No one knows what the “chance of rarer materials” means. Is it an absolute increase? A fractional increase? Or does it work like Magic Find, in which it increases the value of the ‘roll’ to determine which loot table you are in (e.g. common, with no chance of rarer materials versus uncommon with a tiny chance, or rare, with a small chance).
I don’t disagree, but I’d also suggest that it’s absolutely valuable to have someone look into this and post here if they don’t think it’s behaving as expected.
If nothing else it increases the odds of an official response.
Can we please get a working Mac Client [Merged]
in Account & Technical Support
Posted by: SlippyCheeze.5483
I noticed that time of play makes a big difference. I was having a fairly good tournament until one night that I played late and lost 10 of 12 matches. Huge skill difference between teams in losses.
Advocate for slower queues, for better matchmaking. The larger the difference in MMR, the more likely you are getting the player-desired “fast queue” rather than “fair queue”, which sucks, but … it’s a factor of life.
At least you can take solace in the fact that you are gonna be placed on the “wins because high MMR” side as much as the “loses because high MMR” side.
This is not the point. The system means two people with the same skill can need 10 or 500 matches to get to their MMR and some might give up or never be able to play that much games in a season. For people with bad luck it becomes grinding while for others it´s easy to enjoy their right position from start and both have the same skill.
If it takes 500 matches to get your MMR settled, something is dramatically wrong.
…but, really, I think that the issue is that you don’t understand how quickly Glicko-2 MMR converges on a solution. The number of placement matches, 10, wasn’t actually picked at random, y’know?
If you wanted to take a guess, guess how many matches it will take on average before you are in the right ballpark of MMR? I’ll wait … and done. Did you guess 10? I certainly hope so.
It’s also notable that 89 matches is the “magic number” for convergence in a 5v5 competition. So, 500 is only about 5 or 6 times an overestimate.
Finally, do you really think you would converge more quickly if everyone started at MMR 350, full stop? Alternately, do you have some other proposal for what might reduce the time to accuracy as a starting point?
The only point I agree with you on is that, yes, some players may not continue to play if they feel their MMR does not accurately reflect their skill. We have already seen plenty of support for that in people here, talking about “ELO Hell”, etc, because they were used to a system that did not reflect skill, and now it does … they don’t get the rating they “deserve”.
lol turrent engi, nice. agree for the most part but it still seems like glicko cant handle the complexity of gw2 and mmorpgs in general, namely enemy team buying a team mate to forfeit (or a solo actor that’s salty), people running a build they don’t know how to play, and bad rotations. things like this all contribute to a loss that a player has no control over.
See of course i can agree with you that the system not being able to correctly give each player there “perfect rating” but im pretty sure we can all agree that it is giving a near accurate rating.
It’s always entertaining when someone says that, then throws up examples of CS:GO or some random MOBA as being so much better … when they use the same system.
Really, though, most of the problem people have with MMR is that they see one instance they don’t think is fair, and extrapolate to MMR being entirely broken. Because they don’t imagine that one AFK player could get just as easily placed on the enemy team next team, leading to exactly the opposite layout.
I won’t pretend to have done enough thinking about hard resets and the inner workings of a solid Glicko system to talk about those things, but I would like to see the up-and-down of that number reflect actual performance, rather than just the win or loss.
It’s worth noting that it is not just “thinking” that backs Elo and Glicko, it’s actual hard math. Probably a mostly pedantic point, but … it’s worth noting that these have had significant thought and math and refinement applied in the real world.
Anyway, the key issue with your idea is that it builds a metric into your score changes. Once you have a metric, people play to the metric — and you get what you measure, nothing else.
If taking top stats is worth something, you get builds for it: the “top DPS” glass cannon build, the “nothing but stability, and resurrection” build, all of which contribute less to winning the match, but more to the personal score of the player involved.
The only thing that, ultimately, reflects skill without forcing particular playstyles on people is … win/lose. Not how you get there. Which is why these systems pretty much all end up falling back to it.
Also useful to remember: they also result in good matchups, using only that metric, so this is a tested and valid way to measure skill.
Also also useful to remember: no matter which game you point to as having a “better” MMR system, it’s almost certain they use the same basic math that GW2 does.
You people have such a short memory, DH for the first 2/3rds of 2016 was deemed the worst/most useless profession in pvp and was the only profession to receive a F ranking from the ESL player poll
I’ve been especially entertained to see both “warrior OP, nerf them” and “thief OP, nerf them” threads here, concurrently, both citing the utter impossibility of the other profession ever beating them.
When people say “the cloud” they really mean “other people’s computers”; sounds like what happened in your case is that one of the dozen or so routers or computers between your client and the ANet servers had a problem.
Then the problem was fixed, but whatever third party unrelated to either of you that runs it. Which was this strange, “nothing changed on either end”, problem that just went away.
Please dont call it skill rating its so inaccurate. It could be called skill rating if it represented YOUR SKILL but it doesnt. 20% of the so called skill rating is you while 80% of it is random 4 players you get.
In each individual game, sure, it does — and your win or loss is determined in the same part by the five random players you face off on the other team. Over many games, it reflects your skill.
You seem to be making one of the classic mistakes, which is to look at a single event, then complaining that something that measures skill over many games is inaccurate.
You kinda forgot that the random 5 players that you are facing can have up to 600 more individual skill rating than you. Which can lead to – If u cant beat someone with 500+ rating, you wont progress.
It can be called skill rating if the matchmaking is perfectly balanced and that will never happen in gw2 due to population problem.
To your, so called classic mistake:
Does it make sense that my skill rating that reflect my skill ( thats what you think i guess) will go drastically down in over a day. We are talking about 250 rating here. If the skill rating perfectly represents your skill that situation should never happen unless my playstyle went really downhill or the matchmaking of the game is bad.Hmm let me guess which one will it be….
I didn’t forget that the random five players on the other team can have 600 more MMR than you … just like I didn’t forget that the players on your team can have that 600 more MMR.
The problem with “ELO Hell” or “MMR is inaccurate because team composition” as ideas is that every single thing you say is keeping you down because it can randomly happen to one team or the other … applies equally to both teams.
You are also mistaking “MMR is accurate” for “MMR is perfectly accurate, all the time”, which your argument would break, yes. However, why did your skill go down 250 points in a day?
- you had an off day, and played really badly today
* you have high volatility because you have not played in a while
* you have high volatility because, despite randomly matching with players better and worse than you, you still lose when expected to win, and win when expected to lose
* your MMR is higher than your skill, and it’s still moving toward the correct value (which takes ~ 100 games)
* you switched your build or class to something else, and suck at it compared to the one you played before (right now, at least.)
* you duo-queued with someone with a very different MMR, and that made it harder for the MMR algorithm to home in on your skill
* you played at a super-quiet time, so a broader distribution of MMRs in the games, leading to increased volatility
Ultimately, though, your MMR going down means just one thing: you lost matches you were expected to win, and didn’t win matches you were expected to lose. The people you were randomly matched with played better than you did.
Going down a long way means one of two things: either your personal volatility is high, or you consistently played badly despite it being low. Either way, your MMR was lowered to reflect the quality of your play.
Sorry, I know that’s not nice to hear, but … ultimately, your MMR will wiggle up and down a little around the target point, because it is always an approximation of skill. If it changes dramatically, though, that’s about you, not the game.
Since 3 different teams work on the Living World Story maps, and rotate through the cycle, I would think the team that worked on Bitterfrost Frontier had the most time to complete their map.
Everyone knows that assumptions about why your favorite bug happened are more significant than realities of development.
I usually see that right after the game has loaded. After playing for a while it goes away although that is still only in some locations.
That exactly describes the file being cached in memory — slow to load the first time, but fast afterwards, until you hit something not yet cached which causes it to be slow.
Except that is like that even when going to places I haven’t been to in weeks.
Probably more likely that at startup there are many things that need loading and these things have an incorrect priority set on them so they are being loaded later than intended.
…or the textures and models are also used elsewhere in the game, so are preloaded? Anyway, it’s just speculation, ultimately.