Guild Wars 2

The main argument for dropping esports support is that right now, it’s a money toilet. Flushing the money doesn’t benefit the game or the player base as a whole.

It would be one thing if the PvP scene was healthy or growing, but it’s dwindled ever since HoT released. People quit watching because HoT power creep made the game boring to play and even more boring to watch. You can’t grow or maintain a community if the game isn’t fun. And when everyone sees how awful PvP is, it discourages them from trying it.

And it didn’t just shrink in viewers and casual players. Top players left in droves, and the only “top” players left are the ones who made money from playing. Why have esports if there are so few players to compete?

The money has to come from somewhere. In addition to the prize pool, you also have to consider the money to host events, for people to staff them, airfare, etc. You can argue that it’s the marketing group’s money, but that money is fungible and could have been allocated elsewhere. They could’ve done more ads targeting PvE for example. Or it could have been allocated to hiring a few more developers or improving the communications team.

This season’s rating-based system is far better than the awful progression-based system of the past year. It still has a few issues though:

• Decay The implementation isn’t very good. It should only affect the top tiers, stopping once decayed reating has reached a set number. Decay should be based on amount played in a week rather than consecutive days not played (e.g. playing 20 games one day and 0 for the next 5 should not cause significant decay). Some part of decay also needs to be permanent.
• Leaderboard The very top is sparse and volatile. Having a leaderboard based directly on rating will show the same volatility. A better system for a leaderboard could probably be created, but isn’t a huge issue as the people at the top should be there. A better decay system will address a surplus of alt accounts.
• Rewards Ascended shards of glory need to be smoothed out over a longer period of time, with no cap. Repeating the final chest should award some shards.
• No Team Queue The ability to queue as a full team needs to return in some form.

NOT problems:

• Placement Matches Placement matches just mask the relatively dramatic rating swings in your first few games and serve as a way to convey that they may not be the most evenly matched games.
• Rating System in General It’s working fine. People who complain about their rating are simply not as good as they think they are. Many people have shown that’s possible to go from bronze to platinum in a reasonable amount of games.
• Profession Stacking Players should communicate and adjust their team before the match starts. If stacking a profession is too powerful, then that should be addressed with balance changes.

Magnito.6187:

I actually remember myself almost always teamqueueing back then, just because it was better pvp. (and that wasn’t always with a team myself)

Same. The old solo queue was complete crap. It was full of the “solo queue heroes” who didn’t care about working as a team to win the match. Meanwhile the team queue, even if you queued by yourself, was people who realized they needed to work together.

Solo/Duo queue just encourages the “solo queue hero” mentality: try to 1v1 everything, put winning fights over taking/holding nodes, ignore the rest of the team, etc. Then they turn around and blame the rest of the team for their loss.

The reason solo/duo queue won a simple majority was because players blamed playing enemy teams for their losses, rather than their own lack of skill and teamwork. It’s the same people who now whine that duo queue and class stacking are unfair.

You’re probably seeing people with temporary decay.

So what do you think MMR is if it’s not your rating? The matchmaker doesn’t make up a composite number based on rating.

None of the above.

The real reason GW2 e-sports died is HoT power creep. It removed skill and replaced it with spammable passives, which created unkillable monsters and stalemate games. ANet has chipped away at the power creep, but it’s still high. Most of the competitive players who couldn’t make consistent money from tournaments simply quit.

And just like the HoT power creep sucked to play, it sucked even more to watch. Viewership plummeted since every match was the same stale comps over and over.

Doorbreakers have to have a strong down side to encourage risk vs. reward. Much of the current problem is that you can easily keep doorbreakers alive (get a support next to them, they live for a long time) and go through both gates. The best way to deal with this is to prevent 5 supply of doorbreakers from getting through both gates. Each one would need to have higher gate damage though.

In comparison, archers are the slow and steady option, but if you lose one early, it’s not as large of an impact. Archers also need continued babysitting, but it’s not as intensive as with DBs. DBs you have to watch closely, but only until they do their job. Additionally, they keep going, so a single archer can do a lot of work if left alone.

For a comparison of damage-break or CC-break, look at communes on Temple of the Silent Storm. They’re damage-break and lead to clutch fights and important decision making (do you commit to the buff or troll it and take points, negating the buff’s benefit). Right now, someone with pulsing stability just starts the channel and can’t be stopped. Any build which which is going to channel has enough defenses to tank damage during the channel. And if someone is constantly poking you, they’re not doing something else on the map. Keep in mind that I did say channels would need to be faster in order to compensate.

1. No. Part of competitive GW2 is being able to play a few builds well.
2. No. Not enough players to make full team queue work outside of prime time. Full 1-5 player queue will come back in S6.
3. Core isn’t abyssmal; it’s pretty good overall. The problem is HoT specs are far too strong.
4. Agree
5. Fix #3 and you don’t need this.
6. Ranked rewards need to spread out more. They’re all early on and then substantially less afterward.

Expected Results

• PvP population is healthy, though that may be bias in people who took the survey.
• PvPers play because they like PvP. Having rewards doesn’t hurt.
• The pip system sucked.
• Many players overrate themselves.
• Balance and diversity is a problem, which are primarily due to HoT power creep (class stacking is a symptom of this).
• Current MMR system works fine.

Tier Distribution
The biggest surprise to me was the tier distribution. It should be Gaussian. The assumption was that it centered on 1200, but it’s clearly not (even accounting for bias in survey takers). I’m wondering if the rating scale is actually centered on 1500 (most Elo and Glicko systems use that value) and ANet’s soft reset toward 1200 actually skewed people downward. If that was the case, seeing a roughly even silver (centered on 1200) and gold (centered on 1500) makes sense. I hope it’s not, since doing a skew left of center will make it harder to reach legendary in each successive season.

Leaderboard
Despite all the complaints, the leaderboard is doing exactly what it’s supposed to be doing: displaying the people with the highest rating in order. Top players with alt accounts should have high ratings on those alt accounts, even after few games. Rather than kittening progress on alt accounts, it seems like a complementary system to rating is needed for the top players.

@OP

Your video is terrible. You don’t understand PvP or stronghold’s problems. Your video was just a facade for a GW1 GvG love-in. In fact, you never presented ANY real ideas on how to fix stronghold. You just said “make it like GW1 GvG”.

Let me help you with a real analysis…

Conquest
Conquest has huge amounts of depth, map awareness, and decision making, despite it being fairly simple (even you said it, OP).

• There are four popular maps. Temple of the Silent Storm is well liked and produces some of the most dynamic and interesting matches.
• Foefire isn’t as overwhelmingly popular as it seems. When players rank their favorite maps, Foefire exhibits a love/hate pattern. With the map selection system, when around half of the votes split between 2 of the 3 other maps, of course Foefire looks to have amazing support.

GW1 GvG Problems
GvG and even the player run ports of it into GW2 exhibit the big flaws with GvG as a competitive PvP System. That’s why ANet didn’t port it into GW2.

• GvG causes players to congregate in one spot. This leads to “staring contests” followed by either AoE spam or massive concentrated spike damage and CC. Look no further than GW2’s Courtyard map to see how it plays out.
• Large groups hinder the GW2 action-based combat style. GW2 combat centers around knowing when and how to avoid powerful attacks and when you use powerful attacks of your own. When many abilities are being thrown out at once, it destroys that decision making.
• GvG snowballs once a single fight is won. The losing side is pushed back and rarely recovers. In choosing conquest, ANet was looking for a mode where winning a single fight wasn’t the end of the match. Conquest’s design makes larger advantages from winning a fight harder to maintain.

Why the Popularity?
So why do Foefire and GvG seem to be popular?

• Mob play caters to casual players and emotion. You get to see lots of numbers. With more actors in one place, poor play is marginalized and less obvious. When it snowballs in your favor, you feel dominant. “I won”, but “it’s not my fault” when we lost.
• Map design facilitates the previous bullet.

GW2 uses WvW (or at least tried to) to cater to this audience. Desert borderlands failed not because of mechanics, but because the place was a maze where it was easy to fall to your death – more time running around doing nothing instead of being in the ball.

Stronghold Problems
The single biggest problem with stronghold is that all the design elements encourage a playstyle of a PvE race and discourage fighting players. Your PUG may not notice, but any moderately organized team does. You never even have to control territory.

Stronghold Fixes
Fixing it completely will be an incremental approach, but this is the bare minimum:

• Doorbreakers die after one successful hit. Now you can’t carry your initial supply through both gates; players now have to get supply and they’ll fight over it.
• Damage breaks channel; reduce channel times. It’s too easy to stability channel things and completely ignore players. The trebuchet removing all stability was a band-aid, and a terrible one at that. If damage breaks channels, you’re forced to win the territory first.
• Archer gate damage buffed. Archers continue on their path after casting an attack (no chasing). They’re useless now. Players can kill guards just as fast and archers barely scratch gates, even in large numbers. Archers are too easily distracted and due to “stop until target dies” mechanics, they can be distracted indefinitely.
• Mist Heroes buffed in their primary role – getting into a base and killing a lord. They suck as a comeback mechanic. The time you wasted channeling it and going back to a fight could have been better spent in the fight itself. Increase their move speed and gate damage at a minimum.

Sinmir.6504:

Having Stronghold back. If people want to play that mode it doesn’t harm those playing Conquest.

Nope. You can’t have a ranked system where some people don’t meet others because of mode selection. If it came back, it would have to be a map choice alongside conquest maps.

Problem is, stronghold is terribly implemented and was never fixed despite mountains of player feedback during beta weekends. It’s just a PvE race.

Title is nonsense. Glicko2 may not be optimal, but it’s better than any system yet devised.

Score differential isn’t a great indicator of how close a match is. Winning a team fight can cause a shift of 100 points, but this can swing back in another direction later (just look at some ESL matches). Strong secondary mechanics like on Legacy of the Foefire and Temple of the Silent Storm further muddy the waters.

This thread is so useless.

If you have “bad” stacked classes, then one of you should be switching if you really want to win. If a class is stacked because it’s too good, then it needs nerfed. There is no need to restrict class stacking.

Your tier pretty much means nothing except for legendary and maybe platinum. Stop worrying about it.

Sooloo.1364:

Thing is how would you know how good they are, why do you keep making asinine assumptions about something you know nothing about. When 3 of your team are playing to lose you are not going to carry that, and nobody cares if a pro player can do it. That proves absolutely nothing. Until you yourself can go through what they are going through you have no right to tell them where they belong.

So if this “trying to lose” is so rampant, it would affect both teams and your rating gains and losses would end up being roughly even, no?

Or you’re just not as good as you think you are.

The people who have top rating with just a handful of games are smurf accounts of ESL-level players. Having those ratings isn’t wrong.

Malediktus.9250:

So should I not play against better people if I win many games in a row to prove myself?

Yes. And the system does do that.

But I think my rating is moving slow because I already played 294 matches this season, so the algorithm considers it as settled.

It only stays settled if you perform to the expectation of your rating. If you have unexpected outcomes (such as defeating people equal to or higher than your consistently), your rating moves up and the amount it increases with each move and the amount it moves each time will increase slightly.

Keep in mind that it’s unlikely for someone’s performance which is “well known” to change dramatically (multiple hundreds of points) overnight. That’s why the system doesn’t all of a sudden bump you by large amounts; instead it tracks your performance vs. the expectation, but also takes into account that it is less sure of your true rating.

Malediktus.9250:

I played over 10 matches today (soloQ) and won 80-90% (try harded a bit) of them. But my MMR barely increased, it is really frustrating.
In old seasons I would have gone up half a division by this because of win streak bonus.

Who you play is just as important as the outcome. If you beat people with lower ratings than you consistently, your rating won’t (and shouldn’t) go up much. If you consistently beat people equal or above you, your rating will rise.

One more

Good

• Smooth out ranked-only rewards – in particular Ascended Shards of Glory. They should be rewarded for continued play in ranked, not stop after you reach some threshold.

Rashy.4165:

The win/lose streaks should effectively change a player’s volatility to allow for quicker changes, before settling again, but only for the truly anomalous cases where placements put them either too high or too low, and for any player who surpasses their true rating significantly.

This is basically what I was referring to. If you’re getting a win streak like the OP indicated (and it was against even opponents), he’ll start to increase faster. But like you said, such a huge shift is unexpected and should stabilize quickly.

The point is that Glicko2 does account for anomalies and that a special “win streak bonus” is not needed.

@OP

You can’t just pick stuff from the forums. A lot of it is the same nonsense spouted from people who can’t accept that they aren’t as good as they think they are.

Good Ideas

• Strong, exponential penalties for repeated DC’s. Restrictions lasting days for repeat offenders. If your connection sucks, you need to think about not playing ranked.
• Invalidating matches for an early DC. Anyone in the DC’ing player’s party gets a loss to prevent abuse.
• Fix bug when you win with a DC on your team and it counts as a loss.
• Better tracking/reporting for when a match doesn’t count. We have constant forum posts where ANet explains why wins + losses != games played.
• Better decay system. Decay doesn’t need to exist at lower ratings. At higher ratings, decay needs to be permanent and require more than one game every three days.
• End-of-Season lock. To prevent smurfs from flooding the top of the leaderboard on the last days of the season, lock leaderboard access a week or so before the end so only players who have completed placement before then can appear on the leaderboard.

Bad Ideas

• Split solo and team queue. Population isn’t enough.
• Replacing current rating/division system.
• Class restrictions. It will hurt queue times and lead to worse games. If one class is being stacked for an advantage, it needs fixed with balance changes. If stacking it is seen as a disadvantage, then players should be swapping for a better comp if they really want to win. Otherwise, they should lose and lose rating.
• Gating ranked access. There’s already a PvP rank restriction to ensure a base knowledge level. After that, let rating dictate matches; that’s its purpose.
• Solo/Duo only queue. Teams need to be able to compete in a team game. Anyone blaming duo queue for losses is making excuses by people who can’t be bothered to function as a team and should be rated lower.

Debatable

• Tweak rating adjustment for win/loss. ANet hasn’t stated exactly how they do it (there are a couple options – some better than others). However, no current rating system in any game excels at rating people in random teams. Coming up with a better system requires a lot of research, and ANet could probably spend their time more productively elsewhere.

So anyone going to provide a real mathematical argument? Or just continue to make kitten up?

Please take your baseless whining to one of your other whine threads.

By the way, the first step to improving is admitting you’re not as good as you think you are.

vove.2768:

Are you for real? GW2 uses rating system invented for chess which is a 1v1 game? xD Anet keeps on surprising me.

So what system would you propose? Rating systems for competition draw all their roots to chess rating systems. And many games use Glicko for rating systems.

JTGuevara.9018:

Of course MMR hell exists! Nothing has changed!

I see your lack of skill in game is matched by your lack of knowledge outside of it.

S1-S4 you were matched within your pip range. You also had safeguards so you couldn’t fall below certain pip amounts. This caused people to pool at these pip thresholds. “MMR hell” happened to a few people who couldn’t escape this pool and had their rating dragged down, making it harder to escape (I explained it in more detail earlier in this thread).

In the S5 system, there’s no safeguard against going down in rating. Consequently, there’s no forced pooling of players by the system. The matchmaker also isn’t artificially restricted by pip range so you’ll have matches with people near your rating most of the time.

SneakyTouchy.6043:

No it’s not, that’s not how Glicko2 works.

Care to point out where I’m wrong?

Win rate alone doesn’t dictate rating. Who you win and lose against is what’s important.

This is already baked into Glicko2. If you’re consistently not performing at expectation (either above or below), your rating will change faster.

Her point pertained to middle of the pack, not the very top. Decay should combat the 10-0 people and force them to play more games (decay system could be better). My concern is someone doing that at the very end of the season.

Please stop spreading misinformation. Understand things before you made dumb posts.

Win ratio is not the metric used to track rating changes. Who you defeat or lose to is what matters.

You continue to earn rewards once you’re at your proper rating (you win about half your games), so there’s no disincentive to continue playing, as you claim. You can also learn to play better and move up. A visible way to track skill evolution is huge.

Your only valid point is the last one. DH has too low of a skill floor, so at lower skill levels, players do disproportionately well.

Crinn.7864:

Glicko2 definitely predates 2013 since there are google hits for it on pages from 2010. 2013 was the last time it was revised. (according to Glickman’s official site)

As an aside I’m rather perturbed that the glicko Wikipedia page only has two references and one of those is a forum post on these very forums.

Go straight to the source: Mark Glickman
http://www.glicko.net/glicko.html

Glicko2 had some issues with the volatility update calculation which were revised in 2013 to their current version.

Glicko2 wasn’t finalized until 2013.

Gwaihir.1745:

Why should you be given a second chance when you failed to evade it the first time?

1/4sec cast time is not reasonably avoidable. The only way to avoid it is to guess at when it will be used. It’s unblockable, so you can’t stop it that way either.

Gwaihir.1745:

Exedore.6320:

So what do you expect? How many games is “enough” to be at the top?

I never stated an amount to be at the top. If you cant see that someone holding there at 15 games is smurfing that’s another issue. I suggest that players simply not be allowed to place into anything above t1 platinum.

So 15 games is not enough according to a number from your posterior. How many is “enough”? I’ll help you out: you need to look at how many games it takes for rating to reasonably converge at the true number. Now go do some research.

Further, is it wrong for players to have ratings at top if that rating is truly reflective of their skill?

@kdaddy.5431

It’s hard to get a good picture of what’s happening without all the information together. There are also an unknown with how ANet implemented Glicko2 (who’s rating sets are compared to determine gain/loss – individual vs. each individual, individual vs. team, team vs. team).

If I were ANet, I’d look into your specific case a bit more, but at this point, I don’t see anything extremely wrong. Some of it seems to be fluctuation at the far end of the distribution from low population.

Gwaihir.1745:

Currently 3 of the top 5 are under 13 games played. Only one of those guys really belongs there at 120ish games played.

So what do you expect? How many games is “enough” to be at the top?

Crinn.7864:

Trueskill is a evolution of glicko. Glicko’s main feature is the use of deviation, where deviation is the matchmaker’s uncertainty that it’s at the right rating. Trueskill uses deviation in the same way, but Trueskill is Bayesian while Glicko is frequency.

Pretty sure both use a Bayesian distribution mechanic, but applied in a slightly different manner.

The other big difference is that glicko was designed for two players, no more no less. Adapting glicko to 2+ player games results in the algorithm equivalent of a duct tape fix. Trueskill was designed with arbitrary teamsize in mind.

In some cases, Glicko’s “duct tape fix” does remarkably well compared to Trueskill. To the point where there isn’t much of a difference. In Microsoft’s own testing on Trueskill, it had trouble in 4v4 games when compared to Elo (not much of a difference).

Also, a lot of these comparisons are against Glicko, not Glicko2. Glicko2 is an improvement over Glicko and did fix some mathematical problems with the model.

I’m not saying Glicko2 is better – but I have yet to see any paper or analysis which shows either Trueskill or Glicko2 being significantly better. All the Trueskill development analysis was against Elo.

GreenNekoHaunt.8527:

Exedore.6320:

“MMR hell” is an invention of players, so it won’t be in any real analysis. MMR hell is from matchmaking (not rating) and only exists because of limitations in the matchmaking pool.

So the actual problem is the low amount of PvP players queuing at the same time and the match making system / rating system implemented wasn’t designed for a) group play b) this low amount of people?

No. MMR Hell happened in S2 and S3 because of hard divisions in combination with pip restrictions which caused extremely low matching pools.

In S2 and S3 (and S1 and S4 to a lesser extent), you could only be matched with people in a similar pip count. Within that pip count, it would try and find similarly rated people. You progressed pip count by wins and regressed with losses, but you couldn’t regress back over some boundaries. This caused players to pool just over the boundaries – the biggest pool being the bottom of ruby and bottom of diamond being second biggest. The weaker players were the ones getting stuck. But if you were a stronger player and hit some bad luck, your rating would drop and you would constantly be matched with this ever-growing pool of weaker players. Since progression was based on winning significantly more than losing, it became very hard to climb out of this pool.

S5 doesn’t have any of these hard boundaries, so MMR hell doesn’t exist.

Moral of the story: Progression-based rating systems (move up with wins, down with losses) don’t work for competitive PvP.

Archon.6480:

Next time you go on an interview. Tell them experience is completely irrelevant. Heck, why not try out for captain of the nearest professional football team. Tell them experience is irrelevant, when they ask why they should consider you.

Employers don’t really look at years of experience (what you’re asking to do); they look at your record and current skill set. A sports team doesn’t hire a mediocre player just because they’ve been in the league for a while.

Agree. Minimap could be a lot better.

1. No. How many games you played has no correlation to skill.
2. See #1.

I’ve only played WoW recently, so I can give some answers there.

In WoW, you have a large casual PvP audience in normal battlegrounds. Winning or losing doesn’t mean a whole lot and there’s not matchmaking to it. For the ranked portion, you have arena and rated BGs, both of which require you to queue as a full team. If you have a full team, players don’t give up individually and are more likely to try and improve in order to continue winning.

Any time you have a partial team queue, you start to see the behavior you see in GW2 ranked games.

You say you have no problems with the system, but then you complain about getting matched with certain people by the same system. Can’t have it both ways.

Or… Or…

ANet lifts the stupid queue size restriction and you can queue with a whole team and not have to worry about people you don’t like.

So in that latest screenshot, what was your rating change? I’m guessing it was single digits.

Shadowstep.6049:

You are so wrong on this one. You are also defending DHs climbing ranks while being terrible lol.

Thief archetype attracts players who think they’re better than they actually are. I’d say only about 25% or less of the thieves I run into actually play it well. Too many keep trying to get a kill on sometime over and over instead of giving up for a better target.

I’m not defending DH’s either. I’m simply stating that class restrictions are not the way to address balance problems. The DH problem is a low skill floor which allows many players to spam and perform disproportionately well.

@kdaddy.5431

Was your friend still in placement matches when this happened? I was doing placement with a friend and we had a clearly mismatched player one game (solo rushes Foefire lord at 100 point mark and dies). I haven’t seen anything that bad since getting out of placement matches.

Top players lose a lot in a loss because they’re rated so much higher than their opponents. This is offset by them winning many more games than they lose. The extremes of the rating system will not be 50% W:L. When you have no one above you [or below you for the very bottom] and few people near you, the games which can be created for you favor one outcome heavily over the other.

In your player example, he could have had decay, which prevented him from showing up in the top. Each game played removes 100 points (1 day’s worth) of decay. So he might have earned 5 points for a win, but gone up +105 with decay removal.

Vykoz.7803:

You prolly remember class stacking in pro-league when they were playing to win?

Yes. The restrictions were put in place after the June 2015 patch (pre-HoT trait re-work) when d/d ele was dominating every other build. Such restrictions never existed for the 2 years prior nor did they need to exist.

The class restrictions were put in place to cover for awful game balance. The leagues needed to react quickly – must quicker than ANet has shown to be capable. However if balance was reasonable, the restrictions would not have been necessary. Once something is put in place, it’s hard to remove. With balance being questionable ever since June 2015, especially with the addition of elite specs, no one has ever thought about removing those restrictions. And pro-league is effectively dead now.

Why should it be more fun for everyone else?

There are two competing complaints:
1. <build> is overpowered and stacking it is the most reliable way to win.
2. <build> sucks when you have duplicates.

The first is a balance problem. It happens when a build is good at too many things without any clear weaknesses. It can only be addressed by hard caps (which is difficult for PUG play without hurting queue times and match quality). But it should really be addressed by out-of-cycle balance changes. The second is more about player skill and perception, but can be overcome with switches. Neither is adequately addressed by a soft cap.

Crinn.7864:

The TrueSkill algorithm is superior to glicko. It approaches accurate rating much faster than glicko, and it accounts for the fact that team are comprised of multiple differently skilled players. TrueSkill does not have any kind of MMR hell.

Source?

Everything I’ve been able to find doesn’t come down definitely for Glicko or Trueskill. In fact, Glicko and Trueskill are very similar in concept. Trueskill has advantages for certain applications (more than two teams in a game, lots of draws), but neither applies in the case of GW2.

“MMR hell” is an invention of players, so it won’t be in any real analysis. MMR hell is from matchmaking (not rating) and only exists because of limitations in the matchmaking pool.