DOTA 2
- Thread starter Bels_sl
- Start date
- 11,313
- 2,420
Because people don't use number of games, but number of wins, which is a terrible value to judge anything. It's like judging player skill based on the number of letters in the name of the player. "What a pleb, he only has 4 letters in his name, I have 12 letters, why am I matched with this shit". People use what they have access to, even though it's pretty much irrelevant as shit.
Also that Na'Vi game3 was pretty nice. Carry veno, Zeus mid, Ursa jungle.
if theres a pause or a long loader sometimes ill be a creeper and check, just to kinda gauge what im up against, or working with game wise and what kind of sense they may have.
Past 150~ wins its all the same to me, but no one under 150 wins should be pitted with someone that has played 2k games no matter if he's better or not. Hell make a "shit league for 2k games ppl that suck" if you must. :\
And besides, I'm talking about horrible players with sub 200 wins that dont understand base mechanics getting queued with me in a vhigh game. It's never been as bad as past 2 weeks...it was fine before. :\
- 11,313
- 2,420
That's stupid, it matches based on the hidden rating, doesn't matter how many games you've played, if you're bad at the game you're bad at the game. The issue lies in the number being shown being bigger even though it's irrelevant and people using it as a measure because whatever. They should just hide the number of wins and be done with it. I'm sure people will find more random shit to complain about, but should cut the stupid complaints by a decent bit.
Delly
Trakanon Raider
- 3,010
- 671
I have 2 Smeevil Treats. Not trading for anything specific and have all the player cards already. I would prefer 1 key for both, but it doesn't have to be that and I'm willing to negotiate.
http://steamcommunity.com/profiles/76561197967245463
http://steamcommunity.com/profiles/76561197967245463
i know how it matches, and for the past 2 weeks its been broken. They even said they're looking into it on dev forums, get off your high horse.
Either that or the massive influx of people from "official release" have made it outta whack until those new people get correct ratings. I just know that for the past 2 weeks the games I've been part of have been for the most part horrible and its long a time period + too many games to call it a "fluke".
- 11,313
- 2,420
And I didn't say it wasn't messy. I said 90% of the people crying about it cry about it for the wrong reasons. You said plenty of people post about it on dev.dota but a lot of the posts are idiots using shitty numbers to backup their argument. Matchmaking has always been messy, it might be even more nowadays since they just changed it, but it doesn't make "I have 1000wins I shouldn't be matched with people with 100wins" a good argument regardless.
Also their antismurf system can fuck up, maybe a new player gets lucky, pick riki, kills a lot of people in his first few games, and the game will attribute him a high rating based on that performance, and then you have a new player in a high game and he gets trashed. Shit will even up eventually but the large influx of new players create a larger number of these "wrong placement" cases.
My point is just, using Wins as a measure for anything is basically a retarded argument. In no way can you use wins alone as a basis for any sort of ranking, regardless of the numbers. And yes that includes 1win vs 1000wins. You can't say someone with 1000wins is obviously better than someone with 1wins, because of smurfs, previous gaming experience and the fact some people are awful at the game and have 1000wins but 1500losses. They played 2500games and they're still fucking awful, possibly more than a completely new player who's actually trying to improve and been churning bot games. Even using the win:loss ratio is not a guaranteed accurate representation because of the rating they're being played at. 50% winrate at low rating isn't as good as 50% winrate at high rating.
As I said, just hide the wins, or show the ratings if they want to go that way. Or do it the LoL way, have a separate queue for "ranked solo", and show only that ranking. Keep normal games for people who don't want to see the rating(even though there's a hidden rating in lol for these games too). I think that's probably the safest way to go about it, it won't hurt people's feelings because they can still play without a visual rating in the other mode. With that said, even with the visible rating, there's dozens and dozens of threads complaining about the ELO system in LoL and how it's "unfair" and how they keept getting matched with "the worst players ever" and the dreadful term "ELO Hell". Even if Valve's matchmaking was perfect(and it's certainly not), people would still bitch about it.
Dandain
Trakanon Raider
- 2,092
- 917
The major difference that is occurring in matchmaking now is when friends queue in partial stacks. Specifically wide skill gap queueing. Before it would just take an average skill of the players in the game and put you at that ELO. Which basically was the point of the spreadsheet - you've got to make the game bottom heavy for the weakest player(s) in the stack. Now if I queue with a 50 win friend, the other team will have a 50 win champ as well. Even if he's a solo dude. These are cases where 50 wins literally is the first 50 wins of their dota career. So with a stack of 3 players who are high/vhigh - we may end up grouped with a vhigh player with his low normal buddy as the 5th. The other team ends up with the roughly same scenario - in an attempt to give both teams a low win or low experience or low elo (hidden) whatever the reason. Basically - whoever's low elo guy is less shitty or got good advice from his friend to pick something with simple good easy to use spells tends to determine the game.
We had a double queue with a gyro player, who was our safe lane farmer who went aghs first, and had less cs than our offlane - we took it for granted the people in the game weren't going to be that new. He never built a bkb, and he was 1 roundable by qop the entire game - he was completely hostile and not open to suggestion. Guy had 40 wins. Another game a player I've played plenty of group queue games with brought in a friend who goes jungle prophet - and he procedes to have a 30 minute midas shadow blade, and repeatedly tp's on top of a huskar when he has no items stuns or disables to try and manmode him. Obviously this results in no kills and it completely distorts the games unless you can get to a 4 or 5 stack - then the problem at least for your team is mostly solved. It seems that a 4 stack has a much less % chance to get a low experience 5th. And they just match you against another 4 or 5 with the same style of extra.
These are the new shit games - and instead of the stacks being responsible for the hard games (like if we take 2 vhigh and 2 normal players from before) the game gets tossed in high and the normal guys struggle - the 5thplayer is a high player solo player - and everyone on the other team is probably a roughly high skill player unless there happens to be an identical composition stack queuing.
That's the entire part about how the new system not only takes into account the average - but more specifically it looks at the lowest skill player on the team - The current solution is throw in a similar skill player onto the other team - its a little better for straight up queueing with your friends but it sucks sometimes for the random solo or duo queue players who get tossed in to complete said teams.
TLDR; The worst games in matchmaking are no longer something that the stack itself can know before they hit queue.
We had a double queue with a gyro player, who was our safe lane farmer who went aghs first, and had less cs than our offlane - we took it for granted the people in the game weren't going to be that new. He never built a bkb, and he was 1 roundable by qop the entire game - he was completely hostile and not open to suggestion. Guy had 40 wins. Another game a player I've played plenty of group queue games with brought in a friend who goes jungle prophet - and he procedes to have a 30 minute midas shadow blade, and repeatedly tp's on top of a huskar when he has no items stuns or disables to try and manmode him. Obviously this results in no kills and it completely distorts the games unless you can get to a 4 or 5 stack - then the problem at least for your team is mostly solved. It seems that a 4 stack has a much less % chance to get a low experience 5th. And they just match you against another 4 or 5 with the same style of extra.
These are the new shit games - and instead of the stacks being responsible for the hard games (like if we take 2 vhigh and 2 normal players from before) the game gets tossed in high and the normal guys struggle - the 5thplayer is a high player solo player - and everyone on the other team is probably a roughly high skill player unless there happens to be an identical composition stack queuing.
That's the entire part about how the new system not only takes into account the average - but more specifically it looks at the lowest skill player on the team - The current solution is throw in a similar skill player onto the other team - its a little better for straight up queueing with your friends but it sucks sometimes for the random solo or duo queue players who get tossed in to complete said teams.
TLDR; The worst games in matchmaking are no longer something that the stack itself can know before they hit queue.
http://www.twitch.tv/joindotared/b/435286233
The Defence game 3 of finals. Starts at around 2:15. Dat team.
The Defence game 3 of finals. Starts at around 2:15. Dat team.
- 11,313
- 2,420
Yeah that seemed to be the case in the few games some dev guy checked in a thread that was bitching about new MM and he explained it works that way, which is pretty retarded. Mostly it's stacks ruining games, as usual I wanna say cause even before the patch it could be pretty shitty to be with a stack as the last player, but especially stacks of varying skill levels. I guess mostly it wouldn't be as much of an issue if there were 2 queues, the "do whatever" queue where you can team with your friend and proceeds to feed off the enemy low rating guy and a queue which for example doesn't allow queuing if your rating is too different, with a visible rating, for people who want better balanced games. I don't even know if that queue would take longer to start, there's hundreds of thousands people playing atm but it still seems to take as long as when there were only 5000 players in the game. I would reckon splitting the playerbase into 2 queues wouldn't hurt queue times even that much because it'd parse the available players faster to make balanced matches.
Anyway the current system is pretty retarded, there's no question there, I just don't like the trend of people bitching about wins or levels because these don't mean jackshit pretty much. Reddit is littered with these threads "so I'm rly gud player in normal bracket I have lots of wins and then I get newb player with 200wins in my game and gg matchmaking ded gme volvo".
As a sidenote, slightly related, I had the first intentional feeder since early beta in a game earlier. I don't know what was up but he was unhappy with some shit and just kept feeding over and over. The game was very much in our favor already(complete free farm doombringer, they didn't really have anything too carryish and no one farming well) and I assume they got into a heated argument about who was responsible for losing their trilane vs our highly competitive SK+Pudge dual hard lane. Guy had 600wins so it wasn't a new player or anything, just typical rager but hadn't seen these in a while.
Delly
Trakanon Raider
- 3,010
- 671
I was about to post the same video. Hilarious game and actually laughed a few times during it.
Also here is this
Saw the draft list while I was at work and couldn't wait to get home. Shit like that is why I have a Navi pennant. That game was fucking ridiculous and just great to watch.
- 11,313
- 2,420
My google powers have shown that yes it is possible, but it requires a bit of fucking around with the binds. This explains it simply. You cannot however bind them on the fly like you can in SC2, but at the same time unlike SC2 you only play one map and the enemy always spawns in the same position, so you shouldn't need to rebind them to make use of them, I think.
http://dt.reddit.com/r/DotA2/related...ons_in_dota_2/