Tuesday, May 3, 2011

The Phobia of Inefficiency

Tobold laments the fear of being sub-optimal in MMORPGs, and especially World of Warcraft. Just like about everybody else, I agree. Actually, even Ghostcrawler identified this problem.

[..] I know you need some basis to evaluate potential recruits or even pug members. But I do wish there was some way to turn around this virtual phobia of inefficiency -- this terror of being WRONG -- that we have managed to instill in our player base. I honestly think it's one of the greatest challenges facing the game.

Rift is also running straight into this trap at high speed.

So what can we do to fight this challenge?

1) You can make it harder for players to optimize. For example, by disabling the export of combat logs for statistical analysis. Or by disabling combat/dps meters, the armory, inspecting other players, etc. Evidence from MMORPG-history shows that this helps to a degree.

2) You can remove one-dimensional goals. In WoW, a damage dealer is exactly as good as his damage-per-second. Of course, he should also try to interrupt, but in the end everybody looks at the dps meter. That's concrete and readily available feedback. If a damage dealer had to do much more important things than dps during a fight, it were harder to optimize a build. If fights were completely different and not as similar as, for example in WoW, it would also help.

3) You can make content less predictable. A procedurally generated dungeon with or without traps, with or without undead, would make it much more reasonable to take a specalists for traps/undead with you, even though there might not be any such challenges. Of course, this also requires that players can chose to evade some encounters that they just aren't prepared for. This would add a completely new interesting decision to the MMORPG; and a new minigame: information gathering.

4) To make perfectly balanced challenging PvE content, you need to know how many players are going to try to beat it. However, this also leads to every non-optimized player being a problem. Now, in a non-instanced MMORPG, every additional player would help. Players would love to add more players to their groups, because more is better. Of course, this also helps with immerison.

5) And, finally, you can disable any 'interesting decision' that changes the efficiency of the character. If the game offers enough interesting decisions in other domains, this might even be a good solution.

Also have a look at my post last year Cookie Cutter Speccs. I realize that this blog is now old enough that I have written about almost anything more than once already ;)

6) There is a sixth option. You can reduce the challenge of content. The less challenge, the less incentive to optimize.


  1. "2) You can remove one-dimensional goals"

    I think that's the most important thing. Having efficient ways of doing things is not the problem; it's a problem if there is only room for one way. Ideally things would be balanced so that no matter what you do a build could be effective in a unique way.

  2. "1)" Is opposite what I personally seek. I want play with competent players. Damage meters can help measure that .Without them I know something is off as mobs dying slowly but I wouldnt be able to tell who is the culprit

    In pvp its much more extreme. In fact extreme to such degree that without tools I would never play with strangers. -on average (just like everything else) -they completely suck

    That means I would have to spent resources (time and effort ) to get into hardcore guild. The casual players will never get there, hardcore guild dominates. Casuals quit (or avoid that part of game altogether)

    DPS is just something easily measurable ,without it player still have skilled and unskilled ones. Its just harder to separate one from another which leads to more frustration not less

    In fact big problem with LFD is exactly this - you cant choose who you group with by skill. If you could I bet it would be much more pleasurable experience

    Inefficiency is bad mkkay?- We need to eliminate it ,not find various ways to hide it under the carpet . Dog shit under carpet will still stink

  3. A SMALL step in the right direction would be announcing things like interrupts (the way that some addons do now).

    So every time you successfully use your interrupt, everyone in your party sees it. That would make them think "hey, maybe I should be interrupting."

    Things like this would encourage stats other than (or in addition to) simple DPS.

    Maybe even announce when your threat level gets too high.

  4. One-dimensional goals gives me an idea. What if you had bosses who reacted dynamically to the party's efficiency? Essentially an AI Director style encounter that watches the players and tunes it on the fly to keep the challenge meaningful.

    Let's say the party has an inefficient healer. An encounter tuned for a high-average healer would reasonably result in OOM and Wipe. But the AI Director detects (by measuring HPS or clicks per second)the healer's inefficiency and lowers the damage output of the event.

    Let's say the party has overgeared dps. For a regular run, it would result in easy mode, an AoE fest without consequence. The AI Director detects this and spawns some adds, increases the speed at which the boss drops movement-motivating fire, etc.

    Now, reward the party for pushing the difficulty of the encounter. More points, higher ilvl gear (or even more drops!). For each level the AI Director increases the difficulty, increase the reward. Tune the introductory fight to be easy enough and you have an encounter that can be cleared by the inefficient and inspires pride in the efficient (I was able to get Boss X to the fifth tier of difficulty! Realm first!)

  5. I like Daniel's comment, as a theoretical direction to go in. Valve likes its AI Directors (L4D series), and they work really well, or at least I enjoyed those games a lot.

    AI Directors *also* create randomised content, which when done well stays new and fresh for a long time. I can play L4D and L4D2 missions a lot more times than I can run WoW heroics without getting bored.

    I suppose MMO players might try to game it, by doing low DPS at the start or some such, but perhaps there are ways to design around that.

  6. My theoretical design against gaming the system would be sub-optimal rewards. Sure, we could beat this boss by keeping our efficiency low to trick the AI Director into running the fight at minimum difficulty. But if we push ourselves we could vastly improve our reward for beating this encounter. Give the players a string of carrots, each bigger than the next if they just keep straining their necks to reach it.

    Another defense against gaming the system would be setting a sufficiently high minimum difficulty. High enough that you couldn't beat it without basic effort. Any further effort is rewarded further.

  7. I think one of the problems is that there can be calculated best in slot gear for each build. This happens because the top gear isn't randomized except in whether or not you'll ever get to loot it. I'd like to see more diablo 2 rare style loot or even the unique style loot bits that had some randomized properties.

    One of my other complaints is the hard limits on raid group sizes and because we have instances now regular group sizes. In EQ you could practically bring as many people as you wanted to most encounters, at least in 2004 when I quit you could. Having more people does make encounters easier but it also decreases anyone person's chance at loot. So groups will optimize themselves when they can to increase their chances of loot.

    If you combined the above with bosses dropping loot on a sliding quality scale which rewarded players completing the challenge with fewer members with better loot I think we'd have less forced min maxing.

  8. I think the real heart of the problem is the single-minded focus on raiding as the only endgame content. If crafting, trading, exploring, scouting, political maneuvering, and so on were all part of the game, then players wouldn't feel the pressure to play a certain way. The min-maxers could go raid and more casual players could run shops, explore, harvest, and so on.

    But even if the endgame is limited to running dungeons, I'd love to see them be made more dynamic (your point 3). For example, bosses could spawn with randomized sets of skills and equipment so players can't choreograph a strategy in advance. And since different skills would be useful against different types of bosses, there wouldn't be any universally "optimal" build.

  9. Min-maxing of gear/talents etc is the logical consequence of the fact that many MMOs are very stat dependent. People will always do it because it significantly impacts overall gameplay.

    The solution would be to make game less stat dependent. For gear-grind oriented games like World of Warcraft it won't happen anytime soon though.

  10. On Sunday we went rifting, doing Expert Rifts and for once I had the rather delightful experience of being able to simply add anyone who logged on to the raid. As a raid leader it was wonderfully liberating.

    Raids originally were aimed at the hardest of the hardcore but supported large numbers. If your dragon spawned at midday on a Wednesday it was a matter of grabbing everyone you could to go kill it.

    Even in WoW Vanilla the struggle to find 40 people made raid leaders broad-minded. Your best player was probably leagues ahead of your 40th best player, especially if you weren't a server first type guild. The stereotype of the afk autoshotting hunter arose because we knew those guys were in our raids not pulling their wait but it was hard to replace them and we still got the job done.

    A 10 man raid is the antithesis of this and, with Cata, is where WoW has gone. 25 mans are dying out because of the phobia of inefficiency. Raid leaders know they can find 5 good dps, but finding 15? That's a big ask.

    I'd like to see raiding that isn't number capped. So we can solve the problem of low dps by adding players not improving players.

    This is your point 4) and it's very fun for the raid leader, and might encourage more people to raid lead if they get to enable fun rather than being fun's gatekeepers.

  11. Added a 6th option. Also, Stabs, I agree that non-instanced content shines, because everybody is happy when another player wants to help.

    But you cannot produce perfectly balanced, challenging content in a non-instanced way. Or only if the content scales and in that case the additional player doesn't help anymore.

    What you can do, however, is to offer perfectly balanced content, not on the inividual encounter level, but on a higher level.

  12. I was surprised at how well the EVE Online design worked; not as well as fanbois would say but still there was not much ability to have wtfpwnmobiles. A two month old player in a small frigate has no chance to kill a $6000 super ship. But the supership might not could catch them and maybe not hit so small a target if they did catch them. Upon reflection, I think this is because it is so much easier to do this in PvP than PvE. But especially in non-random scripted PvE.

    In any case random boss abilities and attributes make it harder to optimize.

    I think the most promising thing to investigate would be to increase the dimensions. I.e., if you had ten damage/healing points to spend and you could put up to 5 in fire, arcane, nature, shadow, elemental, ... and similar for resistances. Instead of Mages always want more spellpower, warriors strength. If there were several dimensions to improve then choices would have to be made. And if the bosses had random attributes it might not get optimized as easily. Maybe sometimes 500 shadow DPS and 300 nature DPS is better than 1000 fire DPS. whereas 1000 spellpower is always > 800 spellpower.

    In terms of decreasing complexity, I see the alternatives:

    1) More choices / dimensions - there are many dimensions to improve not just spellpower. Perhaps also include tradeoffs (improved armor lowers speed) or diminishing returns.

    2) Talent trees are only things that do not affect target dummy DPS: undead, trapping, ancestral swiftness, cauterize.

    3) No talent trees: note that reducing /eliminating the complexity of gearing/specing is *NOT* the same as reducing the challenge of the content. What if everyone in the game got the same talent build for their specialization (e.g. Fire Mage)? Everyone would complain, but it probably would not change many builds for people doing heroic raids. And it would help the new/ignorant players. Sinestra would not be any less challenging if all the Fire Mages had the same talent build. It would be a less rich game; and it would skew the game even more towards twitch over knowledge but it would appease the no min/max proponents