Ladder Site Online...

Discussion of all aspects of multiplayer development: unit balancing, map development, server development, and so forth.

Moderator: Forum Moderators

Post Reply
User avatar
Rigor
Posts: 941
Joined: September 27th, 2007, 1:40 am

Re: Ladder Site Online...

Post by Rigor »

the history of wins/losses would be saved, because its small and is necessary for ranking and score. the replays are rather big and the disc space is reached and until nobody comes up with a more elegant solution this seems to be the one we have to take.
User avatar
milwac
Posts: 29
Joined: April 9th, 2010, 11:40 am

Re: Ladder Site Online...

Post by milwac »

I'm with the people who'd want to keep the replays. Reasons being, we might need the information about factions and units and maps later on to build statistics. Also old replays could be played on 1.8 or 1.9 as well if we just changed some WML (?) Not too sure about this though.

My suggestion will be, if the hosting service currently being used is free and cannot be upgraded anyway, create another free hosting account and dump all the old replays there. The date, time and player nicks are all on the replay so it won't be that hard to look for them later. (Download links in the DB can be changed to this new alternate server, not too necessary at the moment)

This suggestion is still a hack though. I'll do what moloch suggested, but at the same time if donaters stopped paying, the server can't shut down.
nelson
Posts: 91
Joined: March 19th, 2008, 11:15 pm
Contact:

Re: Ladder Site Online...

Post by nelson »

Where is the ladder site currently hosted? Who is currently donating the server space? Most crucially, how much space does it currently have?

It may be possible to get someone to donate a larger server space, but we need to know how much space etc. the current ladder setup occupies, and how much space we would need to keep it running smoothly for a while.
chains
Posts: 76
Joined: January 9th, 2007, 5:02 am
Location: Portland OR
Contact:

Re: Ladder Site Online...

Post by chains »

There was an archive of ladder replays on the server taking up hundreds of megs from 09 and before in a folder called PreviousLadder. There were totally inaccessable from the web and not a single person ever complained that they weren't there. So, I'm pretty sure they are copies of files that are already on the live site. I removed those backup replays, and the server should be functioning again.

The ladder is running on a shared server and consuming up to 80% of the server providing downloads at busy times. I could move the ladder to my Linode VPS account, but it's fine where it is as long as someone is taking a backup of the database and replays.

What the server needs is an admin who takes care of the replays via FTP. Currently the code stuffs all replays into a single folder called "replays" which has gotten so big that it won't open via FTP. The code should split these by month so that the folders don't get unmanagable. Also, there's no reason to maintain 10,000 replays that no one ever watches again. Something needs to be done about cleaning up old replays. Ideally, the files should also be managed from the admin interface of the ladder so the admin can archive and remove old replays to a zip file which could then be uploaded to source forge or some other place.

Another solution would be to use the real wesnoth replay archive so that people can download replays from the Wesnoth server. This removes the need for file management on the ladder server.

Rigor, if you need the FTP login for the ladder please email me.
Kolbur
Posts: 122
Joined: April 29th, 2009, 9:33 am

Re: Ladder Site Online...

Post by Kolbur »

Good thing the ladder is running again. Thank you chains for your efforts.

@milwac:
Good job with the Trueskill rating there, looks really good! I would be interested to see some rating graphs for players that are less consistent to see how their rating changed over the time. Let's say for thefish, nelson, skb, plk2 or neki. They have the most games and their elo ranking fluctuated considerably.

@All:
The ladder may be running again but there are still a lot of improvements needed. Everyone who is interested in helping in some way can message me, Rigor or Doc Paterson and we will PM you the link to our forum that's exclusively for the ladder. This thread is already way too cluttered and we can discuss the future of the ladder a lot better over there. Some people already found their way there and have started. :)
For everything else regarding the ladder this thread will still suffice.
User avatar
milwac
Posts: 29
Joined: April 9th, 2010, 11:40 am

Re: Ladder Site Online...

Post by milwac »

Hi

I was busy with work so couldn't post this earlier. Here are the graphs for the players mentioned :

http://xntrick.comuf.com/wesnoth/trueskill_graphs.htm

Cheers!

PS: Nice to see the site up and running again.
User avatar
Quetzalcoatl
Posts: 207
Joined: March 18th, 2009, 3:26 pm

Re: Ladder Site Online...

Post by Quetzalcoatl »

Great work on the whole TrueSkill thing :). Do you still have that code that draws charts? I thought it may be interesting to see progression of Owlface and species_8472 as both of them picked strategy to minimize risk by playing mostly weak players (mostly is a bit abusive in context of Owlface :P).

Cheers
Q
Ten soldiers wisely led will beat a hundred without a head.
User avatar
Doc Paterson
Drake Cartographer
Posts: 1973
Joined: February 21st, 2005, 9:37 pm
Location: Kazakh
Contact:

Re: Ladder Site Online...

Post by Doc Paterson »

Quetzalcoatl wrote:I thought it may be interesting to see progression of Owlface and species_8472 as both of them picked strategy to minimize risk by playing mostly weak players (mostly is a bit abusive in context of Owlface :P).

Cheers
Q
About these two "picking a strategy to minimize risk," I'd like to say a few words. ;)

As for species_8472, I think all one has to do to disprove this is to go to his profile: http://ladder.subversiva.org/profile.ph ... ecies_8472 and click on "opposition." You will notice that he is a combined 6-0 against Dauntless, c4rL0s, Michaelmax, zako, and Zarak (aka Shachi). More on the artfullness of the XX-0 accomplishment in a bit. :)

The Owlface case is pretty simple: Most of those matches occurred about 4 years ago, right after the ladder was created, during a time period when there were very very few good players at all partaking. I happen to know that the vast majority of those Owlface games were with people who joined Owlface's "open to anyone" ladder games. So as for these being "strategies to minimize risk," I really don't think so. The Owlface thing, I think, became more of a "kill the weak" project/joke ("Owls eat mice, they don't eat wolves..." :)) during the last 5-10 games, as the 30-0 milestone came into view.
I think it bears noting that there have been many accounts over the years that have gone the "kill the weak" route, but none that have achieved the 30-0 milestone, besides species_8472 (whose accomplishment is, imo, perhaps the greatest feat that the ladder has ever seen; a real masterpiece). Now, even if these two accounts had, in fact, been intentionally targeting the "weak," I think that their records would *still* indicate top tier play. And why is that?

Wesnoth is the kind of game where the RNG can drop you in the midst of a very, very hard to win situation, at a moment's notice. No need for me to say this, because we all know the deal- You're sometimes going to have those +40 %, -30% games, and even against a novice, that can be very challenging (in some people's opinion, practically impossible) to win. Perhaps this is hard to convey, but I feel like succeeding with great consistency at a (usually) less-challenging task (such as beating weaker players at Wesnoth, which gives significant weight to randomness) can still show a high degree of skill, and I don't think that the above ranking systems are *that* far off in showing their merit. I understand that some players see accounts attempting to prey upon weaker players and think, "I could beat them...." Now maybe they could, and maybe they couldn't, but I think that beating players like Owlface at their own game would be another matter entirely, trying to actually achieve a higher rank by doing better than 30-0 against weaker players. Would be interesting to see someone else try to do this..... ;) (Won't be me trying though- I have too many other things to do these days. :P)
I will not tell you my corner / where threads don't get locked because of mostly no reason /
because I don't want your hostile disease / to spread all over the world.
I prefer that corner to remain hidden /
without your noses.
-Nosebane, Sorcerer Supreme
User avatar
milwac
Posts: 29
Joined: April 9th, 2010, 11:40 am

Re: Ladder Site Online...

Post by milwac »

I think in cases of all win or nearly all win (less than 10% games lost) scenarios it is very hard to set an upper limit to the rating. Since either a 1700 or a 2800 could have won 30 games against 1500s, we'll never know the players' real strength until s/he lost. Trueskill still works well as the Rd (sigma) is high in such cases. The mean (mu) however is far higher and so the rating (= mu - 3*sigma) also remains high.

Here are the graphs (although I don't think there is anything interesting in them other than the obvious):

http://xntrick.comuf.com/wesnoth/truesk ... o_loss.htm
User avatar
Quetzalcoatl
Posts: 207
Joined: March 18th, 2009, 3:26 pm

Re: Ladder Site Online...

Post by Quetzalcoatl »

Nah I never had in mind to undermine ones achievements but OTOH this is a bit different approach on climbing ladders than what some other players done in the past. Anyways TS obviously looks a lot better than ELO but I wonder what are impressions of more of you and most of all if you think its good enough (if yes / no pls try to explain you perspective on the whole matter in more detail).

Cheers
Q
Ten soldiers wisely led will beat a hundred without a head.
Gallifax
Posts: 131
Joined: October 23rd, 2006, 5:36 pm
Location: Who cares?

Re: Ladder Site Online...

Post by Gallifax »

@Doc

It was certainly much easier back then,than nowadays:)
User avatar
Quetzalcoatl
Posts: 207
Joined: March 18th, 2009, 3:26 pm

Re: Ladder Site Online...

Post by Quetzalcoatl »

Quetzalcoatl wrote:...
Meh I hope at least somebody reads my posts :augh:.
Spoiler:
Results for dynamics factor of 1/4. What do you think of them?

Cheers
Q
Ten soldiers wisely led will beat a hundred without a head.
User avatar
Quetzalcoatl
Posts: 207
Joined: March 18th, 2009, 3:26 pm

Re: Ladder Site Online...

Post by Quetzalcoatl »

Some extra stuff (charts you most likely want to take closer look at (or at least at some of them)) goes here.
TSCharts.zip
(783.75 KiB) Downloaded 241 times
Cheers
Q
Ten soldiers wisely led will beat a hundred without a head.
The Black Sword
Posts: 373
Joined: October 13th, 2008, 4:35 pm

Re: Ladder Site Online...

Post by The Black Sword »

I haven't had a chance to look through the TS system yet. I'll probably try to look at it soon, if only to answer my next question. :P

On a purely selfish note though, can anybody explain why I (and Dauntless too looking at it) appear to have dropped so much under this system?
User avatar
milwac
Posts: 29
Joined: April 9th, 2010, 11:40 am

Re: Ladder Site Online...

Post by milwac »

@ Quetz : Please re-run your implementation with the modified games history file that I sent you with the original parameters. We need to see if the implementations are well understood and consistent first, and then go ahead with parameter changes. Some of the ranks do look good I must say, but this should not come at the expense of a wrong/inconsistent implementation.

@TBS: As you can see the system is currently being tested on and we're experimenting on the default parameters. Ratings can vary immensely in different systems due to these parameters. For example in Elo, a little tweak in the K values will change the ratings significantly. In Trueskill however this is not the case. Please don't be worried too much. I think it'll be better for us to release a running trueskill rating page on the ladder website while the current Elo system is still maintained, so that people could get used to it and point out bugs/modifications to the implementation before it is made the de-facto rating system for the ladder. Also AFAIR Dauntless is placed at #1 amongst currently active players in the results I published.

It'll be great if more people could look into the TS system and come forward to double check our implementations.
Post Reply