As I said, I think PP is, at heart, a miniatures games with ballistics, so in this case I evaluate the quality of the AI on the basis of how well it emulates the challenge that can be offered by a human player.
In an RPG, I would probably not do that, unless there were actually some mechanics that made it all work satisfyingly.
I completely agree with you on SM Civilizations though, they shouldn’t have gone full board game and kept at least some simulation elements. Not sure if it’s so much attributable to AI as to game mechanics though.
Of course, that’s why the AI needs some “help” to present some challenge to the player, either because it “cheats” (has more or better troops, or is unaffected by the fog of war), or because the sides are completely asymmetric. The latter is definitely the case with all the XComs. As far as I’m aware, it’s not clear whether or how the AI cheats in PP, but I wouldn’t rule it out.
Before calling out BS, please quote me in full
as you can see I did caveat my assertion with except when there are no obstacles. This is because as soon as there is an obstacle in the way of the LOS the player can’t determine (you can guess, and the more your play the better you get at guessing) a priori how exposed the target is going to be before moving to a certain tile. This doesn’t happen in FXcom, where exposure is solely determined by whether the cover is partial or total (and the aiming angles doesn’t make the determination much harder).
I consider myself as human player and I am pretty sure I do evaluate exposure and it isn’t always trivial. So why should I quote some additional conditions if I know better what I do and what I do not? If it makes a difference to you:
Still manure, I feel quite confident navigating between obstacles. Result isn’t always as I imagine it but most of the time I’m able to find a better position. I find it natural and I can’t imagine playing without mobility towards flanking the enemy.
I’m talking about ballistic system and finding the better position in PP. It’s natural, as the LoS, model hitbox and direct targeting are much easier to judge for a human than abstract percentage system. In very non-trivial situations I still evaluate exposure to make decision if I should move or don’t move at all. Or if can’t find better exposure, but can find a place with more-less the same maybe I should still move a bit to be less exposed to enemy fire.
But I struggle to figure out what are you trying to prove? That human isn’t capable of tactical approach so enemies also shouldn’t do it and therefore current AI is pretty good? AI doesn’t evaluate even trivial exposure. It would fail (and fails in a similar conditions) at snowball example. Some LoS? Fire.
Yeah, as I said, you can guess with some degree of success the exposure from a given tile after having played the game for a while.
Must the AI do the same thing? IMO, if it doesn’t, it doesn’t make it useless. But if this ruins PP for you, by all means, create a topic, or a ticket on canny and make your case. If it’s not so hard and other players and the devs like it, it might get done.
I will say though that it looks to me like you and BoredEngineer want the AI to do opposite things: you want it to play more like a human player and BoredEngineer apparently wants it “to stay in character”. I would think that to stay in character crabs shouldn’t care about cover much…
Certainly the AI has some help - And that’s the point, from what I’ve seen players who say that the game is too difficult aren’t saying it’s difficult because they AI is ‘too good’, they’re saying it because they feel that the help which the AI is being given is ‘too much’.
Well, here is hypothetical train of thought when you see that AI heavily prioritizes attacking the guy with a sniper rifle over others:
Perhaps Pandorians have schools where young hatchlings learn from the start what is the difference in weapon classes and which is the most dangerous. They are trained for years so that prioritizing targets becomes a second nature overruling any thought of self preservation or tactical behavior.
There is some sort of “commander” entity that controls each Pandorian during the battle using perhaps telepathy. These commanders are either not very bright or have goals which are much higher level that a regular soldier of PP can’t comprehend or interpret.
Both 1 and 2 and the training and commander perception are so strong that using any sniper rifle, even masking it as a crowbar still produced the same result - enemy clearly knew who was carrying what, there has to be some explanation to this but PP yet have to find it.
In the AI code, enemies are prioritized by the class of the weapon they are using and some weapons have too high priority compared to other variables like proximity and danger.
First 3 have no grounds in my understanding of the game and never communicated in the game, so I’m left with number 4. Which is equivalent to seeing enemies spawn in front of you in the shooter, cars driving at 3x their top speed to catchup with you in a racing, enemies having no concept of stamina while you have to manage it and etc. These are all things that sometimes are necessary but it’s important to make them less visible, they make a cheap impression of your game and can worsen opinion of the player even when non warranted. For something to be a masterpiece or a great game, these have to be cleverly masked or covered up by other mechanics. Like for example in Skylanders, when vehicle get airborne because of collision or a ramp, the game will calculate the exact way how to naturally rotate the vehicle so it always lands on wheels like it was just meant to be. Even those who noticed it, where amazed how it was almost impossible to break this behavior or make it looking unnatural.
Reminding the player on every step of the game that this is no more than a computer program is the opposite of a masterpiece or a great game.
Not that I have noticed, no. Most of the time I get the feeling they go for the closest soldier around. BTW, weren’t you the one using only snipers?
Anyway, there is a commander unit off the battlefield (that’s part of the lore) and the Pandorans are former humans, so there were sufficient reasons to put a stop to that particular train of thought before it went off the rails into Panda nursery school territory.
Seriously though, there is not one right way for a game, or a great game to be. There is no laundry list to check one by one the items, features, or elements it must have. Any game - great games and masterpieces included - will get some things right, and some things not. It will pay attention to some aspect of the gameplay experience and neglect another. It will please some players and displease others. For example, and BTW, I like your thoughts on metagaming - I just disagree when you present them as the only valid approach to the “right” game design (or the right game design for a particular genre, setting, franchise, etc.)
Yeah, you are right. I forgot that telepathic command means that Pandas have to behave like the grunts in F.E.A.R.
Where do you want it to be shown? (or rather where does it have to be shown in order to satisfy the universally established critieria on the right game design)?
Guys, this is one shallow well you are trying to draw water from.
Why don’t you offer your vision of how you want the AI to be? Give us a turn, like there are x crabs and y tritons (who survive the alpha strike from PP unscathed), what do you want them to do? Are they carefully moving from cover to cover, and covering each other? Do they prioritize self preservation and hide? Do they set up a coordinated defensive position? What do you want them to do?
Grunts in FEAR are also under telepathic control, so good point here. Glad we’re on the same page.
Sarcasm aside (or just a little bit to the side), they’re after snipers because they’re former humans and carry on some of memory and are under telepathic command of highly intelligent yet sinister being. They stand still and shoot cover because apparently during shooting they’re just dumb crustaceans without any command and with suppressed human genes and memory. Makes sense
So the game doesn’t communicate that and it’s just your own explanation? Do you see a problem with that?
We did in other numerous threads. Moreover, why we do we even have to do that as we are not designers of the game? Telling what we want them to do, implies designing the whole motivation layer for the Pandas behavior. Some consistent idea of why they are doing what they are doing. You don’t even have to use some cutscenes or words for this, case and point aliens in Factorio - drawn to pollution and destroy any source of it. Granted it doesn’t have to be so simple but the main point is that this clear motivation is essential for world building and subsequent game mechanics.
Maybe they have quantum brains to predict outcomes but a primitive central cortex to execute commands, so they are smart beyound imagination but being limited to a primitive form of expression appear inconsistent and silly.
EDIT: To answer your question - it should be shown on screen, preferably in gameplay. A book on basics of movie creation is enough to learn how mindless hive minds are presented in comparison to group of free thinkers. Can be done in both animation, visual appearance and interaction. We have sounds too, a hive mind drone doesn’t scream to encourage his allies. An individual with self preservation can cover and scream in fear. An individual with empathy can get enraged by seeing comrade die or rush to help him.
If your antoganists are some sort of war machines like robots in Terminator, then viewer will expect efficiency of machine over ingenuity of an individual.
Btw, not sure why your brought FEAR to this. the AI in FEAR is based on individual GOAP system, so you are practically fighting individuals with some rudimentary communication in between each other. Each has a distinct goal and checks GOAP tree to find the next action which gets him closer to the goal. The lore there touches the motivations of the soldiers, not their individual actions in tactical situation.
Which would dictate a tactics very similar to what some players do - soften up as many enemies as possible then trigger panic loop. Focusing on a single individual does exactly the opposite. Either grand commander is just stupid or simply doesn’t exist.
Sarcasm aside, the first point is to figure out whether you want the AI to emulate a human player, or you want it to do something else - just to simplify things, what I call “to stay in character”.
If it is the former, it is a question of detecting those AI behaviors that you consider don’t emulate the human player well so that they can be checked, discussed and possibly altered.
For example, the AI strongly prioritizing certain targets - which, TBH, I haven’t noticed and haven’t read about in the forum either- , might go against staying in character, but arguably it is something that a human player would do.
Shots at partially covered targets? Perhaps right for staying character, perhaps suboptimal for a human player.
Because this a discussion between players, I don’t work at the complaints department.
It’s a great idea, but as far tactical battles are concerned this wasn’t done in any XCom game, hasn’t been promised by the devs and it’s the first time I read of anyone suggesting it on the forum.
When it comes to tactical battles the premise has always been more or less the same: the aliens are “slaves” to some superior species or being and do its bidding (the mediate or immediate goal of these beings being world domination).
The game lore makes it very clear that they are not mindless drones.
The Pandas pretty much follow the same template as the other XComs - the creatures are “slaves”, so they have some individuality and self preservation instinct but it’s rather toned down.
EDIT: took out the spoiler as I can’t hide it on the phone
In Xenonauts your first research item is “Alien Invasion” that explains what is going on. It explains that the final goal of the invasion is not clear but it’s evident that aliens are rebuilding their fleet to be able to use it in atmosphere. Therefore ships come in different sizes and carry different missions. Small ufos, such as scouts, interceptors and etc. will be abducting people and animals, attacking ships, airplanes and small military installations. Larger ships will carry terror missions, bombing runs and direct attacks on major military installations. This is clearly communicated on the strategic map.
When it comes to tactical battles there is a clear difference between terror missions, base attacks and cleaning up downed UFOs. While the tactical behavior is generally the same, the troops composition is not. Most of the aliens will behave consistently to their class and abilities. This is where XCom actually shines as troopers like Officers tend to take a backrow and let grunts be on offensive. Back in Xenonauts, an officer on a downed UFO will tend to stay inside in safety, the same with key personal but an assault units will seek to fight opposition outside as soon as possible. On top of that you have racial and class difference, like discs and androns move differently on the map. That makes them behave as individuals. So while they are the slaves to a superior alien race, they are not drones and act accordingly.
This is not something that you promise to players, that is something for the development team to be used as pivot. When animator asks what kind of animations we need, the answer shouldn’t be “Yes!” but what specific behavior such animation should convey. The choice of that behavior comes from the a larger model of what your setting is. The same model implies the programmatic behavior of the units. Which behaves their abilities and etc.
Let’s add some meat into this in the context of PP. I personally have no clue what the grand idea is. But let’s say Pandorian goal is simply to expand and increase population. They have a biological limit to how deep in the ocean they can go so the next possibility is land. On land they face opposition in a form of survived people and several militant organizations. Luckily these organizations are not aligned and therefore the humanity is divided. The logical way to approach land invasion would be to focus efforts on a single factions and to prevent others from helping them do a smaller scale attacks on them to keep them distracted. Strategically, opposition requires heavens production and population capacity to provide resistance to invasion. Therefore heavens should be systematically destroyed. As it takes a lot of time to grow and train a human being, destruction of materials can provide a better outcome than focusing on manpower. This is reinforced by the fact that humans don’t launch a large scale counter attacks involving bombardment or heavy assault vehicles and operate solely by small groups of troops. Which means there are no strategic land points that require immediate attention. So slowly expanding your territory and systematically eliminating heavens is the way to go. This type of expansion is very suitable to a biological and chemical warfare as Pandorians don’t give a crap about farming or a complete destruction of the industrial capacity as Pandorians are not going to use factories that humans build. Therefore in a current setting a typical heaven attack should consist of dozen of mortar chirons bombing heaven into oblivion, while smaller crabbies just covering them. It’s evident that this might create a very one sided missions, so some other parameters have to be changed. Like what if Pandorians, just like Marsians in the War of the Worlds require high density organic flesh to grow food for their population. So now, the strategy will be different, Pandas have to surround and attack heavens in such a way that organic matter can be easily harvested and destruction of materials would be done only as offensive raid operation. Later followed by a larger organic harvesting operation.
This is getting pretty long and is taking too much of my time. I hope you get a gist of it. This type of stuff can go pretty deep but not doing it is just lazy as there is a ton of materials on writing and concept design. At this point it’s rather pointless anyway as it has to be thought out way before a single line of code is put in place or if team has resources to do major changes to re-align everything. It can’t be done perfectly but doing it even at minimum largely removes contradicting or en-coungruent game mechanics or behavior from the game. Like I still have no clue what is the basis for different factions having a different specialization for weapons besides a generic desire of the designer to have “different” factions. A typical “people from the north” or “desert nomads” in fantasy games is at least dictated by environment, cultural borders and general cultural rigidity. In PP this makes little to no sense as all settlements are all over the place and mixed together. Their cultural difference makes no sense either and reminds me of a shitty culture building in Hunger Games where highly technologically advanced society still needs someone to grow cabbages by hand.
Again, you are just providing a self-explanation based on desire to protect the game as this is not what happens in the game. Sometimes crabbies will run off the map as they don’t have any hands left to attack you with, sometimes they will ignore shotgun to the face just to punch someone one time, the chirons are absolutely dumb asses who won’t even move while being shot at for multiple turns. The facehuggers while will try to sneak to you have no concept of lone soldier or tight group of soldiers. No-on has a concept of ambush or changing behavior when outnumbered or outnumbering. Some triton will be hiding in a single rum whole mission, some other idiot will persist on attacking civs while being shot at from all sides, few turns later he is now a sneaky guy. This is just bizzare, like trying to document all characters on Bosch paintings.
