test content
What is the Arc Client?
Install Arc

Class Balance

1246

Comments

  • micky1p00micky1p00 Member, NW M9 Playtest Posts: 3,391 Arc User
    edited February 2020

    micky1p00 said:

    .....

    Or please reply the question that was given.
    Is in your opinion, in PvE, everyone are equally skilled and the only factor is gear? Or "Veterency"?

    Given everyone exactly the same gear, same stats, same power, same class, same time, will everyone have the same result? In terms of damage, and survivability.



    BtW that is not how you "make AI"
    https://openai.com/blog/more-on-dota-2/

    You do not program them with specific moves. This is a "learning machine" you just make them play.
    Isn't that the same argument that you have been making for years? It isn't the gear, or the game design, but instead a "learn to play" problem? If everyone would "learn to play," they would have the same result that you do.
    If you would ever actually bothered to read what I write...... Instead of the usual self righteous vitriolic anger..

    For your convenience here some quotes from a quick search:



    Here one addressed in a post to you btw:



    Others:



    And I hope people will not "learn to play" to my level because then we are doomed:



    BtW we finished the trial as the group we were since that post.

    And while I have many complaints about the game design, which you are free to look for. I will always argue against people who when stuck at some issue, immediately give up and go whine on the forums blaming players who did succeed the same task. Instead of trying to learn or use the plethora of information available to them to improve themselves.
    External blame as first recourse is an interesting thing.

    Yes, many of those that complain can just learn some basic mechanics and they will be better than me, having upfront better reaction time and better awareness. They just feel they shouldn't do it. It took us half year to get ToMM done as a group what others did in a week. Did you see me post about "evil skilled players"?
    Post edited by micky1p00 on
    Guild: The Imaginary Friends
    If you disagree with me: My opinions are my own and do not reflect on my guild and friends.
    If you agree with me: My opinions are totally the guild official policy !
    ;)
  • micky1p00micky1p00 Member, NW M9 Playtest Posts: 3,391 Arc User
    edited February 2020


    Third Yes, skill matters in any action and no 2 people are the same as summer is more hot than winter and the water is wet.
    Also in D&D you do not know the encounter in advance, nor do you know the stats of the enemy.
    Running static dungeons with a cheat sheet reduces skill to reaction time and reaction time is 80% physiology and 20% training.
    So you should really militate to ban Act log and then talk about some little skill involved.

    I always wonder why you keep comparing this game to D&D. You are not new player, nor oblivious. You know very well that the transition to F2P MMO "took its toll" and the game only resembles the PnP in names and not much more.
    With those static dungeons.

    Yes, reaction time (+ timing ) and awareness, there are still so many players who will not move to get CA, or dodge instead of move and so many other things that indeed can be very narrowed down. After the build and 'preparation' is done.

    There is a difference, and it shows. Not surprisingly, for example, there is a player who took a class and in 2 weeks did better than I did in years.

    I'm not sure what you mean by
    "So you should really militate to ban Act log and then talk about some little skill involved."
    I don't follow how ACT is involved, ACT doesn't add or remove from those reaction/timing/awareness skills or you think that because I'm saying that there is a skill difference I imply than I'm better than anyone? If so read my quote a post above. Especially where reaction time and timings are concerned.
    Guild: The Imaginary Friends
    If you disagree with me: My opinions are my own and do not reflect on my guild and friends.
    If you agree with me: My opinions are totally the guild official policy !
    ;)
  • tchefi#6735 tchefi Member Posts: 246 Arc User
    edited February 2020


    As Janne said, that is not how the type of AI I was referring to is programmed. Nobody programs a chess bot to do x in the case of y, its not efficient to do so. In the case of Dota specifically, I mentioned that it isn't there quite yet, but it may be in this year or the next and for the purposes of 99.95% of all players, they would be better off playing against open AI than playing against another person if they wanted to learn.

    Stockfish (the last version) is still very much sort of a "bruteforce", calculating tree of moves in each position during a chess game (+ inspecting a database of known chessgames) to a quite deep rank (the current position can lead to the examination of 90M/sec of possible subsequent positions at various level of depth ), evaluating each one with a ponderation (a -1 position for black is a position where he has lost a pawn compared to white, or the position is so bad that even with equal material it is like black has 1 pawn less) with a min/max idea + some heuristic to make the definitive move.


    @admiralwarlord#3792

    World chess champion Magnus Carlsen is around ELO ~2850, Stockfish is evaluated at more than ELO 3200 i think. It's the same "strength" difference as between the highest grand masters, and someone who are not even in the international master club.

    Neural AI Alphazero and Leela chess are relying on both self teaching and learning from the games they play. They are evaluating far less positions from a given board (less than 0.1% of how many Stockfish is evaluating), but have some kind of "intuition" (and in fact their moves are very much more "human" than stockfish). They kind of try some random combinations of subsequent positions (based on the previous ones they had learned), and improve on those that seem appropriate to determine the move they will make (based on a tweeked Monte-Carlo algo).
    Alphazero was just "taught" the rules of chess, then it played against itself during 9h (44M of games xD), and after that it was able to literally obliterate any human and any existing engine (Stockfish included).
    Alphago (who won against some of the highest ranked Go master) is more kind of evaluating "hot spots" on the board from a very large strategical point of view/level as if he was doing your country weather and temperature report, then play at or around nodes that are hot and avoid putting stones where it's cold (unless it's forced by the opponent move).


    The nowdays equivalent of AlphaGo/AlphaZero in pro-gamer scenes like Dota2 or SC3 is basically working as its chess and go version. Past one point, it will be better and quicker at decision making (which is I think the most important for those kind of games, especially as you can't count on your reflex to beat an AI in a fight :P) with far less mistakes (not that it will be infallible though) than a human. I think on SC3 it was even trying very cheese strats catching some of the strongest human players completely off guard.

    That's kind of fascinating ^^.
  • thefabricantthefabricant Member, NW M9 Playtest Posts: 5,247 Arc User
    edited February 2020


    As Janne said, that is not how the type of AI I was referring to is programmed. Nobody programs a chess bot to do x in the case of y, its not efficient to do so. In the case of Dota specifically, I mentioned that it isn't there quite yet, but it may be in this year or the next and for the purposes of 99.95% of all players, they would be better off playing against open AI than playing against another person if they wanted to learn.

    Stockfish (the last version) is still very much sort of a "bruteforce", calculating tree of moves in each position during a chess game (+ inspecting a database of known chessgames) to a quite deep rank (the current position can lead to the examination of 90M/sec of possible subsequent positions at various level of depth ), evaluating each one with a ponderation (a -1 position for black is a position where he has lost a pawn compared to white, or the position is so bad that even with equal material it is like black has 1 pawn less) with a min/max idea + some heuristic to make the definitive move.


    @admiralwarlord#3792

    World chess champion Magnus Carlsen is around ELO ~2850, Stockfish is evaluated at more than ELO 3200 i think. It's the same "strength" difference as between the highest grand masters, and someone who are not even in the international master club.

    Neural AI Alphazero and Leela chess are relying on both self teaching and learning from the games they play. They are evaluating far less positions from a given board (less than 0.1% of how many Stockfish is evaluating), but have some kind of "intuition" (and in fact their moves are very much more "human" than stockfish). They kind of try some random combinations of subsequent positions (based on the previous ones they had learned), and improve on those that seem appropriate to determine the move they will make (based on a tweeked Monte-Carlo algo).
    Alphazero was just "taught" the rules of chess, then it played against itself during 9h (44M of games xD), and after that it was able to literally obliterate any human and any existing engine (Stockfish included).
    Alphago (who won against some of the highest ranked Go master) is more kind of evaluating "hot spots" on the board from a very large strategical point of view/level as if he was doing your country weather and temperature report, then play at or around nodes that are hot and avoid putting stones where it's cold (unless it's forced by the opponent move).


    The nowdays equivalent of AlphaGo/AlphaZero in pro-gamer scenes like Dota2 or SC3 is basically working as its chess and go version. Past one point, it will be better and quicker at decision making (which is I think the most important for those kind of games, especially as you can't count on your reflex to beat an AI in a fight :P) with far less mistakes (not that it will be infallible though). I think on SC3 it was even trying very cheese strats catching some of the strongest human players completely off guard.

    That's kind of fascinating ^^.
    @tchefi#6735 google's "Alpha Zero" beat Stockfish quite a while ago and it is using a similar technique to Open AI, it is not a brute force solution, as you mentioned. So whilst Stockfish is still a brute force solution, it isn't relevant to what I was saying.
  • tchefi#6735 tchefi Member Posts: 246 Arc User
    edited February 2020
    Damn @thefabricant, got you :P you didn't read my entire post :P

    (edit : by the way i watched AlphaGo vs Lee Sedol live ^^ in 2016 I think ? )
  • thefabricantthefabricant Member, NW M9 Playtest Posts: 5,247 Arc User

    Damn @thefabricant, got you :P you didn't read my entire post :P

    (edit : by the way i watched AlphaGo vs Lee Sedol live ^^ in 2016 I think ? )

    I did, after I responded. :tongue: I thought the 2nd part was addressed to someone else, so I responded to the first paragraph before I read the rest. I also watched it, it was interesting to watch.
  • micky1p00micky1p00 Member, NW M9 Playtest Posts: 3,391 Arc User
    edited February 2020

    chess stuff

    Interestingly, it's not possible to create a full lookup tree into or even most possible moves (with full depth). So even the classic programs used prediction to certain depth, and various decision making mechanism but never a full lookup for the positions.
    Guild: The Imaginary Friends
    If you disagree with me: My opinions are my own and do not reflect on my guild and friends.
    If you agree with me: My opinions are totally the guild official policy !
    ;)
  • tchefi#6735 tchefi Member Posts: 246 Arc User
    edited February 2020
    micky1p00 said:

    chess stuff

    Interestingly, it's not possible to create a full lookup tree into or even most possible moves (with full depth). So even the classic programs used prediction to certain depth, and various decision making mechanism but never a full lookup for the positions.
    Kind of true (though I do think one day an AI or engine will eventually say : after the first move, white win/black win/draw, it's just a matter of power of calculation and time, but at the end it's not so different from tic-tac-toe game. You just have to remind yourself that when you hold your smartphone in your hand, you have more calculating power in it than what was loaded in the appolo modules who landed on the moon).
    That's why the "bruteforce" engine sort is not very prone to sacrifice a piece to gain a stronger position, unless they can find a checkmate after few forced moves, because loosing material usually makes their evaluation negative for the variant in the position. The reason is they "cut" branches of the tree they have evaluated as very bad (alpha-beta selection) before examining it too deeply in order to save some calculation time (that's why stockfish has some heuristic principle coded, to help him, among other things, look at sacrificing pieces ).
    Neural AIs don't have this biais though :P



    (edit : don't talk about chess or go please, i can't help but dive into those conversations and will make the original topic drift very far from what it is meant to talk about :D )

    (re-edit : for a "more complex" perspective than tic-tac-toe game => after googling about draughts/checkers it seems this game had been solved by a hord of computers doing the bruteforce calculations during 20 years. The checkers game ends in a draw if each of the 2 players play only perfect moves. That was the answer 42 for the draughts :P).
    Post edited by tchefi#6735 on
  • micky1p00micky1p00 Member, NW M9 Playtest Posts: 3,391 Arc User
    edited February 2020

    A woman must be on average about 20 years younger than a man to have faster reaction time.
    No outliners. No injury or illness factoring.
    Is this what are we trying to prove because there are plenty of studies that done it already 50 years ago.
    .

    Well, one day I will learn to make short posts, not any time soon unfortunately.

    Luckily for all of us not in 20 and so on, the twitch reaction in NW is adjusted to accommodate us too. So what left is learning the mechanics of the content (which is not reaction skill) and basic game mechanics (which is also non reaction, though reaction will be a limiting factor in execution).

    So even 60 years old, old farts can do ToMM with enough bullheaded stubbornness and the will to do so.
    Guild: The Imaginary Friends
    If you disagree with me: My opinions are my own and do not reflect on my guild and friends.
    If you agree with me: My opinions are totally the guild official policy !
    ;)
  • krumple01krumple01 Member Posts: 755 Arc User
    micky1p00 said:

    A woman must be on average about 20 years younger than a man to have faster reaction time.
    No outliners. No injury or illness factoring.
    Is this what are we trying to prove because there are plenty of studies that done it already 50 years ago.
    .

    Well, one day I will learn to make short posts, not any time soon unfortunately.

    Luckily for all of us not in 20 and so on, the twitch reaction in NW is adjusted to accommodate us too. So what left is learning the mechanics of the content (which is not reaction skill) and basic game mechanics (which is also non reaction, though reaction will be a limiting factor in execution).

    So even 60 years old, old farts can do ToMM with enough bullheaded stubbornness and the will to do so.
    The reason older people have slower reaction time is because they know their bills are paid.

  • finmakinfinmakin Member Posts: 378 Arc User
    edited February 2020
    micky1p00 said:

    A woman must be on average about 20 years younger than a man to have faster reaction time.
    No outliners. No injury or illness factoring.
    Is this what are we trying to prove because there are plenty of studies that done it already 50 years ago.
    .

    Well, one day I will learn to make short posts, not any time soon unfortunately.

    Luckily for all of us not in 20 and so on, the twitch reaction in NW is adjusted to accommodate us too. So what left is learning the mechanics of the content (which is not reaction skill) and basic game mechanics (which is also non reaction, though reaction will be a limiting factor in execution).

    So even 60 years old, old farts can do ToMM with enough bullheaded stubbornness and the will to do so.
    What, WHAT....
    Are you calling me a bullheaded (in my case) 60+ year stubborn old HAMSTER???
    How rude
    :)
    Ogguk Sanctus… Justicar Paladin
    Puny Tank

    Guildleader of NWLC Alpha
  • rjc9000rjc9000 Member, NW M9 Playtest Posts: 2,310 Arc User

    TThis reminds me of a power that gave you a buff when you were shooting over a certain distance (cough, cough). By the way they could easily use Aimed Shot for your purposes and that would be the easiest way to improve the Archer Hunter.

    - increase the damage of Aimed Shot and use it as a trigger for the cooldown reduction feat currently based on Commanding Shot
    - bring Binding Arrow to the Hunter path (which also makes sense as all feat support for roots is on the Hunter side)
    - regive a damage buff to Longstrider's Shot when used over a certain distance.

    Tactical Power is the offical-ish terminology I remember that the game used for things like Guard, or Teleport (Wizard), or Sprint.

    The way I envsioned the power was that after pressing it, you would enter the Aimed Shot kneeling animation and gain the following effects:
    • -3 Seconds base cooldown on ranged attacks
    • +Damage Buff on Ranged attacks equal to your DEX score (so 25 DEX = 25% damage buff while in this tactical power)
    • Can't move until you re-press the button.
    Slotting this tactical power in place of the short dash would lock out your ability to swap weapons, but give you an ranged additional encounter power slot.

    I suppose I'd call it "Steady Aim" (I could have sworn I saw a class feature/manuever in the 5e Player Handbook with the name "Steady Aim") and the reasoning behind this power is so players who want to focus on pure ranged damage can denote "hey I'm playing Archery style" with this tactical power.

    I am not sure if the game can handle players having multiple tactical powers, or being able to slot in some tactical powers over others, which is why I was asking if it was within the scope of the game engine.

    I also would not be opposed to bringing back the buff of Longstrider's Shot as well, though hopefully its (relatively) high buff amount and short duration (idk, 30% for 5 seconds?) would be kept so it puts the pressure on the player to choose their next attacks wisely.

  • thefabricantthefabricant Member, NW M9 Playtest Posts: 5,247 Arc User
    Alternative thread title: Speedrunning derailing a thread.

    I am not sure how we got from class balance to this :neutral:
  • oldtimer#7525 oldtimer Member Posts: 141 Arc User
    I like turtles.
  • aerhythia#3255 aerhythia Member Posts: 173 Arc User



    Sure op healer is better on first runs until you learn the dangeon but you can do it with every healer very easy.

    If someone is low on HP and in danger of dying OP healer certainly isn't the best healer. Unlike DPS (where only having good AoE / bad single target DPS or vice versa isn't useful DPS in general) there's a good balance.

    OP: Preemptive healing (shields)
    Cleric: In general best heals, requires a lot of praying in combat to build divinity
    Warlock: Good heals, cleanse and buff
  • wintersmokewintersmoke Member Posts: 1,626 Arc User
    .
    micky1p00 said:

    micky1p00 said:

    .....

    Or please reply the question that was given.
    Is in your opinion, in PvE, everyone are equally skilled and the only factor is gear? Or "Veterency"?

    Given everyone exactly the same gear, same stats, same power, same class, same time, will everyone have the same result? In terms of damage, and survivability.



    BtW that is not how you "make AI"
    https://openai.com/blog/more-on-dota-2/

    You do not program them with specific moves. This is a "learning machine" you just make them play.
    Isn't that the same argument that you have been making for years? It isn't the gear, or the game design, but instead a "learn to play" problem? If everyone would "learn to play," they would have the same result that you do.
    If you would ever actually bothered to read what I write...... Instead of the usual self righteous vitriolic anger..

    For your convenience here some quotes from a quick search:



    Here one addressed in a post to you btw:



    Others:



    And I hope people will not "learn to play" to my level because then we are doomed:



    BtW we finished the trial as the group we were since that post.

    And while I have many complaints about the game design, which you are free to look for. I will always argue against people who when stuck at some issue, immediately give up and go whine on the forums blaming players who did succeed the same task. Instead of trying to learn or use the plethora of information available to them to improve themselves.
    External blame as first recourse is an interesting thing.

    Yes, many of those that complain can just learn some basic mechanics and they will be better than me, having upfront better reaction time and better awareness. They just feel they shouldn't do it. It took us half year to get ToMM done as a group what others did in a week. Did you see me post about "evil skilled players"?
    Irrelevant armchair psychiatric analysis, being irrelevant, will also be ignored...

    Hello, Bell Curve, my old friend. "Lies, damned lies, and statistics." I have often wondered what exactly some mathematician did to Mark Twain that caused that witticism to come into being. Most people are aware of the two biggest flaws of statistics... that it is so easily manipulated into giving misleading results, and that the numbers become less accurate the smaller the sample being represented. What many fail to realize, until after they begin to study statistics, is that the entire point of the exercise is to evaluate quantitative data. Qualitative data, being unable to be adequately expressed with numbers, have no place under the bell curve. Skill carries no weight. It displaces no volume. Has neither length, nor depth, nor breadth. And those who would be tempted to express skill as a function of time, will quickly be disappointed. Especially in MMO's with a cash shop. The problem with variables which have no numerical expression, is that they have no place in the decision making process when devising new content. Or the correction of old content. The Devs have access to exponentially more data than we ever will, despite Rainier's best efforts. And, I strongly suspect even they would not be able to come up with an algorithm to successfully measure "skill."
  • micky1p00micky1p00 Member, NW M9 Playtest Posts: 3,391 Arc User
    edited February 2020

    .

    micky1p00 said:

    micky1p00 said:

    .....

    Or please reply the question that was given.
    Is in your opinion, in PvE, everyone are equally skilled and the only factor is gear? Or "Veterency"?

    Given everyone exactly the same gear, same stats, same power, same class, same time, will everyone have the same result? In terms of damage, and survivability.



    BtW that is not how you "make AI"
    https://openai.com/blog/more-on-dota-2/

    You do not program them with specific moves. This is a "learning machine" you just make them play.
    Isn't that the same argument that you have been making for years? It isn't the gear, or the game design, but instead a "learn to play" problem? If everyone would "learn to play," they would have the same result that you do.
    If you would ever actually bothered to read what I write...... Instead of the usual self righteous vitriolic anger..

    For your convenience here some quotes from a quick search:



    Here one addressed in a post to you btw:



    Others:



    And I hope people will not "learn to play" to my level because then we are doomed:



    BtW we finished the trial as the group we were since that post.

    And while I have many complaints about the game design, which you are free to look for. I will always argue against people who when stuck at some issue, immediately give up and go whine on the forums blaming players who did succeed the same task. Instead of trying to learn or use the plethora of information available to them to improve themselves.
    External blame as first recourse is an interesting thing.

    Yes, many of those that complain can just learn some basic mechanics and they will be better than me, having upfront better reaction time and better awareness. They just feel they shouldn't do it. It took us half year to get ToMM done as a group what others did in a week. Did you see me post about "evil skilled players"?
    Irrelevant armchair psychiatric analysis, being irrelevant, will also be ignored...

    Hello, Bell Curve, my old friend. "Lies, damned lies, and statistics." I have often wondered what exactly some mathematician did to Mark Twain that caused that witticism to come into being. Most people are aware of the two biggest flaws of statistics... that it is so easily manipulated into giving misleading results, and that the numbers become less accurate the smaller the sample being represented. What many fail to realize, until after they begin to study statistics, is that the entire point of the exercise is to evaluate quantitative data. Qualitative data, being unable to be adequately expressed with numbers, have no place under the bell curve. Skill carries no weight. It displaces no volume. Has neither length, nor depth, nor breadth. And those who would be tempted to express skill as a function of time, will quickly be disappointed. Especially in MMO's with a cash shop. The problem with variables which have no numerical expression, is that they have no place in the decision making process when devising new content. Or the correction of old content. The Devs have access to exponentially more data than we ever will, despite Rainier's best efforts. And, I strongly suspect even they would not be able to come up with an algorithm to successfully measure "skill."
    In all that great musing, what do you actually wish to say? That there is no difference between players? Everyone are equal and the only difference is what? Or we just musing my usage of the bell curve? That skill is qualitative and hence can't be taken into account by the devs?

    Since you studied statistics, and more recently than most of us, remind me what is the common minimal sample size in which statisticians assume the normal distribution indeed in effect (human population, biological samples and so on)?

    Is there dispute that large enough human groups will be distributed by the bell curve or close enough?
    Can we express "skill" as a number? Is it quantitative or qualitative information, especially with a sample size of hundreds. Good question, lets connect the topics in the discussion, what Elo in its essence represents?

    If we don't need to measure in a single players skill, which we don't, I hope there is no dispute about that, can we measure "skill" effects in not direct way? Does the normal distribution still applies? Is there a process in statistics where we convert qualitative data into quantitative?

    For the rest of the readers, interesting tidbit, content is adjusted for some "skill" - twitch reaction times have a minimum.
    One significant measurement of "skill" devs do it by the usage of "difficulty". Difficulty of a content is adjusted with a some target gear range in mind, but how one piece of content can be harder (ToMM) while another much easier (IC) when targeting the same gear range.

    But in the context of class balance this is so not important, as long as it is understood that there is variance in the player base, it and outliers taken into account (which both do, or at least noworries tries) we have no reason at all to measure "skill" directly, we only want the class have similar performance in the relevant subgroups.

    Arbiter is currency again, a good example, outlier up, and lower mid, which implies difficuly curve issue (requirements of "skill") or familiarity.

    PS:
    "Lies, damned lies, and statistics", while made famous by Mark Twain did not come into being by him.
    And psychology would be a better fit, psychiatry is a medical branch and I didn't imply any medical conditions, so that sentence loses a bit of the effect.
    Post edited by micky1p00 on
    Guild: The Imaginary Friends
    If you disagree with me: My opinions are my own and do not reflect on my guild and friends.
    If you agree with me: My opinions are totally the guild official policy !
    ;)
  • wintersmokewintersmoke Member Posts: 1,626 Arc User
    micky1p00 said:

    .

    micky1p00 said:

    micky1p00 said:

    .....

    Or please reply the question that was given.
    Is in your opinion, in PvE, everyone are equally skilled and the only factor is gear? Or "Veterency"?

    Given everyone exactly the same gear, same stats, same power, same class, same time, will everyone have the same result? In terms of damage, and survivability.



    BtW that is not how you "make AI"
    https://openai.com/blog/more-on-dota-2/

    You do not program them with specific moves. This is a "learning machine" you just make them play.
    Isn't that the same argument that you have been making for years? It isn't the gear, or the game design, but instead a "learn to play" problem? If everyone would "learn to play," they would have the same result that you do.
    If you would ever actually bothered to read what I write...... Instead of the usual self righteous vitriolic anger..

    For your convenience here some quotes from a quick search:



    Here one addressed in a post to you btw:



    Others:



    And I hope people will not "learn to play" to my level because then we are doomed:



    BtW we finished the trial as the group we were since that post.

    And while I have many complaints about the game design, which you are free to look for. I will always argue against people who when stuck at some issue, immediately give up and go whine on the forums blaming players who did succeed the same task. Instead of trying to learn or use the plethora of information available to them to improve themselves.
    External blame as first recourse is an interesting thing.

    Yes, many of those that complain can just learn some basic mechanics and they will be better than me, having upfront better reaction time and better awareness. They just feel they shouldn't do it. It took us half year to get ToMM done as a group what others did in a week. Did you see me post about "evil skilled players"?
    Irrelevant armchair psychiatric analysis, being irrelevant, will also be ignored...

    Hello, Bell Curve, my old friend. "Lies, damned lies, and statistics." I have often wondered what exactly some mathematician did to Mark Twain that caused that witticism to come into being. Most people are aware of the two biggest flaws of statistics... that it is so easily manipulated into giving misleading results, and that the numbers become less accurate the smaller the sample being represented. What many fail to realize, until after they begin to study statistics, is that the entire point of the exercise is to evaluate quantitative data. Qualitative data, being unable to be adequately expressed with numbers, have no place under the bell curve. Skill carries no weight. It displaces no volume. Has neither length, nor depth, nor breadth. And those who would be tempted to express skill as a function of time, will quickly be disappointed. Especially in MMO's with a cash shop. The problem with variables which have no numerical expression, is that they have no place in the decision making process when devising new content. Or the correction of old content. The Devs have access to exponentially more data than we ever will, despite Rainier's best efforts. And, I strongly suspect even they would not be able to come up with an algorithm to successfully measure "skill."
    In all that great musing, what do you actually wish to say? That there is no difference between players? Everyone are equal and the only difference is what? Or we just musing my usage of the bell curve? That skill is qualitative and hence can't be taken into account by the devs?

    Since you studied statistics, and more recently than most of us, remind me what is the common minimal sample size in which statisticians assume the normal distribution indeed in effect (human population, biological samples and so on)?

    Is there dispute that large enough human groups will be distributed by the bell curve or close enough?
    Can we express "skill" as a number? Is it quantitative or qualitative information, especially with a sample size of hundreds. Good question, lets connect the topics in the discussion, what Elo in its essence represents?

    If we don't need to measure in a single players skill, which we don't, I hope there is no dispute about that, can we measure "skill" effects in not direct way? Does the normal distribution still applies? Is there a process in statistics where we convert qualitative data into quantitative?

    For the rest of the readers, interesting tidbit, content is adjusted for some "skill" - twitch reaction times have a minimum.
    One significant measurement of "skill" devs do it by the usage of "difficulty". Difficulty of a content is adjusted with a some target gear range in mind, but how one piece of content can be harder (ToMM) while another much easier (IC) when targeting the same gear range.

    But in the context of class balance this is so not important, as long as it is understood that there is variance in the player base, it and outliers taken into account (which both do, or at least noworries tries) we have no reason at all to measure "skill" directly, we only want the class have similar performance in the relevant subgroups.

    Arbiter is currency again, a good example, outlier up, and lower mid, which implies difficuly curve issue (requirements of "skill") or familiarity.

    PS:
    "Lies, damned lies, and statistics", while made famous by Mark Twain did not come into being by him.
    And psychology would be a better fit, psychiatry is a medical branch and I didn't imply any medical conditions, so that sentence loses a bit of the effect.
    I said what I meant. I purposely did not say that all players are equal. That is something that you said. Several times. And, yes I did understand that you put it up as a windmill to tilt against, but still... In your defense of the idea that skill is the reason for the differences in outcome, as opposed to gear, or hours played (veterency) , or the number of runs performed within a certain subset, or (god forbid) the number of place-holders in you Astral Diamond balance, you invoked the Almighty Bell Curve. The problem is... all of those other things, can be expressed numerically. They are quantitative. Skill is not. As such it cannot be measured using the bell curve.

    There is no minimum data set for performing statistical analysis. Most people looking at data would prefer a minimum of 1,000 data-points. But if you are evaluating drive-thru service times at a fast food restaurant, no matter how much the boss would like, that ain't gonna happen. If 90 or 120 responses are what you have, that's your sample size. Good luck.

    The "people" under the bell curve do not eat, or breathe, or take a nap, of have to go potty. They are there to represent a data point. That measures a specific action that the "person" performed. Went thru the drive-thru & got their lunch in fewer than 90 seconds. Drives a hybrid vehicle. Uses a drug, but received a negative result on a drug test. Does not have cancer, but received a positive result on a cancer screening. Completed Neverwinter's newest dungeon wearing gear <IL1000. Most of these are binary. Yes or no. Success or failure. Or actions which take place over a set time period. Things which can be measured with numbers. The Elo system is not an objective measure of skill, but a subjective one. It takes an average of player's performance (wins, losses, and draws) and compares them against other players. Probability based on past performance.

    While some people do attempt to make statistical analysis of qualitative data, the results are more often than not unable to fit into the normal distribution. The data is often separated into categories, do people prefer to drive red cars or blue cars? Or green tractors? These studies often end up with results that are skewed one way or the other. Since the numbers represent feelings, or preferences, they are considered less reliable, Soft data vs. hard data.

    Twitch reactions. Huh. That's new one. I'm guessing something along the lines of How many seconds between the red circle shows up & the player dodges. So, where does skill end, and ping begin? Somewhere in the graphics card, perhaps. Variables. The devil is in the details.
  • gabrieldourdengabrieldourden Member, NW M9 Playtest Posts: 1,212 Arc User
    rjc9000 said:

    TThis reminds me of a power that gave you a buff when you were shooting over a certain distance (cough, cough). By the way they could easily use Aimed Shot for your purposes and that would be the easiest way to improve the Archer Hunter.

    - increase the damage of Aimed Shot and use it as a trigger for the cooldown reduction feat currently based on Commanding Shot
    - bring Binding Arrow to the Hunter path (which also makes sense as all feat support for roots is on the Hunter side)
    - regive a damage buff to Longstrider's Shot when used over a certain distance.

    Tactical Power is the offical-ish terminology I remember that the game used for things like Guard, or Teleport (Wizard), or Sprint.

    The way I envsioned the power was that after pressing it, you would enter the Aimed Shot kneeling animation and gain the following effects:
    • -3 Seconds base cooldown on ranged attacks
    • +Damage Buff on Ranged attacks equal to your DEX score (so 25 DEX = 25% damage buff while in this tactical power)
    • Can't move until you re-press the button.
    Slotting this tactical power in place of the short dash would lock out your ability to swap weapons, but give you an ranged additional encounter power slot.

    I suppose I'd call it "Steady Aim" (I could have sworn I saw a class feature/manuever in the 5e Player Handbook with the name "Steady Aim") and the reasoning behind this power is so players who want to focus on pure ranged damage can denote "hey I'm playing Archery style" with this tactical power.

    I am not sure if the game can handle players having multiple tactical powers, or being able to slot in some tactical powers over others, which is why I was asking if it was within the scope of the game engine.

    I also would not be opposed to bringing back the buff of Longstrider's Shot as well, though hopefully its (relatively) high buff amount and short duration (idk, 30% for 5 seconds?) would be kept so it puts the pressure on the player to choose their next attacks wisely.
    Mmm, I still see the changes I outlined above easier to implement, and anyway you'll be using Aimed Shot a lot which will make you immediately recognizable as an archer. What could be interesting would be the chance to replace the stance change power with a fourth encounter, like spell mastery does for wizards. I opened a thread in the wilds called "State of the Ranger" and would be very interested in your comments there.
    Le-Shan: HR level 80 (main)
    Born of Black Wind: SW Level 80
  • micky1p00micky1p00 Member, NW M9 Playtest Posts: 3,391 Arc User
    edited February 2020


    I said what I meant. I purposely did not say that all players are equal. That is something that you said. Several times. And, yes I did understand that you put it up as a windmill to tilt against, but still... In your defense of the idea that skill is the reason for the differences in outcome, as opposed to gear, or hours played (veterency) , or the number of runs performed within a certain subset, or (god forbid) the number of place-holders in you Astral Diamond balance, you invoked the Almighty Bell Curve. The problem is... all of those other things, can be expressed numerically. They are quantitative. Skill is not. As such it cannot be measured using the bell curve.

    I didn't put anything as windmill to tilt, I don't know what you want here. You jumped in and trying to get to something or somewhere, trying to bend things.

    There was a simple post, and a simple question, what do you want with it. I understand that your will twist everything and anything in some weird attempt to make me look bad in your own perception or to self validate your some perception.

    Here is the best example ever:

    " In your defense of the idea that skill is the reason for the differences in outcome, as opposed to gear, or hours played (veterency) , or the number of runs performed within a certain subset, or (god forbid) the number of place-holders in you Astral Diamond balance,"

    In no way form or shape I ever said that!

    And like I've wrote before, you don't bother to read, or care to understand. I wrote multiple times, and we can do the history dive all again, though obviously it will be a waste of time, that player performance depends on their experience, gear, class and meta synergy and multiple other factors. Skill is a factor, one of many others. Hence in my question it was phrased as "all other things equal", this is how you do analysis of a single variable in a static system, but you should know that . I never in any post "defended" that skill is the difference as opposed to other things.

    Like you yourself tried to "mock" with that mention that I've been sanding players to "learn to play". Cute. But interestingly the game difficulty adjusted in way that if most players will take the little time needed and try to learn the dungeon mechanics or class or what other factors there are, it can easily compensate, for the variance in skill, a run can be slower, with more deaths, but it will work out. Together with other factors like gear, experience, and so on.

    What is your point? You argue against me hypothetically saying all players equal, which I also never said btw, I did say that we all started from the same starting point in the game (though it was in the context of someone who started also at around the same time). Or now you arguing against factors that make players not equal. So you just like to spend time with me on the forums? It will be more pleasant if you will actually try to understand the intended meaning instead of summarizing into your preconceptions.


    There is no minimum data set for performing statistical analysis. Most people looking at data would prefer a minimum of 1,000 data-points. But if you are evaluating drive-thru service times at a fast food restaurant, no matter how much the boss would like, that ain't gonna happen. If 90 or 120 responses are what you have, that's your sample size. Good luck.

    That was not the question. My question was specific, and while it was long long ago, it may surprise you that I did learn statistics, and I knew the pitfalls of various types of research well enough.


    The "people" under the bell curve do not eat, or breathe, or take a nap, of have to go potty. They are there to represent a data point. That measures a specific action that the "person" performed. Went thru the drive-thru & got their lunch in fewer than 90 seconds. Drives a hybrid vehicle. Uses a drug, but received a negative result on a drug test. Does not have cancer, but received a positive result on a cancer screening. Completed Neverwinter's newest dungeon wearing gear

    To start, that is not how Elo works.

    But lets try differently, can we, in the context of class balance, define players skill? Please do look at my original question at which you first replied for that.

    Lets define that, and then move to it's well studied parallel, and well debated as a predictive model.


    While some people do attempt to make statistical analysis of qualitative data, the results are more often than not unable to fit into the normal distribution. The data is often separated into categories, do people prefer to drive red cars or blue cars? Or green tractors? These studies often end up with results that are skewed one way or the other. Since the numbers represent feelings, or preferences, they are considered less reliable, Soft data vs. hard data.

    I suggest you look up the questions I've asked before. They were not asked from ignorance. Statistics has many limitations, but it is a science, and many things can be derived when understanding those limitations.

    To the current context, we do not discuss new things here, reaction time, spatial awareness, learning, decision making time, and other factors are well researched in sports, education, military. Some of them using very modern tools like VR and eye tracking cameras, saw some live, but that besides the point.


    Twitch reactions. Huh. That's new one. I'm guessing something along the lines of How many seconds between the red circle shows up & the player dodges. So, where does skill end, and ping begin? Somewhere in the graphics card, perhaps. Variables. The devil is in the details.

    New? Not at all. Can I measure all of the factors separately? Can we put them in numbers? Can we account for each one of them.
    I'm sure you can look the topic up, the tolerances in tournaments, in different game types, why they do LAN tournaments and so on. What orders of magnitude each factors contributes and so on.
    But do we even care? Here you have a 100% measurable, quantifiable skill factor, once we remove all those other pesky variables, that can be shoved perfectly into a normal distribution.

    So now lets reduce the problem to the following, given the same location, same ping, same hardware, and all the same conditions otherwise, will be a difference in reaction time within different people?
    Guild: The Imaginary Friends
    If you disagree with me: My opinions are my own and do not reflect on my guild and friends.
    If you agree with me: My opinions are totally the guild official policy !
    ;)
  • jules#6770 jules Member Posts: 583 Arc User
    Oh my, @micky1p00 when my notifications popped up this was not what I expected. Thank you a whole lot <3
    - bye bye -
  • admiralwarlord#3792 admiralwarlord Member Posts: 403 Arc User
    I don't know where the topic's heading is going ?!

    The creator of the topic spoke how happy he was to see that all classes are capable of carrying out any content today. I only replied that despite appearing, I disagreed with the current balance of class by citing something he had already mentioned. When I saw that Dev had entered the discourse, I went to read the whole topic to understand what he said in his post and I see an act of elitism and a statement of which I agreed. I don't care if he was in the first ToMM to be performed, I still don't agree with his statement that he is more skilled than the other. In another topic I about the balance of classes I made a challenge for Dev, but it would be useful for players like him who say they are more skilled than others, I said to try to finish ToMM with 1 healer DC, 1 healer SW, 6 DPS that could be GWF GF SW and DC, did anyone? The few GWFs that have the weapons of the Lion set have probably succeeded in another character. I know why I kept playing this trash class while most switched to another. Points of intelligence for them, that unlike me, I preferred to continue playing my class, even though I had a GF Tank and a CW with full campaigns, even though I participated in the tragedy that was the preview of Mod 16 and knowing what would come through front.

    I don't know mountaineering championships well, but I know that the competitors start from the same one to reach the goal that is the summit. I'm sure, since they are professionals in what they do, they trained to do that. Now imagine mountaineer X having the perfect equipment (even more than he should) on the day of the race, and mountaineer Y with the equipment that was told it was possible (when in fact it was not), went to compete, which of the 2 do you think would win?

    I continue to find it strangely embarrassing to say that he is more skilled than the other, given that they never met. If it were me I would have invited us to go to the server preview and we would have tested whether the statement was true or false in a PVP battle. I even searched the PVP leaderbord to see what position he was in, but I didn't find it. It would be nice to see him and his super skillful friends participate in the pvp tournament that the community is organizing, to see how they would fare against random and unscheduled movements from other players
  • cherryman1cherryman1 Member Posts: 337 Arc User
    Back to the original topic: I still see a need to work on class balance. I see some things that need to be improved. That said, it is better than what was originally done at the start of Mod 16 through 17. The balance was probably the worst I have seen in game during these mods. So much so that if you were in the classes that were the best in ToMM at end game your AD is through the roof. Those who were in the less desirable classes didn't get to reap those benefits. In fact most players left from those classes. Even though the balance of classes was comparably bad in mod 15 and earlier. Those mods used a setup that every class was wanted in end game. That isn't the case now based on the combat change.

    Some of the changes I would recommend would be the following for each class:

    CW: small nerf to single target, decent buff to AoE damage
    Barb: small increase to single target, small nerf to AoE damage
    HR: Archer is the worst paragon in game for dps for end game. Needs to be buffed.
    TR: single target is fine, small increase to AoE damage
    SW: look at the multi SW setup that seems to increase the damage too much. decent buff to single target, small buff to AoE
    Fighter: Small increase to single target, AoE is fine. Look at the issues with the squat dance to see if that can be improved. Recommend to be based more like the barb where attacks build it up.
    Cleric: Small increase to single target, AoE is fine.

    Guild Leader: Under the Influence
    Yule (Barb): 53k : Siren (TR): 52k : Torun (DC): 49k : Siren OP (OP): 52k : Siren SW (SW): 50k : Modern (F): 42k : Cherry1 (CW) : 42k Siren HR (HR): 50k
This discussion has been closed.