite strategy by sheer number of a specific one or two ship types that counter it, forcing the player to either use a completely different strategy, or try to use less-ideal ship types together with the primary tactic, to improve the odds of success.

What if, for instance, one AI planet had been seeded a preference of anti-blobbing. It would prefer primarily ships that deal AOE damage, or generally hurt multiple ships at a time. Or what if one planet had been seeded an artillery preference, and it has a higher distribution of long-range weaponry, compromising it's mobility in turn (not completely, but partially). Or Anti-starship preference. Or... well, you get the idea. These would not need to be defined on every planet, but it would be nice to see different planets having different sets of preferences, that are considerably more noticable than the current distribution variance.

Anyways, just thought I'd share the idea. These preferences could change according to what the player does, but they'd be a sort of starting point for the AI. Strong in something, instead of average in everything.

Another benefit of this would be that the AI's ship selection is more limited on these occasions, and picking out strategies against each ship type is much easier than in scenarios where the AI has tens of different ship types. Seriously, I'm not going to bother tactics against such a mess. Blobbing should suffice.

Again, I don't know if things are different in end-game, because I haven't played a game past taking a few planets since late 4.0, but every time I feel like I'm going against jack-of-all-trade planets, and it's been very... uneventful.

[keith.lamothe]

@HitmanN, from the 5.074 patch notes, August 30th of this year:

=== AI Reinforcement Focus ===

* AI planets now pick a primary, secondary, and tertiary type of ship to focus their reinforcements on. 
** It's not purely random, sometimes it picks ships that fit certain roles.
** When picking a new reinforcement ship for the central pulse or a non-special-forces guard-post-pulse, it has a 10% chance to pick any available ship like usual, and a 90% chance to pick one of its focus types.  When picking a focus type, it has 3 times as much chance to pick the primary as the secondary, and 3 times as much chance to pick the secondary as the tertiary.  If that doesn't make sense, don't worry about it.
** Whenever the AIs unlock new ship types due to AI progress, all AI planets "reroll" their focus types.

To the rest of the thread:

FWIW, I sympathize with the desire for tactical blobbing to be less viable, by way of making the alternatives more necessary, in turn making the tactical game more interesting.  And thus, making the game more interesting.

I really sympathize with it.  If I had my way, I'd be trying something like reducing all ship speeds to 1/8th of what they are, all ship ranges to 1/4 what they are, beefing up the danger of attacking a defended position, AOE, firing angles, etc.  All kinds of stuff.  You wouldn't tactical-blob in a universe like that unless you had a death wish.

I'd have a blast.  And not just as a developer, though I expect I'd get a lot of players-dying-everywhere amusement too.

But that's not this game.

As KDR_11k pointed out, the game doesn't give you the tools to efficiently deal well with a lot of incoming AOE.  Nor does it provide the kind of interface support to deal efficiently with firing angles, subsystems-targeting, and all kinds of other tactical dimensions that are wonderful in more tactically-oriented games.

The reason it doesn't do these things is not a design flaw, but the definition of the design: this is a strategy game, not a tactical game.  Its tactical gameplay is a lot deeper than most strategy games of this scale, but it's still second fiddle.

To the extent that the tactical part can be made more interesting without compromising the strategic gameplay, that's fine and I'm all for it.  But past a certain point taking a planet becomes so complex that the time and mental energy that goes into it renders a game that involves taking 15-20 planets beyond the endurance of a large portion of our player base.  Not to mention requiring a great deal of refactoring of the mechanics and interface so that it isn't a constantly-banging-your-head-on-the-desk situation because your fighters aren't getting out of the way of the flak bursts, etc.

Anyway, got a galaxy filter that needs coding...

[doctorfrog]

The reason it doesn't do these things is not a design flaw, but the definition of the design: this is a strategy game, not a tactical game.  Its tactical gameplay is a lot deeper than most strategy games of this scale, but it's still second fiddle.

To the extent that the tactical part can be made more interesting without compromising the strategic gameplay, that's fine and I'm all for it.  But past a certain point taking a planet becomes so complex that the time and mental energy that goes into it renders a game that involves taking 15-20 planets beyond the endurance of a large portion of our player base.  Not to mention requiring a great deal of refactoring of the mechanics and interface so that it isn't a constantly-banging-your-head-on-the-desk situation because your fighters aren't getting out of the way of the flak bursts, etc.

Each time this thread gets recreated, I want to say this, but I'm basically the Donnie of the conversation ("you're out of your element, Donnie!") and don't have the words.

But definitely, this. This, this, this.

The kind of game that I see described by Wingflier's Hunger (this one is going into altplanetnames.txt) is also a game that I'm interested in, and I hope that Arcen will make it someday, but it isn't this one.

AI War 2: Tactics. Or AI War: Gratuitous Space Battles.

On the other hand, there was room for a "defend against the rush" mode in vanilla AI War, maybe there's room for a "tactical chess battles of humanity's fleet" mode as well.

[TechSY730]

Relevant link to this discussion:

Why Does The Human Player Always Have The "Tempo?"

This article covers some of the points being discussed here.

[Wingflier]

I agree with everything Keith just said, including his personal suggestion on how to change the game.

I also agree with him that we shouldn't change the game in such a way that will alienate most of the players.

However, where I disagree with him is that keeping things the way they are on the tactical level.  I think that is going to keep this game extremely niche.  Even I get bored quickly of playing for hours at a time, and only feeling a sense of real danger and challenge once or twice in this duration.

I just don't think that level of pacing is acceptable, even in a grand-scale strategy game.

That's just my personal opinion though, you're welcome to disagree.

This isn't even technically an "anti-blobbing" thread, this is an "anti-sending everything you have to a well-defended planet and expecting good results" thread.  Planet battles should not be like stealing candy from a baby, yet even with all the crazy SFG changes and such lately, it's still not that difficult.  At worst you retreat, wait for them to come to your defenses and kill them anyway.  The planet is clear and you move on.

Increasing lethality would, I think, increase blobbing. Tactical strike teams would be cut down before they could do anything useful.

I think you're looking at it the wrong way.

Increasing Lethality does not give you the advantage, it gives the AI the advantage.  The AI has the superior force.  The reason this doesn't matter as much now is that it takes so long to whittle you down that you can just keep repairing and outlasting them in prolonged planet battles.  If however, every Guardian and Starship they sent took out huge swaths of units before finally going down, you'd really think twice about alerting the entire planet at once.

Think of it like the Japanese Kamikaze planes.  Those weren't any harder to kill than any regular plane, in fact probably easier since they were regularly coming in range of ship anti-aircraft weaponry.  However, their high lethality made them a bigger threat than regular bombing runs on ships, even at the cost of their own lives.

It's the same concept here.  The player's advantage in the current game is that he can easily win a war of attrition during a planet battle.  If each big unit he kills takes many more down with it, you would have to play MUCH different (i.e., not so mindlessly).