Wednesday, May 04, 2011

More Thoughts on Optimality

Choices and Futures

Several commenters took exception to my statement that "for a decision to be meaningful, there must be a right choice and a wrong choice."

Kring states:
I couldn't disagree more. You only have a real choice if all choices are valid.

It's like democracy. Just because you're allowed to vote doesn't mean you also have a choice. You only have a choice if there are multiple good and relevant parties.

I don't really think we're that far apart. Perhaps I'll rephrase and extend my thoughts.

For a choice to be real, it must lead to different futures. If you vote Conservative or NDP, it's a choice because the future with a Conservative government is different than the future with an NDP government. However, the desirability of each future matters.

When voting, neither future is obviously desirable. Some people would prefer the Conservative government, others the NDP. Thus neither choice is right or wrong.

However, this is not the case with optimizing. In one future, the boss dies. The other future, the boss doesn't die. The future where the boss dies is 100% more desirable. Thus, barring any side-effects, the choices that shift the probabilities towards first future are "right", and the other choices are "wrong".

For another example, take a Civ-like game. Both going for military victory or cultural victory are valid choices. But if you're pursuing a cultural victory, randomly producing tanks for no real reason when you could have made a cathedral instead is the "wrong" choice.

Now if it's a single player game, then no one else other than you pays the price for your wrong decision. But in a team game, wrong decisions hurt your team's chances of victory.

Shintar phrases this in a different manner:
Situations where there is a clearly right and a clearly wrong solution are calculations, not choices.

Sure, if that's the semantics you want to use. But it is incumbent on you to pick the best calculation in a team game. And the best calculation is determined by the optimizers.

Good Players and Sub-Optimal Choices

Masterlooter states:
If you have the "optimal" setup, and I have a "sub-optimal" one, but I do more damage than you (with equal gear), then I am more valuable to the raid. But since my setup is considered to be "inferior", I may not even get an invite to the group in the first place.

In my view, this is a bit of a strawman. The truth is that this very rarely happens.

High-skill players almost always use the optimal spec or strategy. There are a few exceptions, but they are very rare. The vast majority of the time, someone with a non-optimal spec turns out to be a low-skill player.

Encouraging medium or low skill players to feel that they are a "special snowflake" and don't need to use more optimal builds dooms them and their group to mediocrity and failure. You have enough trouble with lack of skill, why further handicap yourself with a sub-optimal build?

And even the high-skill player with the sub-optimal spec does her group a disservice. If the high-skill player switched to the optimal spec, odds are she would play at an even higher level.

DPS Meters

There's also a lot of antipathy for DPS meters among the non-optimizing crowd. This is probably uncharitable of me, but sometimes it seems like DPS just want the freedom to play badly and not get called out for it.

They'd rather no one be able to tell how terrible they really are. That might mean they have to take the "effort" to improve. Just let the tanks and healers carry the group and do the work, while they sit back and collect the loot.

After all, it's pretty obvious when the tanks and healers are failing. What's so wrong with having an element that makes it just as obvious that the DPS are failing?

Good players don't worry about DPS meters. You know why? Because they post respectable results. They're an asset to the group instead of dead weight.

30 comments:

  1. "In one future, the boss dies. The other future, the boss doesn't die. "

    What if both futures lead to the boss dying?

    ReplyDelete
  2. I think you completely misunderstood my post, and many of your commenters. The point is not that we don't want optimal. The point is that we believe that *finding* the optimal is an essential part of the game, maybe even "the fun part" of the game.

    Thus ultimately, yes, I very much would like my characters to be optimal. Just like ultimately, at the end, I would very much like to know who the murderer was at the end of a crime novel. What I object against is the skipping to the end.

    As you say, different choices result in different futures. In a game the "bad" future isn't really all that bad. The boss isn't dead at the end of the evening? Well, we'll get him next night. No lasting damage. Thus the essence of "playing a game" is trial and error, where you try out your own ideas, and see whether they work or not.

    Looking up the optimal solution on EJ and BossKillers is like skipping to the last page of the crime novel: You got the final result, but not the fun part leading up to it.

    ReplyDelete
  3. @Spinksville - you're 100% right there. I say, if the boss dies in both futures, but maybe in one of them he takes a minute longer to die but everyone has more fun, I'd take future number 2.

    ReplyDelete
  4. @tobold I wouldn't go so far as to say that EJ spoils the fun of optimizing.
    Some people consider EJ the end of the Novel, others think of it as a starting point.
    I like to have a comparison point so I can figure out how far from the optimal my personal sub-optimal performs, so to say.

    Now as to what can you expect from other players, well if they are in my guild I'd like them to be able to justify their choices compared to the so called "optimal"
    Not everyone can be trusted with the ability to correctly judge the effect of a change in his template without at least some guidance. In fact, most players can't.
    And in some situations you don't have the time to wait for those players.

    ReplyDelete
  5. I think that there is an obvious way to tell when DPS fails without using meters (at least in some fights). The enrage mechanic can easily be tuned to reveal a lack of DPS. If a boss enters an enraged state, then you know that DPS needs to improve.

    ReplyDelete
  6. Choices:
    "for a decision to be meaningful, there must be a right choice and a wrong choice."
    When *I* see YOU make a choice:
    if the choices are very similar, I learn very little about you and say "there was no *real* choice, there was no meaning." But if the options are very different I say "THAT was a choice, and I learned something about you, very meaningful"

    When *I* make a choice: if the options are very similar, I say "now thats a tough decision, very meaningful". When they are very differant I say "well, thats no choice, one is clearly correct, the other just as clearly wrong, there is no meaning."

    So how meaningful a decision is depends on who is making the choice, me or you; its all about perspective.

    DPS Meters:
    I likes em. I like to look and say either "I did good in that fight" or "I can do better". BUT, I also believe that when viewing them you need to take other factors into account "my dps was low that fight because I was interrupting a lot, so while I had low dps, that was still a good fight."

    ReplyDelete
  7. In the end, it is just a game. I think damage meters and Gear Score ruined WoW. I am sick and tired of people degrading others because during ONE fight, they didn't top out on the damage meters. People use it to bolster their own egos and deride others. This is the majority, not the minority.

    ReplyDelete
  8. "After all, it's pretty obvious when the tanks and healers are failing. What's so wrong with having an element that makes it just as obvious that the DPS are failing?

    Good players don't worry about DPS meters. You know why? Because they post respectable results. They're an asset to the group instead of dead weight."


    Not true for me, but perhaps I'm the dead weight you speak of.

    What I resent isn't the meter itself or the elitist attitude you're revealed above.

    It's the idiocy that I find myself unwilling to put up with.

    Whatever the current "acceptable" dps level is right now, it will no longer be accceptable in a month or two.

    However, if x dps is the top slot or two and kills a given boss right now, that same x dps will kill that same boss in July. But that x dps will no longer be "acceptable"

    This is exactly the situation you find yourself in should you step away from the game for a couple months. Dealing with people who use metrics they don't understand to inflate their own sense of success and importance.

    This is not new. I stepped away after Ulduar in WoLK, but came back when Icecrown opened. The difference then was that I really wanted to see the Lich King, so I sucked it up, grinded out my gear and caught back up.

    I have no such motivation in Cata. If I step off the progression path, there is no reason to attempt to step back on.

    ReplyDelete
  9. Spinks,

    It comes down to an issue of shared burden at that point. If both futures end with the boss on the floor, then you have to ask yourself if you're ok with an encounter where healing, tanking and execution have to go right for 9 minutes instead of 6 so that the Subtlety rogue, the feral druid who thinks Savage Roar is too troublesome to bother keeping up, and the Mage who gems pure haste can have their fun.

    ReplyDelete
  10. I can honeslty say i have never fought a boss and said "OMG im having so much fun getting my ass kicked... lets drag this on another two mins." Maybe its because im a healer and if i can keep a raid alive through all this destruction i expect the dps to do their job and kill the boss as quickly and efficently as the strat allows. I lead soldiers in real life, and i can honestly say i would airborne drop into every mission if i could cause its awesome and fun to jump out of a plane at 10k feet, but is that benifical to the mission? Does it help us kill the bad guys, 98% of the time no, it doesnt. Because an airborne drop is rife with massive supply problems and horrible % of mission success due to even the smallest of miscalculations. Killing bosses as effectively as possible is the fun part becuase their is a successful conclusion to your hard work, wiping all night with nothing to show for it is.... well, not fun at all.

    ReplyDelete
  11. @Animagis: Why are you conflating fun with playing stupidly?

    ReplyDelete
  12. @klep
    he isn't (I think)
    He is saying that for some people, playing fun just happens to also be stupid.
    "Stacking haste so that my spells come super fast makes my animation cool and I think thats fun!" Ok, but it happens to be stupid as your DPS is now super low.

    @Hyperian
    I agree. But sometimes I can make the fight shorter by interrupting, which will give me a lower DPS. SO, the DPS meter says I suck, BUT the fight was shorter. I am not suggesting that you contradicted this, just taking the excuse to point it out. :D

    ReplyDelete
  13. @Brian Carpenter: In a party in WoW, there is only one person tanking, one person healing but 4 people doing the damage so it's not possible to tell which one lacks DPS just by observing how long the mobs live.

    @Tobold: I like your metaphor: not every detective book reader is a "whodunnit" purist; and there are people who do not require a clean sheet and can have fun tuning their build starting from a premade build.

    ReplyDelete
  14. Having the optimal spec, gems, et al, isn't an indicator that someone knows their own class. It's an indicator that someone can google "Shadow priest DPS rotation" and click on EJ.com - copy/paste the info there.

    We're no longer teaching players how to think, we're teaching them that it's OK -NOT- to think.

    Someone that learned their class by trial and error, would naturally have their setup a bit off from the EJ boys and girls. It would be very similar of course, but there would be differences that were largely unnoticable unless you crunched some really big numbers - such as Crit instead of STR gems, or what have you.

    Note, that I make a distinction between a player "doing whatever they want", and playing poorly - they are not mutually exclusive. Set PERFORMANCE benchmarks, expect players to meet them. If they can meet your 15k DPS requirment as a Holy Priest, who are you to say they are "doing it wrong"?

    Judge on performance, not copy/paste technique.

    ReplyDelete
  15. It's funny that you don't see this "I should have the choice to be suboptimal" in other team sports.

    No one gets away with, "I should be allowed to throw the football underhand because that's more fun for me." No one stays on the soccer team with "but I like trying to play with just my knees." You want to play basketball with dress shoes on? Good luck finding four other guys willing to be on your team.

    So why do people feel like they are entitled to this in MMOs?

    ReplyDelete
  16. In my experience, the boss never dies "a minute later", at least on challenging content.

    For a new boss kill, the choice is never between a 6 min kill or a 9 min kill. The choice is between a 6 min kill or a wipe.

    ReplyDelete
  17. Rohan,

    I completely agree with you. I have given up on people in my guild (I lead us) as I and everyone but one or two believe that the most fun and single goal we should pursue is killing bosses. If someone finds more enjoyment out of doing less POTENTIAL dps (because they are in a sub-optimal spec, they COULD easily do more dps, regardless of boss kill) when the boss kill is not guaranteed, we are not seeing eye to eye in what our time should be spent on for raiding.

    If the boss kill is in the future regardless of the spec, then it's farm/easy content and this argument is pointless as we screw around and play with random specs on farm content all the time. I assumed that this conversation was aimed at the situation where boss kills are not guaranteed and thus you are looking for every edge to do so.

    ReplyDelete
  18. I still think that any optimization, that must be made, is not a choice because you are forced to do it to be successful.

    Let's compare the shaman class with the mage class. The shaman class has two DD tree, enhancement and elemental. One is melee, the other is a caster spec. They require different gear. In the history of WoW one of those two specs was always much much more sought by raids (if I remember correctly). But both specs do have their use. If your class is a shaman and you would like to play it as a DD you have a CHOICE. It's unlikely that you're being asked to switch from one DD spec to the other. Nobody will flame you because you play the weaker spec because that spec has useful utility the other doesn't have.

    Now with a mage you have 3 ranged spell casting DD specs. The decision betweem fire and frost is not a choice, it's a CALCULATION. If you would sign up to a raid as frost you would get flamed. And rightfully so, because frost sucks. (At least it did some time back. I didn't check 4.1 changes.)

    The game needs more CHOICES and not CALCULATIONS, methinks.

    ----------------------------------------

    Besides that I understand the request very well. For me, creating decks is the most fun part in Magic: The Gathering. Playing comes secondary.

    I had a lot of fun trying to find MY BIS set in WoW - that's the set of items *I* can realistically obtain as next step. Spending hours in wowhead and with a spread sheat and Simulationcraft. That's a part I really enjoyed in WoW.

    Unfortunately this part of the game already died with the ToC patch where every item had nearly the same stat.

    ReplyDelete
  19. Besides that, it's completely nonsence to assume that a human being would be able to test a few builds and play styles with a current WoW character to find out the most efficient spec and rotation. That's bullshit. Just switching two points in your fire spec and casting 20 fireballs on a dummy will not tell you which spec is better.

    It is impossible to find the knowledge currently found on EJ about your class by a human being alone. No one is able to do that himself. This information can only be collected by running many simulations with tools like Simulationcraft, a tool created by a crew, not a single individual, and reviewed by even more.

    There are people attacking a training dummy FOR HOURS just to identy the crit chance of his pet. No one has enough time to do all those tests himself that had to be done to create Simulationcraft.

    Today, the game is that complex that you can't just try some builds.

    To bring back the "let's try some build and pick the best one" would require to "dumb down" the current mechanic a lot. It would mean to go back to the day of "spamming shadow bolt".

    And if they would do that it would take the Internet about 5 minutes to present the best spec. And if you wouldn't read it, the raids wouldn't pick you or call you a slacker.

    The problem with this whole optimization discussion is that there are people who would enjoy something that cannot exist anymore. It died within the Internet. There is no point in wishing that to come back, it won't because it can't.

    ----------------------------------------

    But there are new choices in the game.

    We know the perfect "rotation" for patch werk which ... doesn't exist anymore. And as soon as something happens you'll have choices.

    As a warlock your choices are:
    - You can life tap of you have any use for the mana and can afford to loose health (normally something bad happens when you're forced to move).
    - You can renew corruption early.
    - You can cast fel flame to renew Immolate.
    - You can renew curse of doom.
    - You can cast shadow fury for some damage.
    - You even have to decide to abort your cast and move or stay in the fire till your cast is finished.

    These are things that cannot be calculated in advance as there are way to many options.

    But, of course, "arcade choices" are not the same as "preparation choices". I feel your pain.

    ReplyDelete
  20. > Phelps said...
    > It's funny that you don't see this
    > "I should have the choice to be suboptimal"
    > in other team sports.

    But in a soccer team you also don't have the "BRB door" or "AFK pizza arrived" or "wife aggro" or "AFK bio" or the most evil of them all "rl comes first". :)

    ReplyDelete
  21. @Phelps
    "It's funny that you don't see this "I should have the choice to be suboptimal" in other team sports."

    It's funny that the best players in the history of some sports are those that pretty much said, "screw what everyone else is doing, I'm going to try something different" - rather than sticking to the same treadmill as everyone else. Michael Jordan, David Beckham, Michael Vick to name a few. They game they played was litterally redefined after their presense - explicitly because they DIDN'T do what everyone else was doing.

    (not necessarily @Phelps)
    The accepted "best" is just that - what's accepted. That doesn't mean that there isn't something that works just as good, or something that might even be better.

    Performance is objective, everything else is subjective.

    ReplyDelete
  22. @Masterlooter

    Are you suggesting, then, that even a small percentage of people who want to show up to group content in suboptimal set-ups are, in fact, equivalent to the Michael Jordans and David Beckhams of WoW?

    ReplyDelete
  23. My issue has never been optimization, but play style. This comes through more with Rift, I think, then WoW, but it's still present in some cases (melee vs ranged dps shaman, even various mage trees). If I want to play a certain way, I have to hope that that tree or that soul is currently at least viable, or I'm stuck doing something that isn't as fun. If you're talking fully optimal, that ups the stakes even more. I'm fine with cookie cutter specs, but not at the expense of sidelining entire play styles.

    ReplyDelete
  24. Thus ultimately, yes, I very much would like my characters to be optimal. Just like ultimately, at the end, I would very much like to know who the murderer was at the end of a crime novel. What I object against is the skipping to the end.

    Poor analogy. Is reading a crime novel a team-based activity? No. Imagine instead that you work in a police department, and are trying to solve the murder yourself (along with other detectives). You get an anonymous tip that the killer lives on the 3rd floor apartments across the street. Instead of acting on this information, you decide you want to canvas the neighborhood instead. Your "exploration" hurts everyone around you, no matter how fun it may be to you personally. My analogy my be a bit more bombastic, but bottom line is that you necessarily have fun at others' expense when you "explore" in group activities.

    As you say, different choices result in different futures. In a game the "bad" future isn't really all that bad. The boss isn't dead at the end of the evening? Well, we'll get him next night. No lasting damage. Thus the essence of "playing a game" is trial and error, where you try out your own ideas, and see whether they work or not.

    Excuse me? No lasting damage? You get 9/24 other human beings across the country to all log on at the same time to do an activity that is impossible to even see otherwise, and there is "no lasting damage" when the boss doesn't die at the end of the evening? Do you not value your own time, or the time of other people? That is perhaps the most self-centered, selfish attitude I have ever heard of in the context of WoW... worse than the GearScore or the raiding "elistists," whom at least fundamentally care about the group's success.

    Grouping is about making personal sacrifices to ensure group successes. If you are not giving 100%, then you are (though it pains me to channel Gevlon here) leeching from everyone else who is. You are the guy "exploring" new ways to get to work while carpooling, making everyone else in the car late. Explore on your own time.

    Re: Well, Michael Jordan, etc, "explored."

    /facepalm

    I have never required my raiders to spec a certain way, as long as they brought the numbers. If they "explore" and bring better DPS the first time, then go for it. If the remaining 9 people have pick up their slack for 20+ wipes on the mere possibility that they found what 50,000 simulations didn't, no thanks. Bring your A game, or I will find someone who will.

    ReplyDelete
  25. Masterlooter, that is a HORRIBLE analogy. Michael Jordan didn't come up with a new way of playing. He took the best (optimal) way and practiced it so much that he became far better at the FUNDAMENTALS -- the optimal -- than anyone else.

    Jordan, Beckham, and Vick have the strongest fundamental skills in their sport. Vick isn't going out and throwing sidearm to see if it is better. He's throwing the same way as everyone else, with more strength, precision and better timing.

    In other words, they all went with the optimal build and then learned their rotations.

    ReplyDelete
  26. @ Hyperian

    I've also been a soldier for 8 years, so I understand some things about real life tactics, etc. The problem with your statement is that it's an inappropriate comparison; WoW is a game, and what you do is real life.

    You don't do airborne insertions for fun, because real human lives are at stake. If shit goes wrong, someone dies.

    WoW is a game whose purpose, at least ostensibly, is to be fun. When WoW starts being treated like it's a second unpaid job, something is very wrong.

    I have a job that I take very seriously. WoW is for fun; I once quit playing it because it ceased to be fun, and if it again ceases to be fun, I'll quit a second time.

    ReplyDelete
  27. WoW is a game whose purpose, at least ostensibly, is to be fun. When WoW starts being treated like it's a second unpaid job, something is very wrong.

    Raiding and instancing is a team sport. When you are on the team, you are expected to help the team to the best of your ability. They guy who wants to wear dressy high heels on the soccer team is going to make enemies of his teammates, whether he thinks he might do better in high heels or not.

    ReplyDelete
  28. "No lasting damage? You get 9/24 other human beings across the country to all log on at the same time to do an activity that is impossible to even see otherwise, and there is "no lasting damage" when the boss doesn't die at the end of the evening? Do you not value your own time, or the time of other people? "

    You're making the usual false assumption that the purpose of raiding is to kill bosses. It isn't. It's a social group activity for people to do in games.

    It actually doesn't matter what happens during a raid as long as the people present had fun.

    Same way that it doesn't actually matter if your local amateur football team lose a match as long as they had a good game.

    ReplyDelete
  29. Masterlooter may have gotten his analogy with regards to some of sports most gifted players, but his final paragraphs are spot on:

    The accepted "best" is just that - what's accepted. That doesn't mean that there isn't something that works just as good, or something that might even be better.

    Performance is objective, everything else is subjective.


    Back in the days of Vanilla WoW, I rememebr moving my Warlock from a guild struggling to fully clear MC weekly to one farming BWL and 70% of the way through AQ40. I was in full Tier 1 with ZG / AQ20 non-set pieces, but most of the other Warlocks wore 4 or more Tier 2 and had some of the AQ40 non-set pieces...

    In my first full BWL run I outperformed all of those locks by at least 20%. Whether they had the optimal setup or rotation didn't matter - they were perfomring to the 'accepted' level required to clear the content.

    WoW is a funny beast in that right now, at this stage of the expansion, everyone 'needs' to be performing almost to the limits of the capability of their class if they want a run because people want to steamroll all of the content. Once the content is even slightly outdated and a majority of the players outgear it, then what is considered optimal by EJ will no longer be 'required' to get into a run. Then no-one will care if you're a fresh 85 with a 329 iLevel because they won't even notice whether you're there or not.

    It happened in Vanilla (to a lesser extent due to the small difference in the gear through Tiers 1 to 3), BC and Wrath and will continue with future patches and expansions. It's starting to happen right now with heroics where CC is the exception rather than the rule…

    Now is the time that you can start having 'fun' (in heroics anyway) trying that off-the-wall spec or rotation because what is 'accepted' is no longer the optimum.

    ReplyDelete
  30. You're making the usual false assumption that the purpose of raiding is to kill bosses. It isn't. It's a social group activity for people to do in games.

    It actually doesn't matter what happens during a raid as long as the people present had fun.


    @Spinks, I now see the disconnect you are having with the opinions of those here and elsewhere; You raid purely for the fun of it, regardless of outcome (as you stated). Many other people, such as myself, do not. I find the most fun in the game to be killing bosses. My fun IS measured upon the success of defeating bosses, most specifically those who are presenting a challenge to us at present (progression). If someone can't help me kill new bosses, they are not helping my raid group have fun and thus should go raid in their sub-optimal spec with you as you will possibly enjoy raiding with them and they would enjoy raiding with you. You are not right or wrong, just different. I want to play professional sports, and you want to play a pick-up-game at the local yard.

    @masterlooter, your analogy of MJ/DB/Vick falls short when you realize that game changers do exist in WoW, and they are accepted as viable players- Remember when HaT rogues (a clearly inferior spec at the time) generated extra combo points with each crit in the party for EVERY rogue? Stacking 5 rogues in a 25man led to ridiculous numbers for a few weeks. When you raid in a sub-optimal spec and do VERY WELL, blizzard fixes you, as it is not "working as intended."

    ReplyDelete