🛈 Announcement
Greetings! The Division forums are now archived and accessible in read-only mode, please go to the new platform to discuss the game
  1. #191
    CategoryTheory's Avatar Senior Member
    Join Date
    Feb 2015
    Location
    Tokyo, Japan
    Posts
    675
    Originally Posted by RichardOshea Go to original post
    ...it's because the math assigns a probabilistic value to an event that may occur and this video describes what did occur.
    The model we use in this forum is probabilistic only and exactly where the game itself is probabilistic. Tests such as yours are good for checking that model against the game's model, but if you're not doing that I'm not sure what value they provide, especially if you can't explain why you're getting the results you do. (It could be an error in the model, but it could equally just be something you've not described in your loadout or your your testing conditions.)

    While the math can certainly describe much it doesn't describe everything so a video shows us and an example is needed for that.
    Everything that happens in the game must be describable in math because that's all the game uses, too. It's calculating stuff similar to how you've seen it done in these very posts, just a lot faster.

    Trading 12% CHD and 5% Damage to Armour for 8% Damage to Out of Cover -- Even losing 6% CHC to get that 12% CHD back-- doesn't persuade me....
    Well, that's not going to persuade anybody who understands the damage model unless you explain things like whether the additional 7.2% expected Wdmg plus the 5% DtA is worth more or less than the 8% DOOC. Even if you measure both against exactly the same enemy that ensures all of those proc, i.e. against armour out of cover, the answer will be different depending on the build. So the result you get doesn't automatically extend to anybody else; it's basically useful for you only, and for that build only.

    As for losing 6% CHC, that's an artifact of your particular build, not a necessity. 10% from the watch and 15% from weapon mods (the most efficient place to add CHC) you need only 35% CHC from your gear (less if you have CHC on your weapon, such as with an SMG). Getting that from five brand buff slots, five attribute slots and three gear mod slots should be no problem at all. You can do it with just six of the eight attribute/gear mod slots if you don't want to use any of the brand buffs, but that might be considered a slightly pathological build.

    So what to do about 70m of armour coming your way(AKA Chunga or Heavy)?
    Well, in that case you know your damage to armour buff will do nothing, if several other experienced players on this forum are right. (I've not checked myself, but I trust them when they say that Chungas do not proc DtA.)
    Share this post

  2. #192
    Originally Posted by CategoryTheory Go to original post
    The model we use in this forum is probabilistic only and exactly where the game itself is probabilistic. Tests such as yours are good for checking that model against the game's model, but if you're not doing that I'm not sure what value they provide, especially if you can't explain why you're getting the results you do. (It could be an error in the model, but it could equally just be something you've not described in your loadout or your your testing conditions.)



    Everything that happens in the game must be describable in math because that's all the game uses, too. It's calculating stuff similar to how you've seen it done in these very posts, just a lot faster.
    Now we know the two agree; That's what I'm seeing anyway. A useful example to check for bugs etc.


    Originally Posted by CategoryTheory Go to original post
    Well, that's not going to persuade anybody who understands the damage model unless you explain things like whether the additional 7.2% expected Wdmg plus the 5% DtA is worth more or less than the 8% DOOC. Even if you measure both against exactly the same enemy that ensures all of those proc, i.e. against armour out of cover, the answer will be different depending on the build. So the result you get doesn't automatically extend to anybody else; it's basically useful for you only, and for that build only.

    As for losing 6% CHC, that's an artifact of your particular build, not a necessity. 10% from the watch and 15% from weapon mods (the most efficient place to add CHC) you need only 35% CHC from your gear (less if you have CHC on your weapon, such as with an SMG). Getting that from five brand buff slots, five attribute slots and three gear mod slots should be no problem at all. You can do it with just six of the eight attribute/gear mod slots if you don't want to use any of the brand buffs, but that might be considered a slightly pathological build.
    Indeed it looks like it comes down to a build by build analysis. I wonder though, as we don't start from a mathematical ideal but almost in our underpants with a tooth pick to charge at the enemy with, if these choices could be any other way, so a build decision has its time and place until you hit that ideal(?).

    Originally Posted by CategoryTheory Go to original post
    Well, in that case you know your damage to armour buff will do nothing, if several other experienced players on this forum are right. (I've not checked myself, but I trust them when they say that Chungas do not proc DtA.)
    I think the point is that a Chunga in a Legendary is much like a dog trotting onto the runway during Dark Hours; a priority target that the team focus on collectively. You can create a one shot build for them I guess but then we're just back to the simple fact that builds are different and do different things. But I'm truly not trying to row with you or prove anything. I'm just looking at the build I have and the gear I have and answering the question you've put. An interesting question, one that for me personally has interesting consequences. Different build different outcome.
    Share this post

  3. #193
    Sircowdog1's Avatar Senior Member
    Join Date
    Mar 2016
    Posts
    3,953
    I'm confused. In those two videos, it appears that the Fox's kill was doing more damage per bullet than the W&H vid.

    Did I miss something?
    Share this post

  4. #194
    CategoryTheory's Avatar Senior Member
    Join Date
    Feb 2015
    Location
    Tokyo, Japan
    Posts
    675
    Originally Posted by RichardOshea Go to original post
    Now we know the two agree; That's what I'm seeing anyway. A useful example to check for bugs etc.
    If you've found that the model here matches up with what you see in game, that's great to hear! And yes, it's certainly useful to have people testing this. I just hadn't seen your calculations to compare with what you saw in the world.

    Indeed it looks like it comes down to a build by build analysis.
    Well, as is so often the case when trying to build an understading and/or intuition for how things work, it often starts with a case-by-case analysis. But from this one can hypothesize general rules and then check those against cases until one can at least use them as rules of thumb and have some sense of where they apply and where they don't.

    In the case here, I'm currently feeling that the general rule is "always try to increase the smallest multiplier," but of course looking at expected value of the multiplier and its buff as based on your estimate of the frequency of that multiplier proc'ing, or the situations in which you want more damage.

    So that leads me to easily imagine a situation in which W&H kneepads do outperform Fox's Prayer: when they would be the second pair and thus giving +5% DtA and when DOOC is significantly higher than DTA. (And of course when you're anticipating seeing enough armoured enemies, or want to prioritize damage on them for whatever reason.) This could happen if you have lots of DOOC on your weapon, for example, and little or no DtA in your build otherwise.

    I wonder though, as we don't start from a mathematical ideal but almost in our underpants with a tooth pick to charge at the enemy with, if these choices could be any other way, so a build decision has its time and place until you hit that ideal(?).
    I don't really understand what you're getting at there. What I see here is basically just an optimization problem with the usual real-world complexities of preferences that are not well defined and having to guess at the values of some variables.

    I think the point is that a Chunga in a Legendary is much like a dog trotting onto the runway during Dark Hours; a priority target that the team focus on collectively.
    Well that sounds like my preference for more DOOC than some would consider optimum because I want to make sure I can kill rushers quick, and I don't mind at all losing a bit of long-term DPS in order to do so. (Lack of long term DPS makes me take longer to get through a mission; lack of short-term DPS on rushers may keep me from getting through the mission at all.) There's no absolute answer to such things unless one can come up with absolute values for one's preferences, and we generally don't know that. (Nor is it binary: there's no amount of armour for most people where they would say, "well, I'm fine with that and need no more, but one armour point less and it would be unacceptable.") So we play around with the figures, along the lines of "let's try a little more here and see how it works out" and never really reach the truly optimum value. But one should still aim for that, rather than aiming lower, because you unlikely ever to go further than you aim.
    Share this post

  5. #195
    Originally Posted by Sircowdog1 Go to original post
    I'm confused. In those two videos, it appears that the Fox's kill was doing more damage per bullet than the W&H vid.

    Did I miss something?
    No I hope not, I mixed the links up the first time but they should be correct now. You can pause your way through the math if you want as it's all there to see. Per bullet tells us the wrong thing though. The target takes x to kill so what we want to know is how many bullets and how fast it dies. RPM tells us TTK -- assuming a hit (and ignoring the variability of criticals and stacking proc).
    Share this post

  6. #196
    Originally Posted by CategoryTheory Go to original post
    I don't really understand what you're getting at there. What I see here is basically just an optimization problem with the usual real-world complexities of preferences that are not well defined and having to guess at the values of some variables.
    And you'd be right to describe it that way but it can't be played that way. We need to be careful to not put the cart before the horse by remembering that from that math we have to create a character, in game, with the exact gear for it to have non-theoretical value. I'd say that its theoretical truth is dependent on one's ability to create it. Even then I think it would be a category error to say that a great mathematical build guaranteed any level of certainty when it came to measuarable performance as you can put a great build into not-so-great hands after all, and that difference in performance is a spectrum. Consider a group match and ask where the teamwork is on that build?



    Originally Posted by CategoryTheory Go to original post
    Well that sounds like my preference for more DOOC than some would consider optimum because I want to make sure I can kill rushers quick, and I don't mind at all losing a bit of long-term DPS in order to do so. (Lack of long term DPS makes me take longer to get through a mission; lack of short-term DPS on rushers may keep me from getting through the mission at all.) There's no absolute answer to such things unless one can come up with absolute values for one's preferences, and we generally don't know that. (Nor is it binary: there's no amount of armour for most people where they would say, "well, I'm fine with that and need no more, but one armour point less and it would be unacceptable.") So we play around with the figures, along the lines of "let's try a little more here and see how it works out" and never really reach the truly optimum value. But one should still aim for that, rather than aiming lower, because you unlikely ever to go further than you aim.
    I agree. I prefer that damage to armour, and that's in some way linked to the fact that I've always used shotguns to deal with rushers; a solution for me was already in place. I'd take us back to the point that these things have to be built and, regardless of what's optimal, that has constraints we can do nothing about.
    Share this post

  7. #197
    The nice thing about rushers, is that their path is more linear than their counterparts - so it's far easier to hit the head [or chest for Outcast] as they approach. Or in the case of Hyena, their bag, which conveniently applies Confuse to everyone near them [except when you accidentally hit the bag when they're next to you].

    I'm now curious about the Out of Cover on Dogs/Chunga.. and for me, the question would be if a Third W&H would be better than Fox's on a non-Rifle build - as my priority point are those fellas. Though it remains to be seen if I can kick my PfE habit

    Played last night for the first time in a long time, and I think I'll be dropping FarCry5 to put it back in rotation [FarCry was fun for the first ten hours, but it's lacking a good enough story or shooting to keep going].
    Share this post

  8. #198
    CategoryTheory's Avatar Senior Member
    Join Date
    Feb 2015
    Location
    Tokyo, Japan
    Posts
    675
    Originally Posted by RichardOshea Go to original post
    RPM tells us TTK -- assuming a hit (and ignoring the variability of criticals and stacking proc).
    No need to ignore criticals; that's exactly what the concept of expected value is for. There will be a lot of variability if you're looking at a single enemy that dies in a few bullets, but over dozens of bullets it won't be far off, and over hundreds it should be within a few percent.


    Originally Posted by RichardOshea Go to original post
    And you'd be right to describe it that way but it can't be played that way. We need to be careful to not put the cart before the horse by remembering that from that math we have to create a character, in game, with the exact gear for it to have non-theoretical value.
    This is quite the opposite way of how I go about it. I don't use theoretical values; I use values from the game. This whole thread started because I had Fox's Prayer in my inventory and on a couple of builds, and I realized perhaps that there were a lot of other things in my kneepads list that, at least on my first crack at analysis, I would never use in preference to Fox's Prayer.

    Essentially, I'm just trying as closely as possible to do exactly what the game does. The game adds and multiplies numbers; I add and multiply numbers; if we both do it in the same way we'll both get the same result.


    Even then I think it would be a category error to say that a great mathematical build guaranteed any level of certainty when it came to measuarable performance as you can put a great build into not-so-great hands after all....
    That doesn't matter in the slightest. We're not comparing things run by two different people, we're comparing two slightly different builds as used by the same person. Given builds A and B, where B does 5% more damage than A but is otherwise substantially similar (particularly in terms of playstyle), you can expect that any place you have a player running A, if you give him B instead in that same place he'll do 5% more damage. If you say that B works badly in situation X, well, compare it with A in situation X, not A in situation Y.

    I prefer [DOOC] to armour, and that's in some way linked to the fact that I've always used shotguns to deal with rushers; a solution for me was already in place. I'd take us back to the point that these things have to be built and, regardless of what's optimal, that has constraints we can do nothing about.
    But you can do (and are doing) something about those constraints. If you want to use a shotgun on rushers for one or more of the many good reasons to do so, you know that you'll have DtA from it, and you can then know that if you want to maximize your damage output you should be making sure you're adding non-DtA multipliers from other sources in your build rather than adding more to the DtA multiplier. (Roughly; there are other considerations of course.)
    Share this post

  9. #199
    Sircowdog1's Avatar Senior Member
    Join Date
    Mar 2016
    Posts
    3,953
    I'm surprised you guys feel it's necessary to make special tactics for rushers. They already move in a predictable path and have less hitpoints than other targets. They're some of the easiest to kill enemies in the game.
     3 people found this helpful
    Share this post

  10. #200
    Oatiecrumble's Avatar Senior Member
    Join Date
    Jan 2016
    Location
    London
    Posts
    4,850
    Originally Posted by Sircowdog1 Go to original post
    I'm surprised you guys feel it's necessary to make special tactics for rushers. They already move in a predictable path and have less hitpoints than other targets. They're some of the easiest to kill enemies in the game.
    Totally agree, they are by far the easiest to kill, i am always more concerned with the enemies behind cover, Controller's, Mech's, Operators, Throwers, RC's etc etc....they are the ones that cause problems.
     1 people found this helpful
    Share this post