Altitude Ladder Change Log: newest change May 29, 2011 - Page 2

Beagle · #41 02-03-2010, 05:59 PM

Fantastic idea on the Ladder Ventrilo! Even though it seems like most communication in alty can be handled by typing, I think having the Vent would make people use communication more and we'd see much better games.

It would also be funny to hear your accents.

Varonth · #42 02-03-2010, 05:59 PM

Yeah the Rage quit stats aren't accurate as far as i can see it:
http://altitudeladder.net/matchlist.php?id=145
http://altitudeladder.net/profile.php?id=145

As you can see, the last match I played was on 1.2.2010 .
But I also got a Ragequit (btw i just got kicked from the server and I never quit a game ).

Could it be, that you are not just logging the leaves of the teams but every person on the server during a match?

CCN · #43 02-03-2010, 06:22 PM

ragequit stats picking up if you quit on winning screen?

I don't ragequit but had 2, last game I quit on winning screen and now I have 3.

nesnl · #44 02-03-2010, 08:23 PM

A "Rage Quit" is determined in the logs by the type of message the server is receiving when you leave a game. If you leave a game at any point between when starttournament is called and when the game is finished then it will be recorded. If you leave during the awards screen it shouldn't count, but we can look into it. It does not count kicks related to being timed out from the server or things of that nature (so if our server is being crappy and kicks everyone then it doesn't count).

For the most part, this stat is only going to be used to identify habitual game leavers. We realize that a lot of people have 2 or 3 disconnects and that some of them are probably cases where it was something other than just randomly quitting the game. Don't worry about it so much unless you have like 8-10% or more leave percentage.

Sarah Palin · #45 02-03-2010, 08:27 PM

Kudos on the interface changes (and I always knew prog was my arch nemesis!).

Ladder balance unfortunately is only "good" at making sure games are consistently +/- 25. It does not in any way ensure a competitive, plane-balanced game and some of the shortest games I've played have been ladder balanced. I think I prefer cappies for now.

Varonth · #46 02-03-2010, 09:45 PM

Quote:

Originally Posted by nesnl

A "Rage Quit" is determined in the logs by the type of message the server is receiving when you leave a game. If you leave a game at any point between when starttournament is called and when the game is finished then it will be recorded. If you leave during the awards screen it shouldn't count, but we can look into it. It does not count kicks related to being timed out from the server or things of that nature (so if our server is being crappy and kicks everyone then it doesn't count).

For the most part, this stat is only going to be used to identify habitual game leavers. We realize that a lot of people have 2 or 3 disconnects and that some of them are probably cases where it was something other than just randomly quitting the game. Don't worry about it so much unless you have like 8-10% or more leave percentage.

Then please, why do I have ragequits, when I never quit any game?
And I didn't even play a game since 2.2.2010 .

nobodyhome · #47 02-03-2010, 09:51 PM

Varonth, we reran the logs, so that this feature keeps track of all possible quits that you've made (even before 2.2.2010).

As for rage quits that you don't think happened, I will look through the logs of those in question sometime later tonight. I guarantee you that the logs will say you have quit at some point during the game (reentering the game immediately won't save you). It is possible that the logs erronously record a quit when you actually haven't, but there is nothing I can do about that, as logs is all I have to work with.

Flyngbanana · #48 02-04-2010, 12:19 AM

Does it count as a ping kick if you are spectating and get ping kicked?

nobodyhome · #49 02-04-2010, 12:21 AM

No it does not (can you even get ping kicked if you're not playing?). It only counts ping kicks and leaves for when you are actually in a startTournament.

eth · #50 02-04-2010, 12:40 AM

Quote:

Originally Posted by CCN

ragequit stats picking up if you quit on winning screen?

I don't ragequit but had 2, last game I quit on winning screen and now I have 3.

I'm 99% sure it doesn't record on win-screen as I usually quit the moment it hits(assuming I lose).

Also, ragequits was my attempt to be a half-humorous jackass, it's changed to "Leaves" now. Hope that's more correct, lol.

Varonth · #51 02-04-2010, 12:42 AM

Quote:

Originally Posted by nobodyhome

No it does not (can you even get ping kicked if you're not playing?). It only counts ping kicks and leaves for when you are actually in a startTournament.

Yes you can (happened to me 2h ago).

The only time I disconnected during a match was a massiv "kick" where everyone was kicked from the server.

http://www.altitudeladder.net/master...ort=duration_a

I think, it was the second longest match in ladderhistory, but I'm not 100% sure.

Edit:
Everyone of that match got at least 1 leave.

Sarah Palin · #52 02-04-2010, 05:01 AM

Cappies vs ladderbalance vs autobalance. Which one is best?

A tale of two games:

Game 1. TBD Asteroids, ladderbalance. A team of ACE.Curious, FlyingBanana, Hurripilot, King!, and ICE.Carstairs defeated DW.Low, Sarah Palin, Smushface, Ingbo and ICE.Lazy8. The game lasted 15:29. The losing team had made zero bomb delivieries. The game swing was +/-25 points.

Game 2. TBD Mayhem, captains. A team of damndirtyape, DW.Low, FlyingBanana, Sarah Palin, and Ingbo defeated KLF.Zen, SWN.Nightshade, Smushface, King! and protest boy. The game lasted 17:34. The losing team had made three bomb delivieries. The game swing was +/-23 points.

This (anecdotal) comparison helps make a few points:

1. Often, you can look at a ladderbalance output and see a foregone conclusion despite the theoretical "equality" of ELO.

2. People have said "once everyone's ELO settles the ladderbalancing will work." This is not the case because a team's performance is more than the sum of ELO. Even if everyone's ELO represented their true skill level, ladderbalancing could and would still produce "foregone conclusion" games with regularity because of plane mixtures. If a team of the 5 best Loopies (or 5 best Mirandas, etc) in the game faced a mixed team of lower total ELO - even 100 or 200 points lower - the mono team would lose. Captains are less likely to let all the Loopies play on one team (or let all the best whales get picked by the other captain, etc).

3. Captain games are often ELO-balanced anyway, as shown by the swing of 23 points (close to the theoretical ideal of 25).

At this moment, I prefer to play cappy games than ladderbalance. One advantage of ladderbalance is that it's faster (sometimes ALOT faster) but what's the point if the result is a 20 minute game where only one team makes successful bomb drops.

nesnl · #53 02-04-2010, 05:14 AM

I understand your argument, but using 1 example of each situation to prove your case isn't going to convince anyone. More statistical analysis would need to be done with a large sample pool of games that were autobalanced compared to games that were not (and also near a balanced elo).

However, the autobalance will be improved in the future. We are going to add components such as which plane a person most frequently uses, and factors such as average bomb hits per game (so each team gets a competent bomb carrier if there is one). There are many other factors that will be included as well and hopefully it will address some of these problems. There are big things to come for the ladder, so stay tuned.

Tosconi · #54 02-04-2010, 05:17 AM

Sarah P. - I'd rather disagree with you...In the last 2 days I've played loads of LB games, most of which were really nicely balanced.
The main disadvantage of LB is that it doesn't pay any attention on the planes ppl prefer...so that many participants have to chose their 2nd best choice. But all in all i'd say LB functions really well.

ryebone · #55 02-04-2010, 06:09 AM

Quote:

Originally Posted by Sarah Palin

At this moment, I prefer to play cappy games than ladderbalance. One advantage of ladderbalance is that it's faster (sometimes ALOT faster) but what's the point if the result is a 20 minute game where only one team makes successful bomb drops.

The same result as if you played for ten minutes, and waited ten minutes for teams to be picked. I've also been in plenty of captains games where one captain made horrible picks, leading to a quick game. In fact, I was also unfortunately a part of a captained two minute game (search my results: January 31, 3:22). Things like that are unpredictable, and cannot be attributed to one method of teampicking being better than the other.

To take plane choice into consideration along with rating is a monumental, probably impossible task that I don't think will ever be implemented unless someone was being paid to do it. Yes, ELO is a better representation of individual skill rather than teamwork, but this was always the overlying problem with the current system, and we've all come to terms with it being the best we've got.

Also, I most definitely do NOT agree with your assessment that most captain games are relatively ELO balanced. Pulling one game out of a sample size of over 1000 carries absolutely no weight, I don't think I need to explain why.

Stormich · #56 02-04-2010, 06:54 AM

It's a simple solution, learn to play more than 1 plane. There's like 5 of them. The balance in HoN is also done only from rating and there's like 40+ heroes to choose from. No one complains, they just pick the heroes that work together.

Sarah Palin · #57 02-04-2010, 09:00 AM

Quote:

Also, I most definitely do NOT agree with your assessment that most captain games are relatively ELO balanced. Pulling one game out of a sample size of over 1000 carries absolutely no weight, I don't think I need to explain why.

Quote:

I understand your argument, but using 1 example of each situation to prove your case isn't going to convince anyone. More statistical analysis would need to be done with a large sample pool of games that were autobalanced compared to games that were not (and also near a balanced elo).

Well gee we have that. Just look at the aggregate of matches in January.

Quote:

It's a simple solution, learn to play more than 1 plane.

The ladder isn't some procrustean bed that we have to fit our playstyles to. People should be picked for teams based on what they play, instead of having to adjust to being sorted into arbitrary teams.

Elo balance is not so great a goal that it's worth sacrificing competitiveness of games. Admittedly competitiveness is a more subjective benchmark but it could be measured by things like game length, base health of winning team, etc. However you quantify it, I bet it's a lot more closely correlated with plane balance than Elo. All of us can remember long and exciting games that ended with 30+ points changing hands in a massive upset.

Quote:

To take plane choice into consideration along with rating is a monumental, probably impossible task that I don't think will ever be implemented

I agree. And why bother when players can make that judgment themselves.

CCN · #58 02-04-2010, 09:25 AM

Suggestion:
Balance by Autobalance, then if it seems appropriate somoene can switch (e.g. 3 whale vs 0, switch a whale)

Flyngbanana · #59 02-04-2010, 11:23 AM

[] sample size.
One match really doesn't prove anything to be honest. The other team almost got two bombs off but we had some frantic defending at the last minute and/or lucky respawns.

nobodyhome · #60 02-05-2010, 07:39 AM

Quote:

Originally Posted by protest boy

What exactly is a rage quit? I have two listed on my account but to my knowledge I have been present at every game I've ever played in, all the way to the end.

Twice I have volunteered to sit out to even the teams when an enemy disconnects. Could this be the cause?

No, that can't cause it. However...

Quote:

Originally Posted by Varonth

Yes you can (happened to me 2h ago).

The only time I disconnected during a match was a massiv "kick" where everyone was kicked from the server.

http://www.altitudeladder.net/master...ort=duration_a

I think, it was the second longest match in ladderhistory, but I'm not 100% sure.

Edit:
Everyone of that match got at least 1 leave.

I looked into this today and discovered some things in the logs:

Code:

WARN [2010-01-26 20:15:14,379] [Thread-9]: Server hitch detected: 7813.0 milliseconds, free memory is 13.8 / 63.4 MB
INFO [2010-01-26 20:15:15,060] [Thread-9]: Server got disconnect request from Client[name=KLF.zencerator, ip=67.193.141.85:62768, playerId=4]
INFO [2010-01-26 20:15:15,060] [Thread-9]: Removing client 'KLF.zencerator' from 67.193.141.85:62768, playerId=4, message: null
INFO [2010-01-26 20:15:15,060] [Thread-9]: Server got disconnect request from Client[name=|J.C| Snap, ip=67.171.211.37:3927, playerId=6]
INFO [2010-01-26 20:15:15,060] [Thread-9]: Removing client '|J.C| Snap' from 67.171.211.37:3927, playerId=6, message: null
INFO [2010-01-26 20:15:15,060] [Thread-9]: Server got disconnect request from Client[name=.:$E:.DODD, ip=68.195.42.99:27272, playerId=10]
INFO [2010-01-26 20:15:15,060] [Thread-9]: Removing client '.:$E:.DODD' from 68.195.42.99:27272, playerId=10, message: null
INFO [2010-01-26 20:15:15,060] [Thread-9]: Server got disconnect request from Client[name=protest boy, ip=67.217.9.73:27272, playerId=5]
INFO [2010-01-26 20:15:15,060] [Thread-9]: Removing client 'protest boy' from 67.217.9.73:27272, playerId=5, message: null
INFO [2010-01-26 20:15:15,061] [Thread-9]: Handling join request from 66.229.106.192:27272 -> uI@268ea3cb<vaporId=88fde34e-8d18-4edc-ba40-387a992b3644, nickName={$E}zackzingki{T}, loggedIn=true, ip=66.229.106.192:27272, sessionId=7486ed69-1613-43d5-8263-47984c1ae73b, totalPlayTimeMillis=12 days 23 hours 39 minutes 31 seconds, affiliateCode=0, referrerClickPoints=0, referrerOtherPoints=0, referrerAdminPoints=0, referrerUnlocks=[0, 0, 0, 0]>
INFO [2010-01-26 20:15:15,061] [Thread-9]: Server got disconnect request from Client[name=drlove214, ip=76.24.37.159:55008, playerId=2]
INFO [2010-01-26 20:15:15,061] [Thread-9]: Removing client 'drlove214' from 76.24.37.159:55008, playerId=2, message: null
INFO [2010-01-26 20:15:15,061] [Thread-9]: Server got disconnect request from Client[name={ICE}sky, ip=71.59.36.238:27272, playerId=9]
INFO [2010-01-26 20:15:15,061] [Thread-9]: Removing client '{ICE}sky' from 71.59.36.238:27272, playerId=9, message: null
INFO [2010-01-26 20:15:15,061] [Thread-9]: Server got disconnect request from Client[name=|proggies, ip=203.125.178.82:2321, playerId=8]
INFO [2010-01-26 20:15:15,061] [Thread-9]: Removing client '|proggies' from 203.125.178.82:2321, playerId=8, message: null
INFO [2010-01-26 20:15:15,061] [Thread-9]: Server got disconnect request from Client[name=##|Mandrad, ip=213.22.228.246:27272, playerId=1]
INFO [2010-01-26 20:15:15,061] [Thread-9]: Removing client '##|Mandrad' from 213.22.228.246:27272, playerId=1, message: null
INFO [2010-01-26 20:15:15,061] [Thread-9]: Server got disconnect request from Client[name={ICE}Pein+, ip=24.148.119.33:27272, playerId=7]
INFO [2010-01-26 20:15:15,061] [Thread-9]: Removing client '{ICE}Pein+' from 24.148.119.33:27272, playerId=7, message: null
WARN [2010-01-26 20:15:15,071] [Thread-6]: Server hitch detected: 1501.0 milliseconds, free memory is 13.5 / 63.4 MB
WARN [2010-01-26 20:15:15,075] [Thread-9]: Server hitch detected: 696.0 milliseconds, free memory is 13.4 / 63.4 MB

Here was a game in which protest participated. In the middle of the game, the server "hitched" and then kicked a bunch of people out.

Almost simultaneously, Varonth was playing in the other server and this happened:

Code:

WARN [2010-01-26 20:40:44,845] [Thread-6]: Server hitch detected: 8413.0 milliseconds, free memory is 11.6 / 63.4 MB
INFO [2010-01-26 20:40:44,869] [Thread-6]: Server got disconnect request from Client[name=(|||)Maverick, ip=70.112.79.250:27272, playerId=16]
INFO [2010-01-26 20:40:44,869] [Thread-6]: Removing client '(|||)Maverick' from 70.112.79.250:27272, playerId=16, message: null
INFO [2010-01-26 20:40:44,869] [Thread-6]: Server got disconnect request from Client[name=IL|coolcow, ip=77.56.97.92:27505, playerId=9]
INFO [2010-01-26 20:40:44,869] [Thread-6]: Removing client 'IL|coolcow' from 77.56.97.92:27505, playerId=9, message: null
INFO [2010-01-26 20:40:45,373] [Thread-6]: Server got disconnect request from Client[name=lambchopz, ip=99.199.21.240:27272, playerId=13]
INFO [2010-01-26 20:40:45,373] [Thread-6]: Removing client 'lambchopz' from 99.199.21.240:27272, playerId=13, message: null
INFO [2010-01-26 20:40:45,542] [Thread-6]: Server got disconnect request from Client[name=[fLb]BG1, ip=174.89.239.236:27272, playerId=5]
INFO [2010-01-26 20:40:45,542] [Thread-6]: Removing client '[fLb]BG1' from 174.89.239.236:27272, playerId=5, message: null
INFO [2010-01-26 20:40:45,645] [Thread-6]: Server got disconnect request from Client[name=IL|Radium, ip=69.236.87.133:27275, playerId=11]
INFO [2010-01-26 20:40:45,645] [Thread-6]: Removing client 'IL|Radium' from 69.236.87.133:27275, playerId=11, message: null
INFO [2010-01-26 20:40:45,730] [Thread-6]: Server got disconnect request from Client[name=DW|strrrrront, ip=83.84.142.81:27272, playerId=10]
INFO [2010-01-26 20:40:45,730] [Thread-6]: Removing client 'DW|strrrrront' from 83.84.142.81:27272, playerId=10, message: null
INFO [2010-01-26 20:40:45,814] [Thread-6]: Server got disconnect request from Client[name==S.F.S= Varonth, ip=88.71.33.251:27281, playerId=14]
INFO [2010-01-26 20:40:45,814] [Thread-6]: Removing client '=S.F.S= Varonth' from 88.71.33.251:27281, playerId=14, message: null
INFO [2010-01-26 20:40:45,899] [Thread-6]: Server got disconnect request from Client[name=(||)Mikesol, ip=75.40.15.244:30202, playerId=0]
INFO [2010-01-26 20:40:45,899] [Thread-6]: Removing client '(||)Mikesol' from 75.40.15.244:30202, playerId=0, message: null
INFO [2010-01-26 20:40:45,911] [Thread-6]: Server got disconnect request from Client[name=[FN]Eagle, ip=70.188.24.105:27272, playerId=1]
INFO [2010-01-26 20:40:45,911] [Thread-6]: Removing client '[FN]Eagle' from 70.188.24.105:27272, playerId=1, message: null
INFO [2010-01-26 20:40:45,917] [Thread-6]: Server got disconnect request from Client[name=Hmmmm, ip=142.58.9.164:27272, playerId=15]
INFO [2010-01-26 20:40:45,917] [Thread-6]: Removing client 'Hmmmm' from 142.58.9.164:27272, playerId=15, message: null
INFO [2010-01-26 20:40:51,567] [Thread-9]: Handling join request from 70.188.24.105:27272 -> uI@927903b1<vaporId=e47deb48-061b-4c38-b405-8a6fc41a2e8d, nickName=[FN]Eagle, loggedIn=true, ip=70.188.24.105:27272, sessionId=8405a3cb-1df8-4fd1-bf20-924f2136cc1b, totalPlayTimeMillis=26 days 5 hours 21 minutes 31 seconds, affiliateCode=0, referrerClickPoints=14, referrerOtherPoints=100, referrerAdminPoints=0, referrerUnlocks=[0, 1, 0, 0]>
WARN [2010-01-26 20:40:53,394] [Thread-9]: Server hitch detected: 1827.0 milliseconds, free memory is 13.4 / 63.4 MB
WARN [2010-01-26 20:40:53,398] [Thread-6]: Server hitch detected: 7481.0 milliseconds, free memory is 13.3 / 63.4 MB

That server kicked half of everybody out there too.

Here is the particular series of log entries that is of concern:

INFO [2010-01-26 20:40:45,917] [Thread-6]: Server got disconnect request from Client[name=Hmmmm, ip=142.58.9.164:27272, playerId=15]
INFO [2010-01-26 20:40:45,917] [Thread-6]: Removing client 'Hmmmm' from 142.58.9.164:27272, playerId=15, message: null

This format of log entry is the exact type of entry that is logged when some just leaves regularly. Unfortunately, there is no way to tell the difference between the two. So with my the current logging method there is nothing we can do about these kinds of mass server kicks.

With lam's new patch though, hopefully there is a better way of tracking leavers. I'll see what I can do.

nobodyhome · #61 02-15-2010, 12:34 PM

bump for update.

nobodyhome · #62 02-23-2010, 05:15 AM

bumpity bump.

lamster · #63 02-23-2010, 06:06 AM

The real issue here is the hitch -- an 8413 millisecond hitch means the server was literally stopped (not running any logic, neither sending nor receiving packets) for over 8 seconds. Needless to say even a much smaller hitch can cause serious gameplay disruptions and undermine the legitimacy of a competitive match. There's already a way to track these connection timeout leaves in the non-JSON logs: parse the hitch entry and ignore any disconnects that occur within a few seconds of it.

I'll try to remember to add some extra info to connection-timeout-related-leave-messages but parsing the hitch is probably more important and meaningful in most cases. I'll add a JSON hitch entry in the next patch:
type=serverHitch
duration=<float, length of hitch in milliseconds>
changedMap=<boolean, true if the hitch occurred in an update that involved changing the map, e.g. is probably normal and can be safely ignored>

As for getting rid of hitches, that's a hosting issue. If the server box is overloaded (not enough memory, CPU maxed, etc) you will experience hitches. If you detect hitches mid game (e.g. not immediately subsequent a map change) you may want to throw out the game for competition purposes, and you'll certainly want to take a look at the host box to try to figure out what's going on.

X_denied · #64 03-07-2010, 06:09 AM

uh... ladder does not work, custom start does not work. etc

nobodyhome · #65 03-07-2010, 10:31 AM

Was down for today, should be fixed and back up now.

blln4lyf · #66 03-08-2010, 12:05 AM

Still doesn't work, fyi

Flyngbanana · #67 03-08-2010, 08:34 PM

Occasionally the ladder balances it 6v4 for some reason. Did it twice in a row on lost city, we changed map and it was still 6v4. Happened twice again before everyone gave up and quit. The teams had people of equal ratings on either side.

nobodyhome · #68 03-08-2010, 08:43 PM

Quote:

Originally Posted by Flyngbanana

Occasionally the ladder balances it 6v4 for some reason. Did it twice in a row on lost city, we changed map and it was still 6v4. Happened twice again before everyone gave up and quit. The teams had people of equal ratings on either side.

I'll look into this later today. This was on the tbd servers? Next time it happens take a screenshot so I can get a better idea of what went on.

Flyngbanana · #69 03-08-2010, 11:08 PM

Yeah it was on the tbd servers. I'll try and remember to find a screenie if it happens again.

nobodyhome · #70 03-09-2010, 08:51 PM

We have discovered what the problem is with the balancing causing 4v6s and 5v7s. I will fix this tonight, but for now, to avoid this problem if you see it happening, please change your name so that it does not include the characters \ or ". This could theoretically cause things like 10v0s and 12v0s if enough people in the game had these.

tgleaf · #71 03-09-2010, 09:02 PM

Quote:

Originally Posted by nobodyhome

We have discovered what the problem is with the balancing causing 4v6s and 5v7s. I will fix this tonight, but for now, to avoid this problem if you see it happening, please change your name so that it does not include the characters \ or ". This could theoretically cause things like 10v0s and 12v0s if enough people in the game had these.

Poor \/\/()|_|=_||\/|/\)(

nobodyhome · #72 03-11-2010, 12:56 AM

Just posting to say that the 4v6 issue was fixed as of last night. Feel free to change your name back to /-\\/\/|=$<>|\/||= or whatever again.

If you see the issue pop up again, please post a screenshot here.

nobodyhome · #73 04-28-2010, 09:29 AM

bump for update

CCN · #74 04-28-2010, 11:58 AM

Quote:

Originally Posted by nobodyhome

Just posting to say that the 4v6 issue was fixed as of last night. Feel free to change your name back to /-\\/\/|=$<>|\/||= or whatever again.

If you see the issue pop up again, please post a screenshot here.

http://xkcd.com/327/

nobodyhome · #75 06-01-2010, 02:59 AM

bump for teh lulz (and for some minor changes).

Sunaku · #76 06-01-2010, 03:07 PM

Quote:

You can now have more than 10/12 people on a team before you start the game. The ladder code will automatically spectate people at random until there is an appropriate number of people playing, spectating first the people who have just played.

I can't wait to see the drama, thanks ! Unless it happens to me too much, in which case I'll come back insulting you.

nobodyhome · #77 11-12-2010, 10:14 AM

Bump for bug fix.

beefheart · #78 11-12-2010, 11:40 AM

Hooray

(for Dr. Zoidberg)

MajorPayne257 · #79 11-12-2010, 08:58 PM

For reals? Awesome.

shrode · #80 11-13-2010, 04:18 AM

it had a couple instances of freeze in lobby after the tournament started today. aka it worked fine but the map just never switched. admin had to switch map then switch back to lobby to fix it.