Non-Deterministic Combat

Discussion in 'Planetary Annihilation General Discussion' started by ledarsi, February 3, 2013.

  1. ledarsi

    ledarsi Post Master General

    Messages:
    1,381
    Likes Received:
    935
    Supreme Commander and Forged Alliance are both strong games, however they both suffer from a particular type of gameplay pathology that I hope that PA will avoid. Their combat is extremely deterministic, with very little variation between similar situations.

    I think PA should have less deterministic, more interesting combat fundamentals than units as blobs of a certain amount of HP that deal a fixed amount of constant DPS. Making combat basics more interesting also has cascading effects as the war gets larger, with smaller groups of mixed unit types fight one another in essentially endless combinations.


    Pattern Weaponry

    Total Annihilation had a variety of weapons which exhibited patterned behavior other than simply dealing direct damage to the target. For example, the Rocko and Storm fire slow rockets in a straight line. Any particular individual shot will be hit-or-miss, but in aggregate this type of projectile becomes more statistically predictable.

    Weapons with distinct firing patterns with gameplay consequences are also much more interesting to use, and to fight against. A pattern weapon is more than just a justification for constant DPS to nearby enemies, it has a presence in space which has combat effects other than its damage. Spraying machine guns behave quite differently than linear rockets or other patterns, and a particular situation may make each perform better or worse in complicated ways.

    A weapon's behavior can also be based on more than just its shape. A missile weapon might fire three missiles with a short interval, and then need to pause for a while. A sniper might need to train on its target for a short period before it can fire. These types of features do increase the effectiveness of a unit with micro, but they are also easy to automate to be effective enough without player supervision that in large groups, or in large games, micromanagement is usually not worthwhile.


    Wild Inaccuracy

    In SupCom/FA, the excellent unit AI and the responsiveness of most weapons creates a battlefield which is extremely predictable. It creates a very neat, very clean and deterministic war where weapons always hit their targets, and where one unit of type A always kills a unit of type B is precisely 5.2 seconds. What I propose is to have PA be filled with weapons that miss. Very often.

    Inaccuracy is very good for gameplay (at least for physics based games) for three reasons. Firstly, it makes individual fights less deterministic while preserving the broad statistical picture. Generally, players can make effective predictions about what is likely to happen when two forces collide, which is very important. However this assessment is always bounded by some level of uncertainty, which is also good.

    Secondly, low accuracy very naturally discourages huge force concentrations in one location in favor of more efficient distributions of smaller groups. A large, tight cluster of units is much more likely to take damage from a highly inaccurate incoming shot, because the shooter doesn't really care which unit it hits, as long as it hits something.

    Thirdly, inaccuracy creates a wonderful tool for creating more complex unit designs that are also intuitive to the player. For example, it is very natural to have units be more accurate at closer ranges, or be more likely to hit a much larger target. Fewer of its shots miss, effectively increasing its DPS. However its weapon is unchanged, and there is no arbitrary damage type or damage assignment based on range.


    Low Cost and High Lethality

    One of the hallmarks of Total Annihilation compared to many other RTS games of the era was the relative low cost of combat units, and how easily almost everything in the game was destroyed.

    Low cost very directly allows players to build more units. Assuming the game's units and mechanics are designed well to create interesting gameplay and decisions about how to use these units, having more units creates more choices for the player, and fills up more of the board.

    High lethality, where units die easily to weapons, is good because it causes attrition. Larger forces suffer casualties even when fighting against smaller or weaker forces. Because a unit remains 100% effective until its HP completely runs out, there is a certain minimum amount of damage required before the effective strength of a group actually changes. This damage threshold should be low; and the way to do this is to have units be killed easily, diminishing the combat strength of a group of units as it takes damage.

    Put simply, even a single unit out on the map might inflict one casualty on even a very large enemy army. If both sides are more durable, then the one unit will be killed, and will have done absolutely nothing to the large enemy force.

    Conclusion

    Supreme Commander and Forged Alliance had quite deterministic battles, resulting in quite uninteresting actual fighting. I propose that PA import a few of the design concepts from TA regarding the low cost and durability, with high damage of units, as well as the unpredictability of inaccurate weapons and . Together, these mechanics discourage deathballs and create more variation in game states between multiple smaller battles while preserving the big-picture predictability of a chess match.
  2. sylvesterink

    sylvesterink Active Member

    Messages:
    907
    Likes Received:
    41
    The issue on TA's side is that it's methods favor micromanagement. Not a huge amount, but enough that it matters. I'm not so sure I'd readily call battles in FA uninteresting either. Perhaps if you zoomed in and watched them close up, sure, but on the larger scale I found them to be every bit as interesting as TA, but in a different way.

    Personally, I think there needs to be a middle ground. The thing that made TA's combat so varied was the variety of units, which is what FA lacked. However, in FA, the weapons felt a bit more correct. The accuracy made the units feel a bit smarter and more effective, rather than some dumb AI spraying its load all over the battlefield, which is all too often what TA felt like. A big contributor to this was that in FA the AI was smart enough to lead its shots, which just makes more sense anyway. All too often I would play TA and thing "WTF, troops, shoot your guns just a millimeter forward and you'll have him!"

    One way to implement the accuracy and intelligence of FA with the weapon variety of TA would be the route Zero K took, with the AI doing the unit micro. A bot shooting a rocket shot it very much like TA, in a straight line forward, but it was smart enough to lead its shots. Slow bots would get hit, but there were fast bots that could "swarm" via their AI and become very hard for the rockets to hit. (They wouldn't actually dodge, just move erratically, which meant some rockets did connect.) The system might not be perfect for PA, but it's a good starting point.
  3. cola_colin

    cola_colin Moderator Alumni

    Messages:
    12,074
    Likes Received:
    16,221
    Nondeterministic gameplay = lets role the dice.

    Nobody wants that. Players need to be able to learn what happens in certain situations. If it randomly changes it would pretty much invalidate any "serious" effort to improve. We could just play some dice-game instead of PA.
  4. igncom1

    igncom1 Post Master General

    Messages:
    7,961
    Likes Received:
    3,132
    Why not have the last 5-10% of a units damage dice roled?

    That way they can still be expected to perform to a dagree, but the rest is up to the way the wind blows and such.
  5. cola_colin

    cola_colin Moderator Alumni

    Messages:
    12,074
    Likes Received:
    16,221
    Make Units have a very high fire rate with a constant amount of 5% missing and it stays perfectly deterministic anyway. If the fire rate is low it can create situations where that 5% dice role cost you the game.

    But Units that miss on a X% base make no sense in a game that actually simulates shots.
  6. ledarsi

    ledarsi Post Master General

    Messages:
    1,381
    Likes Received:
    935
    I think you guys have missed the idea wholesale. "Non-deterministic" does not automatically entail randomness. They are predictable factors that exhibit complex behavior that players can utilize and plan around.

    Inaccurate weapons don't hit "randomly." A shell weapon lobbed vertically that is inaccurate is more likely to deal more damage against a larger cluster of enemies. These are factors that players can control and plan around- very unlike random dice rolls. For example, clustering units together against an inaccurate artillery shell greatly increases the odds of a hit, and increases the likely amount of damage- and this isn't a random effect.

    A shot that goes wide to one side isn't a result of a dice roll going awry which will do nothing because the RNG said so. That shot can hit something else- or even might hit the target the shooter was aiming at depending on how large it is, or how it moves. Spraying bullets is completely different from having an RNG determine only 1 out of every N shots will inflict damage, the rest zero, despite the visual effects. Those shots actually exist on the game board, and have effects that create move and countermove options. Their effects are anything but random, even if some of the time they miss.

    Doing random amounts of damage is even worse than having damage types. Obviously totally random occurrences are terrible for gameplay as then the outcome of a battle, and even the entire outcome of the game becomes subject to totally random chance, and not player choices and actions.
  7. Pluisjen

    Pluisjen Member

    Messages:
    701
    Likes Received:
    3
    I think I'm pretty much on ledarsi's side here. No randomness beyond the starting angle of velocity of the shot, and let the physics take it from there. If it hits a tank head on, damage. If it hits a tank full in the rear, extra damage. If it misses, just see where it lands and apply damage to that.

    As for the micro requirements, I think most of them can (and will) be easily avoided by making smarter AI. Considering we can mod the front end, adding in options to micro specific units shouldn't take long.
  8. syox

    syox Member

    Messages:
    859
    Likes Received:
    3
    ^ this
  9. sstagg1

    sstagg1 Member

    Messages:
    214
    Likes Received:
    0
    Except for the part about location based damage. That would introduce a whole **** ton of micromanagement.

    I personally enjoyed TA battles much more than SC/FA....

    ... but that may have been because the pathfinding made everything a giant cluster-****, and I used lots of splash.

    Also, I rarely/never played competitively.
    Last edited: February 3, 2013
  10. bobucles

    bobucles Post Master General

    Messages:
    3,388
    Likes Received:
    558
    Your basic unit is built around effective and reliable weapons. If this was not true, they would not be basic units!

    However, luck is a tradeoff just like anything else. An advanced unit could very well sacrifice reliability to gain great power. A good example is long range artillery. It lacks perfect aim, but that doesn't matter because you WANT it to deal wide spread damage at extreme range. It's not built for short range, deterministic warfare, because it's taking risks at a much higher strategic level.

    A unit with a shotgun spread sacrifices direct damage to be more effective against multiple targets. This is great for riot units, and can be handy against singular large units as well.

    If you don't like those trades, then stick with more reliable weapons like missiles, death rays and asteroids.
  11. Pluisjen

    Pluisjen Member

    Messages:
    701
    Likes Received:
    3
    Only if you assume that your units will be completely retarded. Micro is generally nothing more than the game forcing you to do basic tasks it could do itself but won't. You can make your tanks automatically move their front towards the enemy, that way no micro is needed.
  12. cola_colin

    cola_colin Moderator Alumni

    Messages:
    12,074
    Likes Received:
    16,221
    I think that's overestimating the capabilities of unit AIs. Making an AI that really takes all micro away is pretty unrealistic. There will always be situations where it is far more effective if the human player gives a few extra commands to increase efficiency.
    If you introduce alot of micro-intense tasks it makes it even more likely that the AI will be overextended. It also makes no sense to introduce i.e. hitregions on units and then make an AI that uses the hitregions perfectly. Why even introduce hitregions if the AI will take care of them anyway?
  13. Pluisjen

    Pluisjen Member

    Messages:
    701
    Likes Received:
    3
    Because a grand scale strategic move can remove the ability to properly respond. It's not that hard to make a unit AI that will automatically turn the front of the tank to the largest cluster of enemies, and that will work in 99% of the situations (being pretty micro light)

    But it won't stop a two-flanked attack or ambushing units from the rear from being lethally effective, regardless of whether the AI or the player will micro them, because there is no right direction anymore.

    That said I don't know if this feature will be in the game, but I don't think that it'd be too micro heavy to be honest, as long as the units are somewhat intelligent about it. (And of course you'd need a broad "rotate this way" command or such)

    I think it'd be pretty low priority though, so it probably won't get in anyway :)
  14. torklan

    torklan New Member

    Messages:
    15
    Likes Received:
    0
    Here is a simple solution that helps the "My blob is bigger then your blob so I will all ways win" issue that seems to be at the core of this argument. Give every unit a base 5% chance to miss. Increase that chance if one or both units are moving.

    Then give all units a 10% chance to critically hit doing 50% more damage in that one attack.

    By adding these random elements into the game we don't need to add a layer if micro unless you want to get an additional bonus of being missed but that adds an equal chance of missing your targets as well.

    That simply give us a chance to have a smaller army win against a slightly larger one if they get lucky.
  15. KNight

    KNight Post Master General

    Messages:
    7,681
    Likes Received:
    3,268
    No, Luck should have nothing to do with a smaller army winning, better tactics, Unit Composition and (to a degree) Micro is what should be determining the outcome to a battle.

    Mike
  16. syox

    syox Member

    Messages:
    859
    Likes Received:
    3
    No critical hits please.
  17. stevenside

    stevenside Member

    Messages:
    113
    Likes Received:
    1
    To be honest, some weapon systems could have a bigger random factor than others. Like Flak. Flak is a weapon that has a high fire rate, but a random aim that does AoE anti air damage. If there could be systems like these incorporated into the game, i wouldnt be unhappy. Flak is one thing i'd like to see (long lasting black flak clouds in the air, hated the lame looking FA flak)

    MIRV missiles which randomly splits up and hits different spots around the battlefield could be a cool mid-longrange weapon system that can be incorporated into a army.

    Also, the physics of the game, they have already mentioned that terrain deformation will be in the game. Using craters and hotspots is some of the micro that might appear into the game, but that micro is so simple that anyone with simple logic in their mind can use it.
  18. cola_colin

    cola_colin Moderator Alumni

    Messages:
    12,074
    Likes Received:
    16,221
    That's a valid point. It would be annoying to deal with the direction the tanks are facing, but with the AI the player would only have to take care that the units are not being overrun from multiple sites.
    But I don't see this feature coming and tbh I wont miss it. There are mo important thing.


    The better player is supposed to win. Not the one who gets a lucky dice-roll.
    Gameplay can only suffer from dice-roles.
  19. Pluisjen

    Pluisjen Member

    Messages:
    701
    Likes Received:
    3
    Agreed. And I'm guessing it'll be there in spirit if you look at the way armies will (hopefully) align, with long-ranged, soft-shelled vehicles at the rear and tough, short ranged units at the front. You'll still be screwed if your opponent comes from the rear because you'll be catching shells with your artillery while your tanks can't reach the attackers with their little cannons.
  20. ledarsi

    ledarsi Post Master General

    Messages:
    1,381
    Likes Received:
    935
    A good example of a basic weapon that is less deterministic than combat in SupCom/FA is the EMG from Total Annihilation, used by the Peewee, Flash, and Brawler.

    The EMG fires continuously in a stream at the target, with fairly good accuracy, but it does miss. It also can be blocked by walls or wreckage, and doesn't arc over terrain. Consequently, throwing 100 units with EMG's into a battle against 100 enemies equipped with EMG's is not as predictable as doing the same 100 vs 100 test in FA.

    However, which side wins and how much they win by is not determined by chance- it is determined by emergence based on clear features of the battle. Players can make decisions based on these features, such as deliberately choosing to fight somewhere with less wreckage if they have the larger army. Or opting to incorporate other weaponry which arcs over obstacles to gain an advantage over an EMG equipped enemy in an area full of obstacles.

    I'm not necessarily suggesting PA import the TA wreckage/EMG/cannon idea, but these features made TA's combat much less deterministic than FA while not being random. In FA, units have a hard circle around themselves within which their weapons are virtually 100% effective, and a battle's outcome is extremely rigidly predictable based solely on the kind and count of the combatants, right down to the precise number of casualties expected.

Share This Page