Strategic Layer Improvements

[wghost81]

And about randomness on strategic layer. Believe it or not, it exists. :smile: April Mutons do not have 100% probability and although it's quite probable you'll see at least one pack in April (I had no Mutons in April at all once), it is also probable you won't see any Mutons in the first mission. And fighting Mutons early April and late April are two different things. :smile:

 

By introducing randomness in mission/alien type, one can unbalance the game very quickly, as loosing early hard mission will mean less resources and more chances to loose another mission.

 

I'm actually thinking about another thing: tie in alien types not to current month, but to UFO type. It can still be random, but with weaker aliens on small UFOs and tougher aliens on big ones. It may not seem to make much difference, but by making harder missions even hard to compensate for late game player progress and still allowing easier missions for rookie training, it can help balance late game difficulty. Another thing is to add UFO type report to abduction mission info (replace difficulty info with UFO type info, for example). It could add another choice to strategic game: reward vs panic vs difficulty. Sure, there is difficulty choice now, but it is not much of a choice really: you want more aliens for more XP and resources. Current mission difficulty affects alien numbers only and it is not much of a difference. But having different alien types on different missions can help with difficulty scaling. I'm saying "can" as this all are just a random thoughts and need a careful consideration. :smile:

Edited May 3, 2014 by wghost81

[Hobbes77]

Dang, I really wanted to stay out of this discussion, but hope the following helps people in thinking about their approach to the problem.

 

Strategy is 1) "the science and art of military command as applied to the overall planning and conduct of warfare", and 2) "A plan of action intended to accomplish a specific goal". [http://en.wiktionary.org/wiki/strategy]

 

XCOM is a game that (on both the Strategic and Tactical levels) limits us in what resource options we have in overall planning and conduct to accomplish the specific goal of defeating the Alien Invasion. This is quite reasonable and realistic given the premise of an unexpected and unknown enemy.

 

Due to the restrictions in what we can mod, we can't really add new resources. We can alter some parameters affecting those resources, such as how rapidly they can be made available, and quantities, but simply altering such parameters in favor of the player will substantially unbalance the game very quickly. And as a general rule, such modifications are generally sought only when 1) the initial play-through is felt to be too easy/difficult to complete successfully, or 2) more "replayability" exploring other potential options for success are desired.

 

The latter reason seems to be driving the desire to find ways to tweak the Strategy component of the game. So, what Strategic resources do we have available and control over (perhaps via mods) to leverage in achieving that goal?

 

* Base layout (Power plants, Access shafts, construction and grouping of similar facilities for bonuses)

* Research (Labs and Projects)

* Engineering (Workshops and Factory Projects)

* Funding (Gray Market, Satellite placement, and FC Missions)

* Recruitment and training (Barracks and Perks selection)

* Deployment of reaction forces (Interceptor bases)

* Intelligence (Satellites, Geoscope/Situation Room updates)

* Mission Management (acceptance and deployment choices)

 

There is one very important distinction to be made regarding strategy and the strategic layer.

 

While I agree with the definition of strategy being "A plan of action made to achieve a specific goal", I prefer the definition of strategy being composed of means, ways and ends. Ways are essentially tactics, what squad to deploy, what weapons for interceptors, etc., while the objectives are essentially to defeat the alien invasion through completing key storyline missions.

 

But when you look at means, then you're thinking specifically not about tactics or strategy, but about the operational level of warfare, which relates to getting the means to achieve your objectives. Things like running Research, Engineering, Base Layout, Recruiting, etc. are not a part of the strategic layer but they are a part of the HQ layer (or operational level), and their management is required for any strategy.

 

So if you remove the operational and tactical (meaning weapons, firing mechanics, explosives, etc.) layers, what is the strategic layer?

 

To me, the first aspect is that there are two factions opposed: the player and the AI, both with competing means, ways and objectives and the strategic layer consists of the interaction between both. And, IMO, if the general game setting is of an invasion by an advanced alien force I find it a bit lacking that most of the time it is the human player has all the initiative, while I'd expect that it would be the aliens dictating the pace.

 

The second aspect of strategy is choosing your battles and where to deploy your forces/satellites to help achieve your objectives (and also secure more means) while denying to the enemy it's own objectives and means. Here there are at least two player (Alien Base or Satellite Rush) strategies and two AI strategies: have XCOM get the Volunteer to Temple Ship or have 8 countries leave the council.

 

Now, like wghost81 has said on her previous post, making any chances will have unintended consequences. So what I would focus would be to slightly increase the number of possible UFOs per month from 2 to 3, while deploying larger UFOs to increase the difficulty of air missions and increasing the odds for retaliation UFOs. This would give more means to the aliens, without large changes such as modifying craft weapons stats, etc., and could possibly make Firestorm research even more important on mid game for interceptions.

 

The 2nd chance is to increase the probability of the 8 countries leaving the Council, and thus of the AI fulfilling this objective and winning the game. This is done by increasing the panic increases from certain events and by adding the Infiltration missions.

 

This would essentially change the AI factor but the player's tactical and base levels remain the same. As for player strategies, that would be up for testing to see if anything worthwhile happens with these changes.

Edited May 3, 2014 by Hobbes77

[Amineri]

I'm going to preface this remark by stating that I really want to see Hobbes succeed in making a viable alternative strategy rework mod that is lighter-weight than Long War has become. With the discontinuation of Warspace and Merciless, Long War is the only remaining "large scale" mod, which just isn't good for the XCOM modding community. So I'm happy to offer any advice (technical or design) that may help -- hopefully it does.

 

 

 

So if you remove the operational and tactical (meaning weapons, firing mechanics, explosives, etc.) layers, what is the strategic layer?

I think that this is an excellent distinction, separating the generic "strategic layer" coded into XComStrategyGame.upk into the operational and strategic layers.

 

We tried with Long War to make smaller changes to the strategy layer, but ultimately the biggest stumbling block we ran into is that the "Strategy AI" really isn't an AI in any real sense. What I mean here is that the aliens on the Strategy Layer aren't really reactive at all to what the player performs. Arguably the X-COM OG was similar. The only reactive thing that the XGStrategyAI code does is generate hunt missions when a UFO escapes. Other than that it's purely scripted based on month/randomized rolls. I think of it more as an "random encounter generator" instead of any real sort of AI.

 

This is very much not true in the tactical AI. The tactical AI makes decisions based on where XCOM troops are positions, whether it can move to flank them, what their health is, what it's to-hit chances are, whether it can one-shot them, etc. It's decision tree very much involves the state of the XCOM forces and decisions that the player makes. This makes for a true "game" in the game theoretic sense. The primary limitation of the tactical AI is that it doesn't perform any multi-turn planning.

 

In contrast the strategy layer simply generates a random set of missions each month (depending on the month and some randomized rolls). It is granted direct access to where satellites are -- abduction and terror missions are disallowed from happening where the player has satellites, while Scout/Hunt missions only happen where the player has satellites. There are two sides making decisions in the context of what the other side does -- instead the alien side is more like an "environment" that is essentially non-reactive to the player's actions.

 

 

two AI strategies: have XCOM get the Volunteer to Temple Ship or have 8 countries leave the council

 

The issue is that in now way are the AI Strategies for winning the game either implicitly or explicitly coded into the game. The "Strategy AI" simply creates a randomly generated set of objectives each month based on a static script. In no way does it make decisions about what actions to take in order to win the game.

 

Also, in vanilla EW there is a third way for the AI to win : winning the HQ Assault mission.

 

--------------------

JL has done a lot of work (and he's done the lion's share of the coding/design in this area) in reworking the StrategyAI to be more reactive to the decisions the player makes, as well as making decisions based on its own condition.

 

I've tried to limit the list below to things that are truly "strategic" in the sense that they are alterations to how the Alien Strategy AI makes decisions, and omitting operational things such as training troops and building your base.

- Added Alien Resource model that determines how many and what type of ships the aliens can field

- Added XCOM Threat model that determines what types of missions the aliens launch and whether they are pursuing other goals are trying to bring down XCOM

- Aliens get new "Aerial Bombardment" mission objective in which a UFO flies around a country dealing damage from the air, increasing panic -- essentially a form of airborne terror attack

- Aliens choose terror locations based both on adjacency to existing held countries and country panic

- Increased panic scale from 1-5 to 0-100 to allow finer granularity of panic results from various alien/XCOM actions

- Aliens can spawn Hunt Missions directly instead of only in response to escaped UFOs

- Changed code that makes UFOs flee when engaged to making it so that they flee only when badly damaged

- Aliens get new Resource Gathering mission that increases alien resources if successful -- these can happen in countries with/without satellite coverage

- Aliens spawn Alien Bases in each country lost to XCOM

- XCOM can launch satellites over lost countries to detect Alien Bases

- Aliens prioritize sending Hunt missions so shoot down Satellites over alien-controlled countries

- Added alien "Retaliate Objective" which creates a UFO that flies to XCOM base to initiate HQAssault mission -- this mission is spawned based on resources/XCOM threat perception -- multiple HQ Assaults per campaign are possible -- losing HQAssault no longer auto-loses the game but has significant consequences to resources

- Add MissionResult enum to allow the AI and panic code to distinguish between failed and successful crashed/landed UFO assault missions

 

 

There's also a whole host of alterations to how certain strategy mechanics work (which I'm not going to try and be comprehensive about), such as :

- Hunt mission success scales based on remaining UFO hull strength

- Terror Missions increase panic by +3 / +1 for each civilian lost (remember new panic scale!)

- Abduction missions only affect one country at a time, and only grant cash rewards

- Expanded FC Request system to request only collectable artifacts, and more frequently grant scientists/engineer rewards

- Countries get innate defense capability that allows them to resist panic increases -- this is improved by granting a country's FC Request

 

 

I've listed these not so much to say that they are "good" or an "improvement" (although JL and I think that they make the game more enjoyable for us), but to point out that such mods are technically possible, as well as to potentially spark discussion about what you want to accomplish.

[wghost81]

The way I see it from storyline point of view, aliens are testing the humanity. And AI "strategy" reflects this: with each month aliens get stronger and missions get harder. Aliens could have destroyed humanity in one strike, but instead they're applying growing pressure on it to allow it to quickly evolve and possibly "ascend". With this in mind scripted AI makes perfect sense, although less fun after several playthroughs.

 

But we can forget about main story and think about increasing replayability only.

 

The main idea for most of the rebalance mods is "it's not realistic, let's make it realistic!" But games are far from reality, especially sci-fi games. :smile: In fact, the most balanced games are the ones, which are very far from reality. That's why personally I'm always a little skeptical about "realistic" mods. But I'm going to try LW beta 7. :smile:

 

Deploying additional larger UFOs will cause panic increase and may be 1 or 2 additional country lost if the player is not ready. It won't be much for a strong game, but can be a very bad thing if you're already loosing. And without scaling down the resources, skipping additional UFO won't delay player's progress. On the other hand, if you scale down resources too much, you'll force player to always try to get additional UFOs, i.e. will make aircraft-related research mandatory.

 

From my personal experience, I don't like to loose. :smile: I like hard games and hard choices, but I also like to be able to better my skills and achieve a clear victory. What I'm trying to say, is that IMO there shouldn't be a situation in game you can't deal with if you have enough skills. Game can still be hard even if you know it's mechanics perfectly (like a chess!) and I hope someone will find a way to make late game in XCOM hard for any player, but still without random punishment and no-win situations.

Edited May 3, 2014 by wghost81

[Amineri]

 

 

From my personal experience, I don't like to loose. :smile: I like hard games and hard choices, but I also like to be able to better my skills and achieve a clear victory. What I'm trying to say, is that IMO there shouldn't be a situation in game you can't deal with if you have enough skills. Game can still be hard even if you know it's mechanics perfectly (like a chess!) and I hope someone will find a way to make late game in XCOM hard for any player, but still without random punishment and no-win situations.

 

We do run across this issue from time-to-time. One of the design philosophies in Long War is that not every battle is winnable, but the war is winnable. JL is the primary driving force behind this philosophy, but I'm on board with it.

 

JL can elaborate if he so chooses, but from my perspective if every individual battle has to be designed to be winnable than it decreases tension. Once a player realizes that some battles may not be winnable, it helps create a bit more fear. It also means that the player needs to account for a possible retreat if necessary to try and preserve as many troops as possible. Another case of "playing the tactical game to win the strategic".

 

 

 

The main idea for most of the rebalance mods is "it's not realistic, let's make it realistic!" But games are far from reality, especially sci-fi games. :smile: In fact, the most balanced games are the ones, which are very far from reality. That's why personally I'm always a little skeptical about "realistic" mods. But I'm going to try LW beta 7. :smile:

 

I agree that only letting "realism" drive the game can end up with a poorly balanced game. On the other hand, completely ignoring realism can lead to very confusing, counterintuitive situations. I general think of these concepts as "narrative" versus "game balance".

 

 

 

 

The way I see it from storyline point of view, aliens are testing the humanity. And AI "strategy" reflects this: with each month aliens get stronger and missions get harder. Aliens could have destroyed humanity in one strike, but instead they're applying growing pressure on it to allow it to quickly evolve and possibly "ascend". With this in mind scripted AI makes perfect sense, although less fun after several playthroughs.

 

But we can forget about main story and think about increasing replayability only.

 

I did go through the exercise of thinking about what it would be like to try and turn the strategy layer of the game into a game where either side could be player by a human, or both sides could. That is, what sort of game rules would have to be added to allow a human to play the alien side at the strategy level and make for an interesting, balanced game? Not all such games need be symmetric to be balanced/fun. One of my favorite games as a child was "The Awful Green Things From Outer Space", which was designed to be assymetric but I found quite fun.

 

Of course, actually making the alien side playable by a human would require a lot more modding than is probably worth it, and may not even be technically possible. However going through the exercise I found interesting in trying to create systems that at least make it appear that the aliens could be following some sort of plan.

 

In terms of a game-play goal for the aliens, my thought was that the aliens don't just "win" or "lose", but can get a score from winning, which the goal is to maximize. In our case we have modeled a "research" resource which coarsely emulates the XCOM research progress. Alien research improves their UFOs, allows fielding of better alien types, improves stats of existing aliens, grants perks to some aliens, increases alien pod size, etc. My best conception for the aliens is to defeat XCOM while maximizing this research resource.

[wghost81]

When I was experimenting with random pods + larger pods + exalt waves, I unintentionally ended up creating missions with 40-50 EXALTs on map. It wasn't very fun, but I discovered it was very winnable: rockets + in the zone = dozens of dead enemies in one turn. The point is, IMO, no matter how hard you'll try to create no-win situations, people will still try to find a way to beat them. Because it's in human nature and in the nature of computer games: we play to win, not to loose. Even in Tetris people still believe they can win. :smile:

 

Vanilla late-game problem is that aliens can't hurt your soldiers let alone kill them. So no injuries and no need for replacements: late game is won by 6 colonels, wiping out aliens the moment they get discovered. This is a problem of both tactical and strategical layer, IMO.

 

In terms of a game-play goal for the aliens, my thought was that the aliens don't just "win" or "lose", but can get a score from winning, which the goal is to maximize. In our case we have modeled a "research" resource which coarsely emulates the XCOM research progress. Alien research improves their UFOs, allows fielding of better alien types, improves stats of existing aliens, grants perks to some aliens, increases alien pod size, etc. My best conception for the aliens is to defeat XCOM while maximizing this research resource.

Although this approach is more realistic, IMO, it creates even bigger "Slippery Slope" effect: if aliens winning, they're getting stronger and harder to defeat, and if humans are winning, aliens loose their research resources and get weaker and easier to defeat. In the end it leads to "artificial" game-design decisions, like mandatory loss of one country to ensure aliens will have at least one base, no-win terror missions and ambush missions. They shouldn't exist in the concept of "equal conflict sides", but if you remove those from the game and make sides truly equal, humans will always win and win quickly.

 

BTW, why stick to unused vanilla concept of Infiltration missions? Why not make this more like OG and allow aliens to randomly place a base anywhere? Chance of discovering a base can depend on satellites and ships activity like in the OG.

 

I just remembered another thing from one of the Julian Gollop interviews. He said that AI should not be smart, but should seem smart. AI in the OG was random and was adjusting to player's performance, so it seemed very smart. But in the terms of modern games AI was just a plain simple cheater at strategic layer. :smile:

Edited May 4, 2014 by wghost81

[Amineri]

 

 

When I was experimenting with random pods + larger pods + exalt waves, I unintentionally ended up creating missions with 40-50 EXALTs on map. It wasn't very fun, but I discovered it was very winnable: rockets + in the zone = dozens of dead enemies in one turn. The point is, IMO, no matter how hard you'll try to create no-win situations, people will still try to find a way to beat them. Because it's in human nature and in the nature of computer games: we play to win, not to loose. Even in Tetris people still believe they can win. :smile:

 

Vanilla late-game problem is that aliens can't hurt your soldiers let alone kill them. So no injuries and no need for replacements: late game is won by 6 colonels, wiping out aliens the moment they get discovered. This is a problem of both tactical and strategical layer, IMO.

 

I agree. No amount of "throwing more aliens" at the player is really going to up the challenge. That's why I created the mods that allow for upgrading of alien stats during the campaign, as well as creating leaders/navigators that can get perks. Aliens can also be configured to have/increase innate Damage Reduction and regeneration. Lots of tools to make the enemies tougher, beyond just adding more aliens.

 

Of course it's easy to go overboard with that kind of stuff, which is why I made it all configurable in the DGC.ini.

 

So far our goal of switching up the difficulty curve seems to be working -- players generally report that the first couple of months are a bit easier than vanilla but then the difficulty starts to substantially ramp up. We might still have some issues with too-quick ramp-up of difficulty in months 4-5, which we're trying to address.

 

 

 

I just remembered another thing from one of the Julian Gollop interviews. He said that AI should not be smart, but should seem smart. AI in the OG was random and was adjusting to player's performance, so it seemed very smart. But in the terms of modern games AI was just a plain simple cheater and strategic layer. :smile:

I kind of feel that's the way human brains work. They aren't actually smart, they just seem smart :p.

The other thing that I remember from a Julian Gollop interview is creating a lot of simple systems that have a fair bit of interactivity with each other, which results in emergent behavior. There's a few other studios out there that do that, like Arcen Games -- I'm sure there are others.

 

 

Although this approach is more realistic, IMO, it creates even bigger "Slippery Slope" effect: if aliens winning, they're getting stronger and harder to defeat, and if humans are winning, aliens loose their research resources and get weaker and easier to defeat. In the end it leads to "artificial" game-design decisions, like mandatory loss of one country to ensure aliens will have at least one base, no-win terror missions and ambush missions. They shouldn't exist in the concept of "equal conflict sides", but if you remove those from the game and make sides truly equal, humans will always win and win quickly.

 

Agreed. I still haven't hit upon a good way to introduce a negative feedback loop to prevent catastrophic failure in either direction. Long War still suffers from it, although I'd like to think that it's not quite as bad as vanilla -- I'm probably deluding myself, though :)

 

 

BTW, why stick to unused vanilla concept of Infiltration missions? Why not make this more like OG and allow aliens to randomly place a base anywhere? Chance of discovering a base can depend on satellites and ships activity like in the OG.

That's a good idea. Aliens could build bases in countries which would use up resources. I suppose an alien base could also generate panic on a continuing bases. Will have to ponder the possibilities...

 

-------------

Anyhow, I hope I'm not derailing Hobbes' thread too much. The original goal was to figure out how to make some imporvements to the Strategy Layer. I daresay it's an ambitious goal, given all the interviews indicating the Jake Solomon and Sid Meier spent a lot of time trying to create a good strategy layer.

[wghost81]

I think our problem is a kind of "experienced player" problem. I mean, it's hard to create interesting game situations for experienced player. While still learning, player makes a lot of mistakes on strategic layer, but the game is designed so he can correct for those mistakes and still win. And it's a good design. But it has a drawback of being increasingly easier for experienced player.

 

I think, negative feedback in it's simplest form can be a solution here. Although it kinda punishes a good player for a good performance, it gets the job done. To minimize this punishing effect easier missions can be allowed late game to train more rookies, as player is expected to take losses (or at least injuries) due to overall harder missions. And this will be up to player: let your colonels to wipe the floor with aliens and then go on harder mission and loose half of those colonels, or let your less experienced soldiers to earn more XP and level up.

 

BTW, about late-game tactical layer challenges. I found that larger alien pods are not as good as it seems, because aliens go for the nearby cover and cluster. One rocket and in the zone (especially with insanely buffed EW snap shoot snipers) is usually enough to kill everyone. So now I'm thinking of utilizing "call for help" mechanics, which is currently implemented for UFO commanders only: active pack can call nearby packs for help. Not all, but 1-2 closest. If combined with random pod placement, this can create a situations when player gets attacked from several different directions, which is harder than just one bigger pod in one direction.

[wghost81]

I suppose an alien base could also generate panic on a continuing bases.

This will allow for easier base detection by the player, which is not good.

[wghost81]

I've been thinking about "Diminishing Returns" and how it affects strategic layer decisions. Although it practically forces aggressive strategy, it is still fun. May be slightly less punishing version of "Diminishing Returns" will help?

 

Without "Diminishing Returns" you can cover all the remaining countries even without satellite rush strategy, as they get cheaper with more engineers and you get more money with more satellites up. It's a very powerful positive feedback, which allows for satellite rush strategy in the firs place. But what if satellite cost will not depend on number of engineers? What if it will stay the same ($100) or increase a little? And what if batch-building satellites will be equivalent to individual building in terms of cost?

 

Another interesting possibility is to decrease Satellite Uplink capacity to 1 satellite (+1 for each adjacent uplink) while simultaneously decreasing engineer numbers requirements to +5 for each uplink. (Because with vanilla +10 it will force player to go for engineers.)

 

And another thing. With aggressive strategy and early alien base assault panic management actually gets easier, as you will usually get only one panic-increasing situation (abduction) and two panic-decreasing situations, one of which affects the entire continent (council mission + terror mission). What if council and terror missions will still happen only once in a month, as they do prior to base assault, and additional mission will be big UFO mission instead?

Edited May 4, 2014 by wghost81