1. We have added a Gift Upgrades feature that allows you to gift an account upgrade to another member, just in time for the holiday season. You can see the gift option when going to the Account Upgrades screen, or on any user profile screen.
    Dismiss Notice

What if combat wasn't RNG dependent?

Discussion in 'Civ4 - General Discussions' started by TheMeInTeam, Jul 2, 2012.

  1. Mylene

    Mylene Deity

    Joined:
    Jan 30, 2011
    Messages:
    4,000
    Location:
    Pangea
    Come on, this doesn't make any sense ;)
    "abusing tech leads", okay tell me..what do we play Civ for? Abuse = advancing in strategy games?

    Anyone realizing that you could build loads of warris in an unconnected city, make them suicide against an AI stack, and then win *every* battle afterwards with same strength units?
    Rng may not be perfect...but it's surely awesome compared with such tactics.

    Oh..HA rush? Annoying spears?
    No problems, build some chariots before HBR, they make excellent suicide bombers and your HAs will be glad that they can win afterwards.
    Hammers mean nothing..this would turn even Deity games into 1 big joke ~~
     
  2. jokulmorder

    jokulmorder Chieftain

    Joined:
    Nov 1, 2009
    Messages:
    81
    maybe i'm missing it, but why not just make it deterministic in a way so that the expected average outcome of battles is the actual outcome. on average, your rifleman will not be taken out by a few swords, but we know that he is not likely to survive 20 swords. if you make it so that the riflemen has the expected outcome of each battle, combat is, on average, no different than the current game. you lose as many units as you would in the average game, but there are no longer games in which you lose 20 cavalry against 4 CG muskets.

    now since the expected outcome is guaranteed every time, people would probably build fewer units on average since the risk of not having enough dudes to kill the AI is much less than the risk of going over by a bit and suffering some maintenance. as such, perhaps attacking should be weakened slightly or the AI gets superior combat stats as the difficulty progresses.
     
  3. vranasm

    vranasm Deity

    Joined:
    Feb 2, 2002
    Messages:
    6,437
    Location:
    Czech Rep.
    hmm this one remembers me about the issue in civ V where 10 warriors could kill tank since you got at least 1 damage guaranteed.

    they tried to fix this in G&K with increasing hitpoints to 100

    @L
    thanks...tough to remember for me, especially this one, since the I use the phrase a lot in czech.

    hmhm when (or while??) I think about it...everyone here from now on learns Czech and posts some czech sentences here and there so I can get some laugh too as revenge :)
     
  4. traius

    traius His own worst enemy

    Joined:
    May 27, 2012
    Messages:
    938
    These are the problems I see
    1. Mostly useless city defender bonuses. City defender is a strong ability that adds tons of power to units with a first strike (Long Bows, archers, etc). I am in the middle of a game where a protective churchill had a CD3 DR 2 Longbow in his city, and I simply lost unit after unit before he even got injured. I think it took 3 mace and 4 elephants to finally kill him, and frankly, I've seen worse for similar units. This would be a lot less in your system.
    Thought about this more, and since he had 2-3 first strikes and 16.5 STR, he would kill quite a few even in your system. The first strikes should have him up to 3 unpromoted macemen/ at least. Nothing on 7, but not terrible.

    2. First strikes are half useless, I think. They should probably be made to happen like normal, before combat starts, to put a SLIGHTLY random feature into your game, so it is not 100% predictable. If you don't like that, then you must consider what first strikes and, more importantly, first strike "chances" do in this game. Do they save some health for the survivor? Do they occur before combat, as before? Or do they give a deterministic expected value of damage before the combat begins (first strike chance of hitting is (FirstStriker-er STR) / (Not-FS-er STR + FS-er STR). This then does [(FS-er STR) / (Not-FS'er STR)] * 20HP damage, if I am not mistaken. You would simply multiple these all through to get expected value for the strike (EV = probability * magnitude of event/data for simple things like this, EV = chance to hit * damage, as defined above). First strike chances would be the same, but also have to multiply through the chance of them happening (I don't know the percentages for first strike 'chances' off hand. 50%? If so, a first strike chance = 0.5 * EV for first strike). I would recommend the first the 'leaving it as normal' or the final 'completely deterministic' methods, myself.

    3. I assume damage done would be 100HP * (Dying unit STR / Surviving Unit STR), which would also result in MAD for Axe v Axe or similar even battles, which would have combat units survive. I just remember seeing an out of context subtract comment somewhere, didn't want to have a unpromoted elephant killing a combat maceman thanks to an oversight using subtraction.

    4. Make sure not to muck with city raider, or it will get unbelievably powerful. It should reduce the opponent exactly as it does now, and save the attacker health the same away. Don't let the modder accidentally make city raider do what it reads like it is supposed to do!

    5. Cats are stronger than ever in your system. not only they do guaranteed collateral damage, they also significantly weaken the top defender EVERYTIME.
    Although one could argue that there are loads of scenarios where you would have kept 'sieging' in the past where now you wouldn't bother because you already have nothing but guaranteed wins for your CR's.

    6. The main problem I see is the stack on stack conflict that has 0% of doing much of anything for a tech weak defender. If you are outnumbered and out-teched in the old game, you are in a terrible position (2 players only and it is over if only due to pillaging, but anything else not necessarily because there could be bribes, backstabs, new villains to distract the aggressor, etc). In the past he could hole up in his city and hope to build up and bring in allies. Not even a city is likely to do anything for him now. I guess what I worry about most is the defender who can get a top defender with a ~66% chance to win and everything else ~40% who only manages to get one kill with his top defender and everything else in the stack just dies.

    As a random aside:
    Custer may have been 'out-teched' by Sitting Bull and his superior repeating rifles. Look it up on wiki. It's really not the same as massed rifles and spear wielding warriors thing. And Stalingrad was retaken by Russian tanks. REtaken. That could still totally happen in this set up (?)
     
  5. rah

    rah Deity Supporter

    Joined:
    Sep 12, 2005
    Messages:
    8,935
    Location:
    Chicago
    While I'm starting to buy into the concept, I still have a problem with equal fights means everyone dies. Someone always walks off the field. OR should anyway.

    Or that you always win a 51 to 49. Yes I agree that the survivors in a close contest would be near death and easy to pick off. But when fighting the AI, once you've taken out their stack(s) those remaining units in cities would just wait there for their eventual demise while just a few fresh units could protect your stack until it healed.

    Maybe a fuzz factor would resolve this.

    Interested to see how people respond to further testing.
     
  6. TheMeInTeam

    TheMeInTeam Top Logic

    Joined:
    Jan 26, 2008
    Messages:
    25,178
    If by "such tactics" you mean something as flagrantly terrible as donating dozens of war success to the AI and eating WW to match then sure, you might prefer the RNG to executing explicitly poor strategy.

    The problem with fodder units is that they're guaranteed losses, and those losses only translate to future wins if you have enough total units. What's to stop them from using some fodder...or in the AI's case simply building units at fodder unit costs due to bonuses? Nothing.

    Doubtful. But god forbid we implicate some unit attack tactical importance into the game.

    Warriors were not the problem. It was extremely improbable to be able to hit the same unit with 10 other units in 1UPT rules. The issue in civ V was that ranged units like archers were guaranteed 1 damage, and could gain the ability to fire multiple times...making an experienced archer threatening to literally any unit in the game (except nukes).

    No such luck in civ IV. If you fodder off 10 warriors, you lose 10 warriors's worth of :hammers:. That's a lot of hammers to give up to take one city/kill one set of units. Of course, assuming they do enough damage to reliably clean the opposing stack.

    Ever notice that all of these historical "upset" battles being used here involve small #'s of more advanced troops losing to superior #'s of enemy forces in terrain/situations that favor the "underdog"? Now recall that such is exactly what happens in civ IV, even with deterministic combat, cut the trash arguments out of the thread, and focus on gameplay :).

    The only promotion I modified was drill, making it a bit more linear in promo quality with the same awesomeness on drill IV.

    If you are so badly out-teched and out :hammers:'d that you can't deal with enemy numbers with collateral initiative in your own territory and a shorter reinforce line, you deserve to die.

    I've noticed some odd behavior where some very close fights result in withdraws when the unit has no chance of withdrawal (both units get redlined). I'll have to see what's going on with that, and also decide if it's bad or not.
     
  7. Mylene

    Mylene Deity

    Joined:
    Jan 30, 2011
    Messages:
    4,000
    Location:
    Pangea
    God forbid we accept that units with 100% health always winning vs. those with 95% is nonsense ;)
     
  8. MeatUnit2

    MeatUnit2 Warlord

    Joined:
    Jan 12, 2010
    Messages:
    174
    TMIT: Have you played the board game Diplomacy? All combat is resolved without dice rolls or other elements of chance.
     
  9. civac

    civac Warlord

    Joined:
    Oct 22, 2010
    Messages:
    108
    You did not answer this idea, tmit. Seems much less invasive than changing the battle resolution completely.

    If you want to stick with the subtraction system can you qualify how it differs from the vanilla system?
     
  10. Ramesses-Rules

    Ramesses-Rules King

    Joined:
    Nov 27, 2008
    Messages:
    837
    Location:
    Sequim, WA
    I agree with all this. The TMIT change wouldn't make much difference for me personally, but it would make the game less frustrating for players who typically go for domination victory. That's reason enough to go ahead and do the change. I believe it would increase "average enjoyment" of the game for the collectivity of Civ-4 players. :)
     
  11. TheMeInTeam

    TheMeInTeam Top Logic

    Joined:
    Jan 26, 2008
    Messages:
    25,178
    Why don't we hear some legit justification for this then? Players have control over whether they're putting damaged units at risk.

    You mean in terms of damage inflicted? I did mention earlier in the thread that possibly damage dealt is too much right now (IE too much guaranteed damage).

    Now, do you have a workable formula for all strength values that gives us the answer to "average expected damage" based on previous odds values? If you have one, I would CERTAINLY forward it to my friend and it would be well worth consideration. Maybe there's something I'm missing and a nice algebraic formula will do nicely for that and make #units needed on average to kill another unit close to what the rng-based game uses.

    We've done the change, but testing it is in its infancy. We nearly started a game where the defender got no defensive bonuses ever before realizing it and fixing it :eek:.
     
  12. plasmacannon

    plasmacannon Emperor

    Joined:
    May 5, 2010
    Messages:
    1,651
    Gender:
    Male
    Location:
    Orlando, Florida
    Hopefully, the units will be made cheaper in pp and maintanaince cost. So we can really get a large stack, without tanking our economy early too.
     
  13. Mylene

    Mylene Deity

    Joined:
    Jan 30, 2011
    Messages:
    4,000
    Location:
    Pangea
    Okay TmiT, one last try ~~
    ..you have 20 Curis (currently). Psycho Alex sits with his stack of 60+ units (Knights, Jumbos, Maces, Siege, x-bows, Pikes etc.) nearby.

    DoW, his big stack moves towards one of your cities. We knew that, our Curis are waiting and roads are plenty. Pikes cannot get winning odds on open terrain..

    So..20 easy wins, nopes 30+ cos all siege dies too.
    Very elegant how my Curis move back into safety after hitting.
    Next turn...baffled Alex moves his stack 1 turn further (yay for cultural terrain).
    Our Curis get another promotion (and healed), and i can most likely pick..should i destroy his helpless Maces & Co now, or heal 1 turn with our brand new super medic that arrived just in time (wining all battles greatly helped).

    Final result: 0 losses, sweet war success.
    Strategies used: 0.001.
    Wishes RNG would be back: 99.99%

    Now lets upgrade Alex..he has Curis too instead of Knights, 15.
    Highest base strength, so they'd defend first..Alex was "smart" enuf to give 6 of them str. 2 instead of flanking. We use 6 silly units (or Cats) first..then repeat our 20 wins.
    zzzZZZzzzz
     
  14. Um the Muse

    Um the Muse King

    Joined:
    Sep 11, 2009
    Messages:
    922
    Location:
    surrounded in the USA
    Mylene does have a bit of a point: you are getting "free" heals thanks to promotions. It might make defenders stronger, too, at least in theory, because they can defend and heal on the same turn as long as they survive.

    As for the formula. You probably are doing a little too much damage compared to the unmodded game, but maybe not as bad as you think.

    If a battle is at all close to equal (but not exactly equal), the winning unit will need four wins (best out of seven). So, you want the odds that the loser would be able to inflict one, two, or three rounds' worth of damage before dying and then average them. To find out what this is mathematically, I believe you'd use a Poisson formula, but let's just split the difference and call it 1.5 (it "should" be less than 2 because the odds of winning are less than 50/50. 1.5 is just easy to deal with).

    Finally, here's the formula per round (copied from the combat explained article): floor(20*(3*A+D)/(3*D+A)). If you're using my 1.5 rule of thumb, change the 20 to 30 and that should be close to what the base game gives you.

    One "problem" is that if two units are exactly equal in strength, then neither will die. A more serious concern is that highly uneven combat will result in more damage to the victor. If you're willing to let the computer do the computation, you can fix this, but I don't know if it's worth the effort. You might want to just tweak relative strengths of units instead.
     
  15. TheMeInTeam

    TheMeInTeam Top Logic

    Joined:
    Jan 26, 2008
    Messages:
    25,178
    It would help if you followed this by actually trying. In the RNG world you get to duplicate that outcome at the cost of a few siege units extra, but with much less healing time after doing so. You can't pretend for a second that you use significantly more strategy with your tech lead when the RNG is involved vs when it isn't; I'm not letting such nonsense slide.

    Moving into opposing territory vs a tech superior force with enough enemy #'s is flagrant stupidity. You seem to have a PROBLEM with flagrant stupidity resulting in bad losses. Why is that?

    Also, best defender code is still in play so cuirassers would not necessarily defend first.

    At any rate the sum of your argument is "the AI is freaking stupid and throws away units, therefore this game needs an RNG so that human players get screwed over by idiot strategies sometimes"...even though at THAT point in the game, your cuirassers are going to win regardless. Hell, in your SECOND example (alex has cuirassers) you might actually take FEWER losses with the RNG. And what will you lead with in both cases? Siege. What will you attack with after doing so in both cases? Cuirassers. What difference is there in your "strategy"? NONE.

    However if Alex got on defensive terrain or made it to your city...or if you pressed into HIS territory and got hit by that stack you'd be in real trouble (dead for certain, actually). Indeed, with the amount of damage you consistently take it might still be a bit challenging to press into his territory and take a lot of cities...because he has the bonuses after all.

    In fact, I assert that your overall "strategy" in war won't change one bit vs the AI, except in the very early game rushes where your "strategy" is to "build as many units to rush as possible and then hope for luck". At tech parity you will find it pretty hard to defend early game unless you can come up with decent #'s on defense. If they get a tech lead on you early on then declare? Baaad stuff. You actually have to think more, not less. Citing a tech lead example where your best unit is better than anything they have is a joke Mylene. That's not a rigorous test of the system at all. It's like me saying how broken the RNG-based system is by pointing out that my CR II and III artillery can mow down the AI with few losses when they have rifles. To quote you: "zzzzzzzZZZZzzzzzzzz". Isn't this a fun way to approach the argument?

    Now what about vs competent players? Still no strategy?

    That...exists in both RNG and deterministic settings and is moot. Mylene's issue is with the "no losses", which only happens when one side has rocks (or silicon) for brains. Basically she's complaining that doing something completely and utterly idiotic is punished slightly more with deterministic combat than otherwise, and therefore the game "needs" an RNG that can have ruinous consequences in other situations.

    So...1.5 rounds? A 1 str unit attacking a 10 str unit would do ~11.8 hp of damage per round using (30*13/33)...for 1.5 rounds this would do 17.7 hp of damage...actually MORE damage than the current model? IMO that's the wrong direction, since compared to the base game things like hill longbows are already gimped to some extent due to the guarantee that they take some damage (even if their first strikes makes them take a fraction of the damage other units take, they still take some).
     
  16. civac

    civac Warlord

    Joined:
    Oct 22, 2010
    Messages:
    108
    The free healing from promotions is actually very problematic.


    There is not a single way to do it. You have choices.

    The problem is to take the possibility of losing an advantageous fight in the RNG based case and incorporate it into the damage system in the deterministic case.

    One way would be to simply assume the stronger unit (the favourite) wins the fight in the RNG based case and calculate expected damage under that condition. This would lower the average damage taken by the favourite compared to RNG case (well, sort of).

    If you do not want this we need to look at the RNG outcomes in which the favourite loses, too. For those outcomes we have to assign a damage value to the favourite. The most natural one would be to assign its hp before the fight (since it loses all of it).

    However, both of the above possibilities are not consistent with the case of 2 equally strong units annihilating each other. Define one of the units as the favourite arbitrarily. Its received damage by the both formulas will be obviously lower than it's initial hp. In practice this would lead to a jump between a 5 to 5 a and a 5 to 5.01 fight. The fewer rounds of combat there are the larger the jump will be. In the extreme case of single combat round the 5.01 unit will win with half of its hp left following the second formula and completely unscathed following the first.

    Getting rid of the incosistency is surprisingly quite tricky. The problem lies in the fact that in a battle damage value of attacker * number of rounds to kill <> the hp of the unit being attacked.

    I shall have to think about this some more.
     
  17. Mylene

    Mylene Deity

    Joined:
    Jan 30, 2011
    Messages:
    4,000
    Location:
    Pangea
    TmiT :)
    I dun care about mp..and i guess most others here don't, there are way better games to play vs. humans. But there are hardly any better games to play vs. AIs ;)

    So yip of course..i care about AI stupidity not being too rewarding.
    Random stuffs happening keeps this game fun, well for me on Deity at least.

    My example shows 2 things: Horse units get even better, cos now healing means more and they can move away from danger.
    And tech advantages make you invincible on defense, and all attacks *can* start with luring stacks out.
    Iam only countering your argument from earlier here, that tech leads matter less cos all units can do damage. Really, it's the opposite.
     
  18. TheMeInTeam

    TheMeInTeam Top Logic

    Joined:
    Jan 26, 2008
    Messages:
    25,178
    I'm thinking that I agree, and might pull the ability to heal with promotions.

    Tech leads definitely matter less on the offensive, and possibly more on the defensive. Keep in mind, however, that you're not going to have a few advanced units shred dozens of weaker units. A BIG part of your cuirasser example was the fact that you could retreat them; any non-mounted unit (or attacking against forest or hills, which would cost your movement point and ability to retreat) would almost certainly cost you cuirassers...if not when attacking the stack directly (defensive bonuses) then on a retaliation attack (you can't move away).

    However, without that "mounted can attack and run away without retaliation due to tech lead" you really do have to be careful about trading an advanced unit for a less advanced one due to exposing your damaged troop. Cuirassers and Cavalry are the only 2 units in this game that get a decent window to do the tactic you describe...but defensive wars don't win games. You still have to attack people, and that still means building a LOT of troops to brave the AI's territory and take guaranteed damage on every attack. I'm not convinced deity would actually be easy for you because of this.

    Think of it this way: when you lose deity, is it usually because a) you get wiped off the map or b) you get stuck in a position where you can't keep up with AI bonuses and war either isn't viable or fails? I think most players who lose deity lose because an AI gets too far ahead of them when they don't have enough land to catch it.

    I'm definitely interested. Balance can warp bigtime based on expected damage, although possibly we WANT high expected damage. I'm also thinking that experience might merit being reworked because you know if you'll win or lose and the current model offsets a bad-damage disincentive with more xp gained in battles where you lose the most health.
     
  19. plasmacannon

    plasmacannon Emperor

    Joined:
    May 5, 2010
    Messages:
    1,651
    Gender:
    Male
    Location:
    Orlando, Florida
    I play multiplayer, but with a friend of mine and I (unteamed) vs the rest of the AIs.
    So, I'm in the group who do look forward to new options for this game.

    Keep us posted TMIT.
     
  20. Nick Carpathia

    Nick Carpathia Unleash the HAARP

    Joined:
    Aug 17, 2010
    Messages:
    1,064
    Location:
    Romania
    Post a link to the deterministic combat mod, I think the Realms Beyonders might be interested (in breaking it :p)
     

Share This Page