Greedy Goblin

Friday, January 7, 2011

Why games are unbalanced?

The Pug update: Omnitron defense system is down too, so we now have 2 (+Algaloth) boss scalps:
Wowprogress says it is enough for the world position 13017. If we assume 20 raiders/guild and 4M people who bought Cataclysm, we are way in top 10%. If you want raiding, but your current "freindly heplfull" guild have problems even in HC Grim batol, or there are simply not enough healers and tanks in your guild, join! After reading the rules of course.

I wrote how terribly Tol Barad is defense-sided. I was wrong. I participated in more successful assaults than defenses in the last days (after the end of 1800 honor times). Tol Barad is still unbalanced, but it's not clear defense-sided, like Arathi would be horde-sided with 4 points, no stables.

"Unbalanced" means that "if two players/teams of equal skill and gear would fight, the one playing with the overpowered side/class would win most of the matches". Tol Barad is completely defense-owned in most realms. It's mostly offense-owned on our realm. This is clear unbalance. But how can the same battleground defense and offense-sided? For long I did not understand this fundamental game design problem. It seems Blizzard developers did not understand it either or they wouldn't make Tol Barad the way they did.

The problem is not that Tol Barad falls into the definition of "unbalanced". The problem is that the above definition is useless. The outcome of a Tol Barad (and a wintergrasp or a zerg vs protoss) battle is highly dependent on player skill/gear despite it is equal for the opponents. It means that if two equally great team would fight for Tol Barad, attacker would always win. Between two crappy teams defense would always win. In Wintergrasp a skilled defender could easily repel an equally skilled attacker, but a bunch of moron defenders would be obliterated by equal morons on the attacker side. A scrub playing zerg can easily zerg an equally useless protoss player, despite on the high-end Starcraft is a perfectly balanced e-sport.

I can give PvE example too: all the high-end players whine that survhunter and unholy DK are overpowered, while elemental shamans, arcane mages and ret paladins are weak. On the other hand if you go to a raid that utterly fails at Algoloth, you find that the damage charts are "owned" by 8K arcanes and rets, while the "op" surv and unholy are below the tank.

The fundamental problem is that the different classes and Tol Barad/WG sides scale differently with player skill (and to a lesser extent, gear). Even a monkey can spam arcane blast and click on barrage if the screen flashes. A really good player can add little difference to the monkey. On the other hand a random M&S playing surv forgets black arrow completely, clips the lock and load, never have energy for explosive as he wasted it on arcane or multi and his pet has no talents. The good player from this is worlds apart.

One must see that the problem is unpatchable:
If Blizzard would simply add a percentage nerf to survhunters to make them equal to arcane mages in hard mode raids, not only the moron, but also the casual survhunters would be significantly below their arcane guildmates.

The class unbalance is inevitable, and the way of unbalance is philosophical choice:
  • In a casual/amusement game the class skills are balanced to the median or average strength. So simply the game company gathers data of all PvE group play, for example Blizzard logs DPS of all players in HCs and raids. Then they calculate the average/median and buff/nerf to make sure that they are equal at all classes. The trade-off is that in the HC realm some classes are seriously overpowered, and their power changes significantly in every little patch. The HCs must accept in a casual game that their class can turn into seriously inferior overnight and developers don't give a damn. You can be HC in such a game but have a geared alt army or be prepared that you can be benched for a whole patch.
  • In an e-sport, the class skills are balanced to the equal performance of the top players. If one class is underrepresented in the world top 1000, buff it, if overrepresented, nerf it. The price is that at the beginner level, some classes just zerg other classes. You can be a casual in such games, but you must restrict your playing to the simple class or accept to be often obliterated without a chance.
  • everything else is a bad game
At least to map development this problem can be bypassed: make a mirror-map, where everything is equal to the two sides. Most WoW battlegrounds are such mirror-maps. Battle for gilneas is almost mirror (WW is closer to ally than horde) and in Strand of Ancients you must play both sides. Tol Barad, and Wintergrasp are not such, therefore their failure was inevitable in a game where both casuals and HCs are present. Making a balanced, non-mirror map is theoretically impossible in a non-e-sport game. If they would give a buff to attackers then in realms with some PvP guilds defender would never win. Leave it this way: on PvE and simply M&S infested realms assault never wins.

PS: attacker's guide to Tol Barad. The defenders spawn in the middle, equally far from all bases and win even if they hold one base. The attackers on the other hand has a graveyard very close to the bases. The defender must levitate and then ride to the base, it takes about 25 secs, while a defender can join a battle 10 secs after being ressed. It means that the slider does not move if the attacker:defender ratio is around 4:5. Since the attackers have equal people as defenders, if they manage to get 1 v 1 on all bases, they win.

So the key to attacker win is to make every battle very slightly winning. If you are just obliterating the defense, you are just losing the battle, because it means that you have too much people at that base, therefore too little somewhere else. The key to victory is if you attack a base, die 5x and finally when you are starting to win, abandon that base, let your less intelligent teammates finish the job and go to the next base. If you are losing badly, call for reinforcements. If the battle is even or even slightly losing, you are fine. If you are "guarding" a base with several people, you are no better than an AFK-er. Leave just lookouts who either have far sight (shaman, hunter) or can stealth who can count the enemy and call for reinforcements. If you leave significant defenses, the defenders just ignore that base and win on the other ones.


Taemojitsu said...

Counter example: the performance of a class in battlegrounds, vs the performance of the class solo.

Classes with a large number of cooldowns are not balanced 1v1 when using cooldowns against a class without cooldowns. They are also not balanced 1v1 when none of their cooldowns are up. However, in a BG with continuous PvP, cooldowns will not always be up, and performance averages out according to the types of players who tend to choose each class.

If one goal of an MMO is to provide interesting choices and varied situations for those choices to apply to, then classes and maps will not be the same. However, by increasing the complexity and number of choices, it's possible to provide similar performance curves despite wide differences, the way many distributions in nature are normal-shaped despite each having a totally separate set of often binary factors.

"Which buttons you press to do maximum damage" is one factor, but does not have to be the only one to contribute to overall damage done if survivability is a factor, and dps itself does not have to be the only measure of effectiveness for classes or specs that are seen as primary damage dealers if actions by players have indirect implications for the choices and survival of other players. The variance of performance versus skill does not have depend solely on "complexity of the optimum rotation", which means that the scaling does not have be so wide that imbalances appear at higher or lower competences of play.

Since non-damage performance is one of these possible factors however, a game design which encourages the popular disregard of non-damage performance will, necessarily, have a wider difference in variance of performance versus skill between classes and maps than a game which encourages players to take the time to consider the more complex factors of performance.

It's probably fair to say that considering these factors leads to higher long-term performance from a playerbase, while using simple metrics is more efficient for a player who is discriminating between potential group members to achieve a short-term goal. Consequently, a game culture that emphasizes short-term goals and disregards the individual improvement of other players will in turn put pressure on the game design tending towards either homogenization of choices or a reduction in the importance of factors contributing to performance that are not easily judged by a metric (..same as last paragraph..), increasing the likelihood of different skill caps as described in this post.

nightgerbil said...

So I take it you think wow is an E sport, given surv hunters are being nerfed because of what they can do in the hands of the professionals, but Blizzard appear to be /care about the way frost mages are destroying 90% of other classes in one v one pvp?

It certainly makes sense when looked at from that perspective. After the nerf breaks our high pve dps it will be interesting to see how many more hunters follow the example of the "casual pvpers" who went and rerolled mages after 4.01. Do you think Bliz would introduce a fix, if hunters stayed competitive in high end arena/rated bgs/raids, but the numbers played dropped to say <5% of total 85s were hunters? If this is an E sport, following your above example the answer would be no and that fits with history if you look for example at subtlety rogues.

Shandas said...

A very interesting read. Still, what can you do if defense zergs one base? You'd have to bring to many people, and they can just zergs the next base when you do that... Or does that fail most of the time (haven't played Tol Barad yet, the answer might be obvious).

Coldroger said...

Just a little extra help if people really want to win TB:

As a resto sham in PvP I often try to 'lead' TB. Technically I'm calling out what happens on the map, nothing more to be honest. If I see that we’re gaining a base, I ask people to move to an other base before we actually take it. And with me 2-3 other people say in the raidgroup too, that helps a lot as well.
The fact that people think M&S won’t listen is wrong. Some are just to stupid to read the chat while attacking someone, others won’t listen to people ‘they don’t know’. The only thing you need is authority.

To achieve this (towards M&S) I found that being in TB often helps quite a lot.
For example there was some guy (something with druid in his name… surprise) who was just calling everyone noobs while we tried to get TB in our hands. After 2-3 TBs someone in the raidgroup said the following: “just /ignore ..druid, if you ask yourself why, just look at his name”. And poof, he was suddenly quiet the entire fight. Majorly because the person who said it, was very often at the battle of TB, and the Moron thought he would really had gotten ignored massively by the entire raid. I think his ego is hurt now, poor guy…

Another way to achieve a little authority is having (sorry to say it) decent/good gear. People look up against others with raiding gear or epic PvP gear (even though it’s not hard to obtain).
People in greens/low blues can have the best tactics in the world, but if they get inspected by (mostly) the M&S they just tell him ‘lolz nowb’ and ignore what he says.

The last thing which had success as well, is while giving orders, put the word please at the end. People (and not only the M&S) tend to react better/quicker if you sort of beg them.
In the real world people react to that word the same way, as if you realy need them. But mostly you only say it because You are the one who benefits from it the most.
Saying please means nothing to me, but if you would have problems ‘begging’ M&S to do something for you, just make a simple macro.

Grim said...


Soon after the battle starts, the defense get initiative. As in - defense decides where the fighting will take place and how many players will be involved. If offense can react quickly and coherently, they win because of the commute time differences explained in the article.

If offense is too slow to react then they lose, because defense just zergs one base, then when they get thrown out of it, they zerg the next base.

The gray area of the capture bar is much longer than the colored areas, allowing defense to switch as soon as a base is overrun and bring another base into gray before offense captures the last one.

chewy said...

If I'm reading your meaning as intended with reference to the difference between e-sport and casual/amusement games isn't the solution to provide both maps but distinguish them between servers ?

As today we have pvp,pve,rp servers is there not a case for e-sport/casual environments the difference being the way the classes are calculated ? This would provide the environment you long for since the M&S would not want to level a player on the e-sport servers.

Ulsaki said...

Whilst it's very true that the dynamics of games can change dramatically at different skill levels, even when sides are equal, I disagree that TB is not unbalanced.

I'll discuss WG first.

Low level WG usually results in an attacking victory because the attackers tend to achieve the steps they need (controlling workshops, attacking the base), whereas the defenders do not usually go after the strategic targets (workshops/towers). At this basic level the attackers can do this with little effort or thought.

Effectively, one side does the strategic thing (albeit badly), and one side ignores it. It's no surprise that the strategic side (attack) wins.

At high level WG, where sides go after towers, workshop, and controlling important areas of the map, both are playing strategically, and at a high level, and this is where the imbalance between defending and attacking can be observed. It's a completely different situation to low level play because you have both sides playing "correctly" instead of just one.

So, back to TB.

TB favours defense because they can all rush a single base, and if they do that they will always have numerical superiority (if it's equal, then all the attackers are at one base and a single defender can take the other two back).

Any defensive strategy that involves taking on an equally sized attacking force is simply a bad strategy.

Attackers can win 1v1 battles, but if defense plays correctly, this situation never occurs. The only way to beat a correct defense is via superior skill.

In short, attackers only have the advantage if defense plays badly. If both sides are equally skilled and use the right strategy, defense always wins. And this is why TB is very unbalanced. The same result also occurs at high level and low level play, unlike WG where the game plays completely differently at different skill levels.

Pulse said...

Excellent analysis. It seems to me that the gray/coloured bar ratio could be exploited to restore balance to different servers. If the defenders win 3 TB's in a row, then the coloured portion grows (and gray shrinks). if the attackers win 3 in a row then the grey portion grows (and coloured shrinks. This would improve balance on different servers with one fix, regardless of which side typically has the advantage.

Note that this doesn't increase faction imbalances, merely attacker/defender imbalances. So the "better" side will still win most of the time. but at least the chances of a week-long defense will be less.

Blizzard did some similar mechanics in WG late in Wrath, maybe a fix like this will help balance TB.

RLWJR said...

Just a quick update on the Korgath (US) guild "PuG Inc", which is running on the exact same rules as "The PuG"......

We are currently up to 26 members. We are all re-rolls, so we are all still madly leveling and looking forward to raiding and PvP together. Some of the best experiences I have had in WoW have come from just running instances with the other guys in PuG Inc!

If you are tired of your social guild drama, are self-sufficient and enjoy playing with the same, and want to raid with competent players, come on over to Korgath (US) and message Kricknard, Raico, Zintix, Varanus or Durantula for an invite! is our small blog.

Lars Norberg said...

Your tactics for Tol Barad attack are just the same as I've been trying to convince my realm of. In Tol Barad it all comes down to numbers.

Sure, you have the skill factor; you could theoretically get a Horde team where everybody are hardcore PvPers vs an Alliance team where all are PvE scrubs in green leveling gear. But that's unlikely. Tactics should be done with the assumption that gear and skill are equally divided between the factions. And when that is the case, the ressurection time is the key to winning. The defenders simply need more time to get back into the battle, and time + people at any given base is what captures it.

It's so simple that it's hard to understand why it's so hard for people to understand.

But I think this, like all other things, boils down to something very specific about people: They only see what's in front of them. They see that there's tons of Hordies at the Slagworks, and naturally they conclude "Need moar at SW!!!1". The only real tactic they understand, is to outnumber and/or overpower. And that is not a tactic. That is to outnumber/overpower... /sigh

nightgerbil said...

"chewy said...

If I'm reading your meaning as intended with reference to the difference between e-sport and casual/amusement games isn't the solution to provide both maps but distinguish them between servers ? "

Your talking about setting up "heroic" servers here, I have read this idea before and Blizz have always refused to do it in the past. I guess because the majority of people would actually HAVE to goto the hc servers or be labeled scrubs. If the goblin logic is correct though, hmmn food for thought here.

Unknown said...

Even with your strategy TB still favors defense. Your example assumes the attacking team has scouts and the defense charges a random base. But if defense has 1 guy on every base, the classic "zerg one base then another" approach will beat your "try to balance fights in our favor" approach every time. While offense has the graveyard advantage on the flag they are fighting at, the defense have that same graveyard advantage when switching targets. If you assume it takes 25 seconds to get to a flag from the middle, then it follows that it takes 50 seconds to get from flag to flag. So, if both teams split up evenly across flags, the attackers have a big advantage. However, if the defense knows how many people are at each flag, they can just use the regular zerg strategy and force you to move around as well. Then, you are forced to play on their terms, where they have 2x of your mobility when changing targets.

Gevlon said...

@Goxy: no, from flag to flag it takes just a bit more than base to flag. Also horde have resurrection timer. So if defense starts zerging one base, reinforcement can arrive in time.

Shintar said...

I have to say I've experienced the same thing as Goxy. Flag to flag might not take twice as long as base to flag, but it does take longer and the defender has the initiative when it comes to switching targets since it doesn't matter to them which base they hold. All it takes is a couple of defenders ressing in the middle and deciding to run over to one of the non-defended bases. Since the slider bar for capture is so short, they'll usually make the attacker lose it before more attackers can move over from the previous fighting spot. Then while some attackers move over to that base to get it back, the next couple of defenders res in the middle and run to cap the third flag... and so it goes perpetually round and round because the defenders are always fast enough to cap an insufficiently guarded node before attacking reinforcements can arrive.

Anonymous said...

Also in Gilneas, while horde has to actually swim through the water to reach WW, allies can just ride over to it on "land", because of the rock in the water at the broken bridge. And the speed powerup is placed on the broken bridge, so in the allies' path.

Anonymous said...

This post makes no sense. A competent defense cannot be beaten in TB. The distance from spawn to flag is half the distance from flag to flag, and since defense only has to hold one spot the chosen spot can be called spontaneously and disable a rapid deployment of reinforcement.

The "little" time that you claim it to be different between attackers moving between flags and defenders respawning and attacking a new flag means that provided enough defenders move to a new point together, the attackers holding said point will be dead before the reinforcements arrive, and upon respawning the reinforcements will be weakening.

A defense that maintains mobility and keeps maximum travel time on the attacking team will not be beaten on a TB defense.

Your strategy seems to assume that defenders will not quickly dispose of a force that can be, at VERY best, 1/3rd of the entire force available; and even that is only possible if the defense team is picking the wrong target. When you then consider that the attacking team also has to overcome a beginning score of 0-3, it becomes an impossible feat.

The only reason that TB currently changes hands at all is because there are enough bads playing that at certain times of the day enough of them get in to lose.

On my server, horde did not lose TB more than twice until the 1800 honor change was implemented.