|
I dont get it. I dont understand what are they doing, what are they thinking and what are they planning.
The matchmaking system is not as terrible as people said, but it was designed so poorly toward the 2 spectrums of the ladder scale.
Few days ago I posted this:
http://www.teamliquid.net/forum/viewpost.php?post_id=19660010
On September 03 2013 04:48 NB wrote:Ok, havent posted in this thread for a while so here i go: WHAT THE FUCK does new players with less than 10 hours into dota2 doing in the fucking VERYHIGH bracket games? Seriously, the last 2 games of mine i got players who just brand new... And these are supposed to be top tier games as some others players actually has 2000-3000 hours invested in. First game: + Show Spoiler +http://dotabuff.com/matches/2964591972nd game: + Show Spoiler +http://dotabuff.com/matches/296496779Like... its not even party matchmaking bc in game 2, all the PL talked was chinese and the pudge keep asking him if he speak english. So in a sense, the dota2 matchmaking is failing miserably by ruining both veteran players experience and new players experience. I am seriously shock. You can notice how this is not well written bc im really.... out of words. I really dont know how to describe what im feeling right now: disappointed, bitter, anger, disbelief,.... And both of them are veryhigh games.... I dont even...
At first I thought its just a rare error, a small flaw in the system and complained about it. But apparently this issue has been going on repeatedly since before TI and may be date back years since they last fundamentally change their philosophy design on the match making system.
Surfing through the absurd, poorly moderated dev forum Misc. section. You would expect several cases of QQ complain posts but occasionally you could spot some people who were truely concern about the game and know what they are talking about. Flaws, error, bug, poor design, you name it. But only until today I stumble upon a case that I, myself, couldnt justify by any scientifical mean.
Take a look at this game:
![[image loading]](http://i.imgur.com/ZVPCWpg.png)
http://dotabuff.com/matches/298665565
Here are the profile of 2 teams, with myself included:
Radiant:
+ Show Spoiler +Invoker ![[image loading]](http://i.imgur.com/Ml3cmnU.png) ![[image loading]](http://i.imgur.com/4l1TzZp.png) WR ![[image loading]](http://i.imgur.com/r196rxj.png) SF(Me) ![[image loading]](http://i.imgur.com/EZZiucf.png) Rubick ![[image loading]](http://i.imgur.com/KgmQ0Yd.png) Furion ![[image loading]](http://i.imgur.com/GhLoHcx.png)
Dire:
+ Show Spoiler +
No I know where you are about to go with this: but... but... number of win does not represent the MMR NB!!! Well shit, this isnt a problem with MMR anymore, its a problem where they are straight up matching new players with veteran and call it an even match.
As you can see, by the rough data those picture represent, both teams should have 'even' team MMR as well as 'even' lowest bound MMR which they recently added as a condition of the match making. My issue is that how does a player with 8 games in his account, 50% winrate, got matched into a veryhigh MMR with other veteran players? Isnt the matchmaking supposed to prevent this from happening?
As a veteran, it is really annoying for me seeing a new players in my game. Its not about winning or losing, its about how me or any other could exploit them to the finest details. The MMR supposed to PROTECT those new comers from players... like me. We are veterans, we play differently, we act differently, we even use different languages to communicated(if you have less than 10 hours in dota2, the chance of you not knowing Clinkz others nick name is pretty damn high). The game will be extremely enjoyable for both parties. We are not talking about 1 team robbing the fun from others. We are talking about games where its so one sided the winning team just want to get it over with and hope to get a better game. And dont even mention the losing side, no one like getting an undeserved lose.
Now the chance of that account being a smurf is there. I mean, lets be real, he does pick SF/pudge/invoker. But in what universe does a smurf with 50% winrate made it to what supposed to be the top 1% of the player base? Is 5 millions people just that terrible or Valve matchmaking has a problem, i will let you decide.
https://twitter.com/LuminousInverse/status/374606837137416192 @LuminousInverse I don't mind waiting for 30mins for a good game instead of wasting 30min of my time where mm fails @dota2 + Show Spoiler +
I think Lumi sum it up best. People would rather play with bot or even not playing than getting an uneven game. Who would enjoy hitting their head against a wall repeatedly? The matchmaking is the CORE of the user experience in DotA2 and as long as it is not functioning properly, I think the game will not be able to grow up to its full potential.
If you are a Valve dev and reading this, I just wana say that im appreciate your time into reading my complain but YOU need to learn from Blizzard wreck train: start communicating openly with your players, your communities' representatives(pros?) and listen to the (filtered)concerns. There will be constructive feedbacks if you wish to listen them.
P/s: also, as i demonstrated in this blog, hiding player win count does nothing. If i want to QQ about something, I could always check out people profile and their hours into dota. If somebody has their profile set to private, I could easily assume that they are experienced enough to have more than 10 hours into the game.
 
|
im not sure, but ive never experienced having a total noob in my games, yes there appears to be way more shitty players in top10 pages these days, but ive never ever had someone "new"
seems like a failure in the smurf detection somehow i suppose.
|
I've be lead to understand that if you stomp your first couple of games it will shoot you very high very quickly. I can see why though, you have to ask, who would your rather protect? The newbies who have to play against a pub stomper? or the veterans who have proven that they will press the next match button?
Idk though, I suck at dota2 so there's always someone floundering around. Usually me :D
|
The "simplified" valve match finding for Counterstrike-GO is annoying too!
|
It's obvious all of those with few hours are playing on their alt accounts. Perhaps part of the match making process involves moving those that do fairly decent as they start out to very high and if they can't handle the heat, they will get pushed down.
|
On September 05 2013 10:53 JumboJohnson wrote:
I've be lead to understand that if you stomp your first couple of games it will shoot you very high very quickly. I can see why though, you have to ask, who would your rather protect? The newbies who have to play against a pub stomper? or the veterans who have proven that they will press the next match button?
Idk though, I suck at dota2 so there's always someone floundering around. Usually me :D
there is a possibility that in game performance could affect your MMR BUT we are looking at a 50% winrate player with less than 10 game. He was 4-4 going into this Very High match and by no reason he belongs here.
Look at sc2 for example. They do have a system called Bonus Pool where points will be distribute to player over time and those point only be spent toward your ranking when you actually win a game. With that system, you could make sure that cases like this never happen.;
On September 05 2013 10:59 thoraxe wrote:
It's obvious all of those with few hours are playing on their alt accounts. Perhaps part of the match making process involves moving those that do fairly decent as they start out to very high and if they can't handle the heat, they will get pushed down.
So by your suggestion, you are wasting other players time for an unranked match in order to calibrate a new player MMR. A placement match that rank you into the top 1% while you have 50% winrate in a small sample size just doesnt make any sense. That not even counting that having a placement match system in place, wasting others players time and ruining their enjoyment in order to calibrate 1 person MMR is just plain wrong.
|
I think you're ignoring the stats behind the wins, maybe he went 15-0 his first game. It doesn't look like the profiles show the scores of the wins and losses. Maybe there's some info that we are missing.
Edit: Also I fail to grasp your bonus pool theory. If fills if you don't play often so it seems that that would just help balance people who play a lot and people who play rarely.
|
Oh wow, game stats are taken into account in MM? Always thought it was pure win/loss or something. I mean, depending on the role, etc...right? :X
|
On September 05 2013 11:42 Aerisky wrote:
Oh wow, game stats are taken into account in MM? Always thought it was pure win/loss or something. I mean, depending on the role, etc...right? :X
Games stats probally are not ussually taken into account, but it appears Valve has an anti-smurf system that works fairly well, a good player that starts a new account rises in rank extremelly quickly. Because of that, a popular theory is that your stats in your first games may be relevant. After your MMR is stable I would say it's unlikely stats play a relevant role.
|
28090 Posts
On September 05 2013 11:42 Aerisky wrote:
Oh wow, game stats are taken into account in MM? Always thought it was pure win/loss or something. I mean, depending on the role, etc...right? :X
From what valve says they aren't taken into account except for initial smurf detection. If you look at this guys second game he crushed on Warlock, also he was queueing with a friend who is a very high player. With the combination of smurf detection and your confidence interval he was probably skyrocketed into the very high bracket. This is quite common as I have tested it multiple times, where I have soloed into very high after 1 game, and have reached page 10 level games after 15-20 wins. My main took like 300 wins to get there because I was legitimately new when I made that account.
Now there are other reasons for low win people to be in high level games other than smurf detection, such as queueing with friends, etc.
|
The problem with these new players is that they haven't played enough matches for the matchmaking to figure out how good they are. They have to placed in games nonetheless.
The easy way to eliminate your problem - bad players in high level games - is to make sure everybody plays more games at the lower ratings before they can be promoted. This way, you push the problem over to the lower rated games, where there will be more smurfs stomping.
The only other option is to have new accounts in an entirely separate queue, which creates its own host of problems.
|
A player should not be put into the top 1% of MM with less than 20 games played unless they win like all of them.
Saw singsing playing solo and he got someone on his team with less than 10 games played who was obviously completely new to the game (they just ran into creeps and died). Valve's system needs to be reworked. I think they are way too far onto the antismurf side of things. They should match potential smurfs with potential smurfs until like 50+ games and not with veterans.
|
Watch the game and then please explain to me how the invoker is the one that lost the game for you
|
As a new player (~ 40 played games) I don't feel that this has been a problem for me.
E: Meaning that I haven't run into players who has way more games played than me
|
LOL! In the first game @14 minutes, Luminous just realizes he has a Ricki on his team, pings him, and checks his score. You can literally feel the despair grow inside of him.
|
Valve has probably the worst matchmaking system in any game I have played and I have played a lot of competitive games. The fact that they don't use something so simple and already in place (elo) boggles my mind.
I feel like Valve is trying to reinvent the wheel when despite what Valve drones think we already have a perfectly working system. League of legends and Hon use straight up ELO and it works completely fine, sure their presentation of the ELO might be a cause for concern but the base matchmaking itself works completely fine, it's simply Valve being Valve and it's sadly ruining the non professional side of the game. The fact that because of the lack of a ladder LoL is the more competitive game when Dota is touted as the more "hardcore" game should raise red flags in all Dota fans.
Unfortunately two things aren't going to happen
1) we aren't going to get a ladder queue with real ranks so we can be sure we are matched evenly
2) we aren't going to see a competent non ranked matchmaking because Valve is showing no intention of fixing it.
Honestly, this is going to cause a "Brood War Effect" soon where the competitive community moves to In houses and ladders and MM is left to the extremely casual or lazy players, mark my words. The fix for this is MMR transparency and an actual ladder.
The "I'm scared of any kind of criticism so hide all my info" cry baby casual attitude the Vocal majority has been yelling at Valve since release is ruining the game.
|
Could you check the history of "pgg" and see if his games are him stacking with anyone else on your team previously?
That could 100% account for him being matched with you, think about it
"pgg" and person b stack together for 3 matches, win all 3 and "pgg"'s mmr is boosted substantially because of person b's existing mmr leading to them vs'ing high mmr players straight off the bat.
edit: the emulator sums it up pretty well, likely better worded than I did but you get the idea
|
28090 Posts
On September 05 2013 15:44 ReignSupreme. wrote:
Could you check the history of "pgg" and see if his games are him stacking with anyone else on your team previously?
That could 100% account for him being matched with you, think about it
"pgg" and person b stack together for 3 matches, win all 3 and "pgg"'s mmr is boosted substantially because of person b's existing mmr leading to them vs'ing high mmr players straight off the bat.
If you look at his dotabuff he has played 7 times with one guy who is a very high level player. That along with smurf detection can skew his mmr quite a bit.
|
You have over 4000 games played, and your complaining about one. These types of mmr mixes are few and far between.
While I agree the mm needs work, the example you have provided is a 'outlier' which we can ignore...
|
Yeah, played my first game of Dota 2 yesterday... Nearly all of the players were complete noobs, apart from Kunkka on my team, who got 31-0. I got the second best score with 8-3...
Means back to Starcraft for me.
|
|
|
|
|
|
EPT![[image loading]](http://i.imgur.com/eoxaomJ.png)
![[image loading]](http://i.imgur.com/LdtSleh.png)
![[image loading]](http://i.imgur.com/t9ym5Lj.png)
![[image loading]](http://i.imgur.com/KTozE2O.png)
![[image loading]](http://i.imgur.com/3FSY8I8.png)
![[image loading]](http://i.imgur.com/B4HPFek.png)
![[image loading]](http://i.imgur.com/85jtUuY.png)
![[image loading]](http://i.imgur.com/a3eDIoX.png)
![[image loading]](http://i.imgur.com/umKQcRB.png)
![[image loading]](https://twitter.com/LuminousInverse/status/374606837137416192/photo/1)
